Tectonic Geomorphology of Mountains: A New Approach to Paleoseismology William B. Bull tS%B|ackweM \9ß Publishing Tectonic Geomorphology of Mountains Tectonic Geomorphology of Mountains: A New Approach to Paleoseismology William B. Bull tS%B|ackweM \9ß Publishing © 2007 William B. Bull blackwell publishing 350 Main Street, Maiden, MA 02148-5020, USA 9600 Garsington Road, Oxford OX4 2DQ, UK 550 Swanston Street, Carlton, Victoria 3053, Australia The right of William B. Bull to be identified as the Author of this Work has been asserted in accordance with the UK Copyright, Designs, and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs, and Patents Act 1988, without the prior permission of the publisher. First published 2007 by Blackwell Publishing Ltd 1 2007 Library of Congress Cataloging-in-Publication Data Bull, William B., 1930- Tectonic geomorphology of mountains : a new approach to paleoseismology / William B. Bull, p. cm. Includes bibliographical references and index. ISBN-13: 978-1-4051-5479-6 (hardback : alk. paper) ISBN-10: 1-4051-5479-9 (hardback : alk. paper) 1. Morphotectonics. 2. Paleoseismology. I. Title. QE511.44.B85 2007 551.43'2-dc22 2006100890 A catalogue record for this title is available from the British Library. Set in 10.74/1 lpt AGaramond by SPi Publisher Services, Pondicherry India Printed and bound in Singapore by C.O.S Printers Pte Ltd The publisher's policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured from pulp processed using acid-free and elementary chlorine-free practices. Furthermore, the publisher ensures that the text paper and cover board used have met acceptable environmental accreditation standards. For further information on Blackwell Publishing, visit our website: www.blackwellpublishing.com Contents Preface................................... viii 1 Scrunch and Stretch Bedrock Uplift 1.1 Introduction.............................. 3 1.2 Pure Uplift, Stretch and Scrunch Bedrock Uplift................. 6 1.2.1 Isostatic and Tectonic Uplift..................... 6 1.2.2 Stretch and Scrunch Tectonics..................... 12 1.3 Landscape Responses to Regional Uplift.................... 23 2 Concepts for Studies of Rising Mountains 2.1 Themes and Topics........................... 27 2.2 The Fundamental Control of Base Level.................... 28 2.2.1 Base Level............................ 28 2.2.2 Base-Level Change......................... 28 2.2.3 The Base Level of Erosion...................... 31 2.2.4 The Changing Level of the Sea.................... 33 2.2.5 Spatial Decay of the Effects of Local Base-Level Changes........... 37 2.3 Threshold of Critical Power in Streams.................... 39 2.3.1 Relative Strengths of Stream Power and Resisting Power............ 41 2.3.2 Threshold-Intersection Points..................... 42 2.4 Equilibrium in Streams.......................... 42 2.4.1 Classification of Stream Terraces.................... 42 2.4.2 Feedback Mechanisms........................ 45 2.4.3 Dynamic and Static Equilibrium.................... 46 2.5 Time Lags of Response.......................... 49 2.5.1 Responses to Pulses of Uplift..................... 50 2.5.2 Perturbations that Limit Continuity of Fluvial Systems............ 51 2.5.3 Lithologic and Climatic Controls of Relaxation Times............ 54 2.5.4 Time Spans Needed to Erode Landforms................. 57 2.6 Tectonically-Induced Downcutting...................... 58 2.6.1 Straths, Stream-Gradient Indices, and Strath Terraces............. 58 2.6.2 Modulation of Stream-Terrace Formation by Pleistocene-Holocene Climatic Changes . 65 2.7 Nontectonic Base-Level Fall and Strath Terrace Formation............. 66 2.8 Hydraulic Coordinates.......................... 69 3 Mountain Fronts 3.1 Introduction.............................. 75 3.2 Tectonically Active Escarpments....................... 79 3.2.1 Faceted Spur Ridges........................ 79 3.2.2 Mountain-Piedmont Junctions.................... 83 3.2.3 Piedmont Forelands........................ 86 vi Contents 3-3 Fault Segmentation of Mountain Fronts.................... 97 3.3.1 Different Ways to Study Active Faults.................. 97 3.3.2 Segmentation Concepts and Classification................ 104 3.3.3 Fault-Segment Boundaries...................... 105 3.3.4 Normal Fault Surface Ruptures.................... 106 3.3.5 Strike-Slip Fault Surface Ruptures................... 113 3.4 Summary.............................. 115 4 Tectonic Activity Classes of Mountain Fronts 4.1 Tectonic Setting of the North America-Pacific Plate Boundary............ 117 4.2 Appraisal of Regional Mountain Front Tectonic Activity.............. 119 4.2.1 Geomorphic Tools For Describing Relative Uplift Rates........... 119 4.2.1.1 Mountain-Front Sinuosity.................. 122 4.2.1.2 Widths of Valleys...................... 124 4.2.1.3 Triangular Facets...................... 127 4.2.2 Diagnostic Landscape Classes of Relative Tectonic Activity.......... 128 4.2.3 Regional Assessments of Relative Tectonic Activity............. 141 4.2.3.1 Response Time Complications and Strike-Slip Faulting........ 141 4.2.3.2 Maps of Relative Uplift................... 145 4.3 Summary.............................. 164 5 Fault Scarps 5.1 General Features............................ 165 5.2 Scarp Morphology Changes with Time.................... 172 5.2.1 Changes in Scarp Height...................... 173 5.2.2 Decreases in Maximum Scarp Slope.................. 174 5.2.3 Diffusion-Equation Modeling..................... 175 5.3 Climatic Controls of Fault-Scarp Morphology................. 181 5.4 Lithologic Controls of Fault-Scarp Morphology................. 184 5.4.1 Fault Rupture of Different Materials.................. 185 5.4.2 Lithologic Controls on an 1887 Fault Scarp................ 187 5.4.2.1 Geomorphic Processes.................... 190 5.4.2.2 Scarp Materials...................... 193 5.4.2.3 Scarp Morphology..................... 194 5.5 Laser Swath Digital Elevation Models.................... 196 5.6 Dating Fault Scarps with Terrestrial Cosmogenic Nuclides............. 201 5.6.1 Alluvium............................ 201 5.6.2 Bedrock............................ 204 5.7 Summary.............................. 207 6 Analyses of Prehistorical Seismic Shaking 6.1 Paleoseismology Goals.......................... 209 6.2 Earthquake-Generated Regional Rockfall Events................. 212 Contents vii 6.2.1 New Zealand Earthquakes...................... 212 6.2.1.1 Tectonic Setting...................... 212 6.2.1.2 Background and Procedures.................. 215 6.2.1.3 Diagnostic Lichen-Size Peaks.................. 225 6.2.1.4 Tree-Ring Analyses..................... 227 6.2.1.5 Alpine Fault Earthquakes................... 241 6.2.1.6 Recent Marlborough Earthquakes................ 246 6.2.2 California Earthquakes....................... 255 6.2.2.1 Calibration of Lichen Growth Rates............... 257 6.2.2.2 Recent Cliff Collapse..................... 258 6.2.2.3 Rockfall Processes in Glaciated Valleys............... 262 6.2.2.4 San Andreas Fault Earthquakes................. 265 6.2.2.5 Lichenometry and Precise Radiocarbon Dating Methods........ 270 6.3 Summary............................... 273 References Cited................................ 275 Index.................................... 305 Preface Uplift by mountain-building forces changes fluvial landscapes. Pulsatory tectonic activity on a range-bounding fault increases relief, changes rates of geo-morphic processes, and modifies the shapes of hills and streams. Landscape responses to uplift occupy a critical time frame for studies of past earthquakes between the brevity of instrumental seismic data and long-term geologic crustal shifts. The appealing challenge for us is to determine how and when nearby and distant parts of the landscape change in consecutive reaches upstream from a tectonically active range front. Each climatic and lithologic setting has a characteristic style and rate of erosion, which adds spice to the scientific challenge. Landscape analyses include the geomorphic consequences of seismic shaking and surface rupture and their associated hazards to humankind. Tectonic geomorphology is essential for complete paleoseismology investigations. Locations, sizes, times, and patterns of seismic shaking by prehistorical earthquakes can be described and surface rupture and seismic-shaking hazards evaluated. This book explores tectonic geomorphology of mountain fronts on many temporal and spatial scales to encourage expansion of paleoseismology inquiries from the present emphasis on stratigraphic investigations in trench exposures. Evaluating earthquake hazards is in part a study of mountain-front segments. Cumulative displacements over late Quaternary time spans create landscape assemblages with distinctive signatures that are functions of uplift rate, rock mass strength, and the geomorphic processes of erosion and deposition. Such interactions define classes of relative uplift. Tectonic activity class maps define tectonically inactive regions as well as fronts of slow to rapidly rising mountains. Fault scarps focus our attention on recent surface ruptures and propagation of active faults. Dating and describing the characteristics of single prehistoric surface-ruptures is important. But now we can link sequences of events and depict sequences of prehistorical earthquakes along complex plate boundary fault zones. Examples here include the Alpine fault in New Zealand and the northern Basin and Range Province in the United States. This book applies a variety of geomorphic concepts to tectonics and paleoseismology. Don't expect landscape summaries for all major mountain ranges. Repetitive descriptions would dilute explanation and application of basic principles. Do expect essential concepts that should help you better understand the landscape evolution of your favorite mountains. Mountain front tectonic geomorphology studies can determine: 1) Which faults are active [Holocene ruptures], 2) Fault slip rates for short time spans [offset landforms] and long time spans [landscape evolution], 3) Time of most recent surface rupture and degree of irregularity of earthquake recurrence interval, and 4) Intensity and extent of seismic shaking. The amount of related literature cited borders on being unwieldy because of topic diversity of and the rapidly increasing interest of earth scientists in these subjects. I had to pick and choose so as to not overwhelm the content with citations of relevant literature. My citations are merely a gateway to related literature. Dating times of prehistoric earthquakes and estimating rates of tectonic and geomorphic processes continue to be of paramount importance. Study methods are changing, and precision and accuracy are improving. Diffusion-equation modeling of fault scarps and stratigraphic radiocarbon dates on pre- or post earthquake material collected from trenches have long been bastions for approximate age estimates. Sykes and Nishenko made a plea in 1984 for better ways of dating frequent earthquakes along plate boundary fault zones whose earthquake recurrence intervals may be shorter than the intervals defined by groups of overlapping radiocarbon age estimates. The rapid development of terrestrial cosmogenic nuclides broadens dating perspectives by estimating ages beyond the reach of radiocarbon analyses and by making surface-exposure dating a cornerstone for studies of geomorphic processes. Tree-ring analyses and lichenometry have potential for dating prehistorical earthquakes with a precision of ± 5 years. Preface Both methods are used here in a study of Alpine fault, New Zealand, earthquake history. The subjects of the six chapters are wideranging. Acknowledging the scrunch and stretch horizontal components of bedrock uplift is assessed from a geomorphic standpoint in Chapter 1. Diverse, essential conceptual models and methods for fluvial tectonic geomorphology are presented in Chapter 2. Contrasting tectonic land-forms and landscape evolution associated with thrust and normal faults are the focus of Chapter 3. Uplift, stream-channel downcutting, and piedmont aggradation are interrelated base-level processes that are used to define relative classes of mountain-front tectonic activity in Chapter 4. The fault scarps of Chapter 5 are incipient mountain fronts with surface-rupture recurrence intervals ranging from 200 years to 200,000 years. Chapter 6 considers how mountains crumble from seismic shaking. It uses coseismic rock-falls and tree-ring analyses for precise, accurate dating of earthquakes of the past 1,000 years and for mapping the intensity of seismic shaking of these prehistorical events. Readers should know basic geologic principles as these essays are written for earth scientists and students of geomorphic processes, landscape evolution, and earthquake studies. This book is appropriate for upper division and graduate-level courses in active tectonics, geologic hazards, tectonic geomorphology, physical geography and geomorphology, engineering geology, and paleoseismology. This project began in 1975 when Luna Leopold encouraged me to embark on selected in-depth geomorphic syntheses using book manuscripts as a career development tool. Global climate change and tectonic deformation are major factors influencing the behavior of fluvial geomorphic systems. Book goals determined my study emphases in a series of projects. "Geomorphic Responses to Climatic Change" (Bull, 1991) revealed pervasive impacts on geomorphic processes of arid and humid regions. This second book examines tectonic geomorphology of mountain ranges in a paleoseismology context. Of course the varied content of this book is indeed a team effort by the earth-science community. Students in the Geosciences Department at the University of Arizona played essential roles in every chapter. Peter Knuepfer, Larry Mayer, Les McFadden, Dorothy Merritts, and Janet Slate were among the many who tested the conceptual models of Chapter 2 with field-based studies. The first true positive test of the fault segmentation model (Schwartz and Coppersmith, 1984) in Chapter 3 is the work of Kirk Vincent. Les McFadden and Chris Menges broke new ground with me for the Chapter 4 elucidation of tectonic activity classes of mountain fronts of the Mojave Desert and Transverse Ranges of southern California. Susanna Calvo, Oliver Chadwick, Karen Demsey, Julia Fonseca, Susan Hecker, Phil Pearthree, and Kirk Vincent helped define the essential aspects for studies of normal-fault scarps of the Basin and Range Province in a vast region stretching from Idaho into Mexico. Andrew Wells kindly provided fascinating details about the sensitivity of New Zealand coastal and fluvial landscapes to seismic shaking. The integration of geomorphic and structural features shown in the Figure 1.12 map is the work of Jarg Pettinga. Kurt Frankel and Mike Oskin shared results and concepts of work in progress and Figures 5.35-5.40. The book project expanded in scope during a decade when a new lichenometry method was developed to date and describe how seismic shaking influences rockfalls and other landslides. Lichenometry projects included expeditions into the Southern Alps and Sierra Nevada with Fanchen Kong, Tom Moutoux, and Bill Phillips. Their careful fieldwork and willingness to express divergent opinions were essential ingredients for this paleoseismology breakthrough. I appreciate the assistance of John King in sampling and crossdating the annual growth rings of trees in Yosemite, and of Jim Brune's help in measuring lichen sizes near the Honey Lake fault zone. Jonathan Palmer introduced me to Oroko Swamp in New Zealand, which turned out to be a key dendroseismology site. Images are essential for landscape analysis and portrayal. Tom Farr of the Jet Propulsion Laboratory of the California Institute of Technology always seemed to have time to help find the essential NASA and JPL images used here. The banner photo for Chapter 2 and Figure 4.14 are the artistry of Peter Kresan. I thank Preface Frank Pazagglia for Figure 2.4, Malcolm Clark for the Chapter 4 banner photo, Tom Rockwell for the Figure 5.28 image, Greg Berghoff for Figure 5.34, Scott Miller for Figure 6.2 and Eric Frost for Figure 6.9A. Formal reviews of the entire book manuscript by Lewis Owen and Philip Owens provided numerous suggestions that greatly improved book organization and content. I am especially indebted to Wendy Langford for her meticulous proofreading and to Rosie Hayden for editorial suggestions. Their thoroughness improved format and uniformity of expression. It was a pleasure to work with the efficient production staff at Blackwell Publishing including Ian Francis, Rosie Hayden, and Delia Sandford. Essential financial and logistical support for this work was supplied by the U.S. National Science Foundation, National Earthquake Hazards Reduction Program of the U.S. Geological Survey, National Geographic Society, University of Canterbury in New Zealand, Hebrew University of Jerusalem, Royal Swedish Academy of Sciences, and Cambridge University in the United Kingdom. Earthquakes! Active Tectonics! Evolution of Mountainous Landscapes! Landscapes have a fascinating story to tell us. Tectonic geomorphology intrigues laypersons needing practical information as well as scientists curious about Earth's history. How fast are the mountains rising? When will the next large earthquake occur? Will the seismic shaking disrupt the infrastructures that we depend on? How do the landscapes surrounding us record mountain-building forces within the Earth's crust, and how does long-term erosion influence crustal processes? Humans are intrigued by tectonic geomorphology on scales that include origins of continents, grandeur of their favorite mountain range, and the active fault near their homes. Let us expand on the purpose and scope summarized in the Preface by elaborating on the structure of this book. I introduce, describe and use geomorphic concepts to solve problems in tectonics and paleoseismology. The intended geographical focus is global application of examples from southwestern North America and New Zealand. A fluvial emphasis excludes glaciers, sand seas, and active volcanoes. I present data and analyses from diverse tectonic, climatic, and lithologic settings so you can resolve similar problems in other geographical settings. This book emphasizes responses of fluvial systems to uplift, or more specifically the adjustments of geomorphic processes to base-level fall. Uplift terminology usage continues to change since the hallmark paper by Molnar and England (1990). Geomorphologists may use uplift terms in a different context than structural geologists. So Chapter 1 is a brief review of terminology and types of base-level change induced by tectonic deformation in extensional and contractional settings. Such crustal stretching and scrunching is nicely recorded by landforms ranging in size from mountain ranges to fault scarps. A variety of useful geomorphic concepts are assembled in Chapter 2 instead of being scattered. Get familiar with these principles. This broad base of essential concepts lets you evaluate and explore new and diverse approaches in tectonic geomorphology. These include a sensitive erosional-depositional threshold, time lags of response to perturbations (changes in variables of a system), types of equilibrium (graded) conditions in stream Photograph of 59,000 and 96,000 marine terraces (Ota et al., 1996) and 330,000 year old mountains (Bull, 1984, 1985) rising out of the sea at Kaikoura, New Zealand Chapter 1 systems, local and ultimate base levels, and the process of tectonically induced downcutting to the base level of erosion. These guidelines are a foundation for understanding interrelations between tectonics and topography in the next three chapters. Chapter 3 compares the landscape evolution and useful tectonic landforms for mountain ranges being raised by slip on active thrust and normal faults. These fluvial systems are affected differently by the two styles of tectonic base-level fall. Strike-slip faulting tends to tear drainage basins apart: a much different subject that is not emphasized here. Some tectonic landforms, like triangular facets, are rather similar in different tectonic settings. But piedmont landforms are much different in thrust-and normal-fault landscapes. Comparable contrasts should be expected elsewhere, such as the countries bordering the Mediterranean Sea, and Mongolia. The next three chapters discuss tectonic geo-morphology for three distinct time spans (Fig. 1.1) of about 2,000,000, 12,000, and 1,000 years. The tec-tonic-geomorphology theme continues to be applications for paleoseismology. The landscape tectonic activity classes of Chapter 4 are based on universal geomorphic responses to different rates of base-level fall during the Quaternary time span. The resulting diagnostic landscape assemblages are defined and mapped for diverse tectonic and structural settings in California. This model could have been created, and applied, just as easily for suites of mountain fronts in Japan, China, Mongolia, and Russia. Fault scarps are the focus of Chapter 5, with an emphasis on the Holocene time span. Choosing to discuss recent surface ruptures in southwestern North America was done in part to hold variations of several controlling factors to a limited range. These include climate and alluvium mass strength. Such studies of incipient mountain fronts can be made just as easily in the Tibetan Plateau, the Middle East, and Africa. New approaches are overdue to decipher the sequences of frequent earthquakes that characterize plate-boundary fault zones. Chapter 6 develops a new geomorphic way to precisely date earthquakes in New Zealand and to describe their seismic shaking. It then tests the model in California. This geomor- Stretch and scrunch bedrock uplift Conceptual models for fluvial tectonic geomorphology 3 Thrust- and normal- Classes of rising Late Quaternary Surface-rupture and faulted mountains landscapes Fault scarps seismic-shaking events Pleistocene Holocene ^^^^ - The past 1,000 years Assessment of potential earthquake hazard and risk ] Figure 1.1 Major topics of this book and their application to paleoseismology. Scrunch and Stretch Bedrock Uplift 3 phic approach to paleoseismology provides essential information about the frequency and magnitude of recurrent tectonic perturbations such as surface ruptures and seismic shaking. Other plate-boundary settings, such as the Andes of South America, Anatolian fault zone of Turkey, and the Himalayas may be even better suited for this way to study earthquakes than my main study areas. This book uses two primary, diverse study regions to develop concepts in tectonic geomorphology for fluvial systems in a global sense. Principal sites in New Zealand are shown in Figure 1.2 and southwestern North America sites in Figure 1.3 together with the links to their chapter section numbers. 1.1 Introduction Continental landscapes of planet earth are formed in large part by interactions of tectonic and fluvial processes, which are modulated by the pervasive influence of late Quaternary climate changes. Tectonics is the study of crustal deformation: the evolution of Figure 1.2 Locations of Southern Alps study sites in the South Island of New Zealand discussed in Sections 1.2, 2.4, 2.5, 2.6, and 6.2.1. This is a grayscale version of Shuttle Kadar Topography Mission image F\A06662 furnished courtesy of NASA and JPL. 4 Chapter 1 Scrunch and Stretch Bedrock Uplift 5 geologic structures ranging from broad transition zones between crustal plates to small faults and folds. Geomorphology is the study of landscapes and the processes that shape them. The influences of vertical and horizontal earth deformation on fluvial, coastal, and glacial processes and the resulting landscapes comprise the domain of tectonic geomorphology. The main emphasis here is on fluvial system responses to tectonic deformation. The challenge for all of us is to more fully recognize and use tectonic signals in the landscapes around us. The consequences of earth deformation by specific geologic structures profoundly affect geomorphic processes and landscape evolution. Conversely, evolution of landscape assemblages can be used to decipher the kinematics of faults and folds. Changes in style, rate, and locations of faulting and folding change the landscape too. An example is the Hope fault of New Zealand where Eusden et al. (2000) describe a 13 km long and 1.3 km wide transpressional duplex structure (adjacent areas of rise and fall) that has migrated northeast along a range-bounding oblique-slip fault that is as active as the San Andreas fault of California, USA. This leading portion of the duplex structure is rising on thrust faults. In the trailing southwest portion, formerly active duplex structures are now collapsing, undergoing a reversal of slip style to become normal faults. Rising geomorphic base levels become falling base levels with dramatic consequences for hills and streams of upstream watersheds. Another example is drainage nets that change as tips of faults propagate (Jackson et al., 1996). Structural geologists need to recognize how tectonic deformation affects erosion, deposition, and landforms. Tectonic geomorphology aids tectonic inquiries on many temporal and spatial scales. Some of us seek to understand how horizontal, as well as vertical, earth deformation affects the shapes of hills and streams in a quest to better understand long-term partitioning of strain along plate boundary-fault systems (Lettis and Hanson, 1991). Others study landslides in order to determine earthquake recurrence intervals and to make maps depicting patterns of seismic shaking caused by prehistorical earthquakes (Chapter 6). Tectonic geomorphology, seismology, and paleoseismology are cornerstone disciplines for studies of active tectonics (neotectonics). Seismology—historical instrumental studies of earthquakes - contributes much to our understanding of crustal structure and tectonics by 1) defining earthquake hypocenters (location and depth of initial rupture along a fault plane), 2) describing earthquake focal mechanisms (strike-slip, normal, and reverse styles of displacement), 3) evaluating the frequency, magnitude, and spatial distributions of present-day earthquakes, and 4) modeling how yesterday's earthquake changes the distribution of crustal stresses that will cause future earthquakes. Paleoseismology — the study of prehistorical earthquakes - utilizes many earth science disciplines including dendrochronology, geochronology, geodesy, geomorphology, seismology, soils genesis, stratigraphy, and structural geology. Tectonic geomorphology is indispensable for complete paleoseis- Figure 1.3 Locations of study sites in the western United States and northern Mexico and their book section numbers [5.5]. B, Pleistocene Lake Bonneville [5.2.3]; BL, Big Lost River [5.3]; BP, Borah Peak and the Lost River Range [3.3.4]; CD, Curry Draw [2.2.3]; CP Colorado Plateau; DR, Diablo Range [4.2.3.2]; DV, Death Valley, Panamint Kanqe, and Saline Valley; FR, Front Range [1.3]; GC, Grand Canyon [2.5.2]; GP Great Plains [1.3]; HL, Hebgen Lake [5.6.2]; l\R, Kings River [6.2.2.2]; L, Pleistocene Lake Lahontan [5.2.3]; LS, Laguna Salada [2.1]; MC, McCoy Mountains [3.2.2]; MD, Mojave Desert [4.2.3.2]; ML, Mount Lassen and the southern end of the subduction related Cascade volcanoes [4.1]; MR, Mogollon Rim [4.2.2]; NPR, North Platte River [1.3]; OM, Olympic Mountains [1.2.2, 5.5]; PIT, Pitaycachi fault [2.2.5, 5.4.2]; PR, Panamint Range, Death Valley, and 105 km2, >105ky >106 km2, >106 ky Isostatic uplift 5crunch and Stretch Tectonics Tectonic denudation or burial >105 km2, >105 km2, >105ky >105 ky Tectonic displacements Local lO'tolO3 km2, lO-'to >105ky Crustal 103to >104km2, 103to >104ky >----- Bedrock uplift Geomorphic processes Surficial 10zto >103 km2 10° to >103ky Subsurface 102to>104km2 10° to ^ky Net change of land-surface altitude Tectonic uplift Figure 1.4 Links between tectonic, isostatic, and nontectonic variables affecting landscape altitudes and bedrock uplift. Feedback mechanisms to isostatic and tectonic uplift are shown with dashed lines. earth deformation usually entails tectonic shortening (scrunch) and extension (stretch) processes. So, rocks move both vertically and horizontally (Willett, Slingerland, and Hovius, 2001). Do not expect crisp black-and-white definitions, because a model of fuzzy overlap is closer to the truth in the world of scrunch-and-stretch tectonics. Erosion of mountains is like rain falling on a marine iceberg; the height of both results from buoyant support. Rainfall can never melt enough ice to lower the surface of an iceberg to the water line. This is because ice melted above the waterline is largely replaced by "uplift" of submerged ice. Sea level is a handy reference datum for uplift of ice or mountain ranges. Uplifted materials may be above or below that worldwide waterline. Altitude is the specific term for height above present sea level, whereas the engineering term "elevation can have several geologic connotations, including uplift. Isostatic uplift occurs because ice is only 90 % as dense as seawater. If 100 tons is melted from the exposed surface of an iceberg, it is compensated by 90 tons of ice raised by isostatic uplift. This is pure uplift because it is not complicated by shearing or tensional failure of ice. Similarly, isostatic uplift of mountain ranges continues despite eons of surficial erosion because continental crust "floats" on the denser rocks of the Earth's mantle. Continental crust with a density of about 2,700 kg/m3 is in effect floating on mantle with a density of about 3,300 kg/m3- a density contrast of roughly 82% (90% contrast for oceanic crust with a density of 3,000 kg/m3). The iceberg analogy is appropriate because materials deep in the earth behave as viscous fluids over geologic time spans (Jackson, 2002). Fluvial and glacial denudation of 1,000 m only seems to significantly lower a mountain range because it is largely compensated by 820 m of concurrent isostatic rebound. Neither ice nor rock landscapes remain the same, unless erosional lowering is the same for all points in a landscape. Relief and altitudes of peaks increase if melt of ice, or erosion of rocks, is mainly along valley floors. Removal of mass above our sea-level datum causes pure isostatic uplift of all parts of the landscape. The average altitude of both the iceberg and the mountain range decreases with time because buoyancy-driven isostasy can never fully compensate for the mass lost by erosion. Chapter 1 A substantial proportion of mountain-range uplift is the result of these crustal isostatic adjustments (Molnar and England, 1990; Gilchrist et al, 1994; Montgomery, 1994; Montgomery and Greenberg, 2000). Isostatic uplift is both regional and continuous (Gilchrist and Summerfield, 1991), and generally does not cause pulses of renewed mountain building. This is done by scrunch and stretch tectonics. A major difference between icebergs and mountain ranges is that mountains do not float in a Newtonian fluid such as water, which has no shear strength. Continental rock masses float on hot litho-spheric materials whose rigidity provides some support. Rocks at shallower depths are stronger (cooler) and respond to changes in load by flexing in an elastic manner. Small, local changes in rock mass will not cause the lithosphere to flex because it has enough strength to support minor changes in load. But beveling of a 10,000 km2 mountain range will indeed influence crustal dynamics. Prolonged erosion has resulted in substantial cumulative isostatic rebound of the Appalachian Mountains of the eastern United States for more than 100 My. Tectonic geomorphologists would prefer to discern how different uplift rates influence land-forms and geomorphic processes, but reality is not that simple. Mountain-building forces may continue long after tectonic quiescence seems to have begun, as revealed by strath terraces (a tectonic landform discussed in Sections 2.4.1 and 2.6) in pretty dormant places like Australia (Bierman and Turner, 1995). Space and time frameworks of references vary greatly for the Figure 1.4 surface-uplift variables. Generally, they are large and long for pure uplift, tectonic denudation, or burial, and small and short for tectonic displacements and geomorphic processes. The predicament is that uplift has two components - tectonic and isostatic. Tectonic mountain-building forces may cease but the resulting isostatic adjustments will continue as long as streams transfer mass from mountains to sea. The best we can do at present is to observe landscape responses to the algebraic sum of tectonic and isostatic uplift. Bedrock uplift = Tectonic uplift + Isostatic uplift (1.1) This seems simple, until we attempt to quantify the Figure 1.4 variables that influence tectonic uplift and isostatic uplift. The term bedrock is used here in a tectonic instead of a lithologic context. Bedrock is any earth material that is being raised, with no regard as to the degree of lithification or age. We should note the "fuzziness" of this definition. Three exceptions are acknowledged; these occur when the nontectonic surficial process of deposition raises a landscape. The most obvious and dramatic is volcanic eruption, which raises landscape altitudes by depositing lava and tephra. Of course volcanic eruptions may also be associated with tectonic shortening and extension. Tectonic geomorphologists are interested in how climate change affects the behavior of streams in humid and arid regions. Mountain valleys and piedmonts undergo aggradation events as a result of major climate changes (Bull, 1991) that change the discharge of water and sediment. We do not class such stream alluvium as bedrock because its deposition is the result of a nontectonic process that raises valley-floor altitudes. Alluvium laid down before the particular time span that we are interested in would be treated like other earth materials, as bedrock. Studies of Pleistocene uplift would treat Miocene fluvial sand and gravel as bedrock. Thirdly, nontectonic deposition includes eolian processes such as the creation of sand dunes. Least obvious, but far more widespread, is deposition of loessial dust. In New Zealand windblown dust is derived largely from riverbeds after floods and the loessial blanket that covers much of the stable parts of the landscape may contain layers of volcanic ash, such as the 26.5 ka Kawakawa tephra (Roering et al., 2002, 2004). Hillslopes where this ash has been buried by 0.5 to 5 m of loess are landscapes where deposition has slowly raised the altitudes of points on the land surface during the 26 ky time span at average rates of <0.02 to >0.1 m/ky. Deposition - by volcanic ejecta, inability of a stream to convey all bedload supplied from hillslopes, and dust fall - is just one of several nontectonic geomorphic processes that change altitudes of points in a landscape (Fig. 1.4). I prefer to emphasize bedload transport rates in this book because bedload governs stream-channel responses to bedrock uplift. Rivers transport mainly suspended load to the oceans and deposit silty sand and clay on floodplains. Dissolved load is bedrock conveyed in solution. Both require little stream power, but the unit stream power required to mobilize and transport bedload reduces the energy available for tectonically induced downcutting of stream channels. Saltating cobbles and boulders are tools for abrasion of bedrock. With suspended load being Scrunch and Stretch Bedrock Uplift flushed downstream it is bedload that is deposited as fill stream terraces and many alluvial fans. Such land-forms are used to analyze responses of fluvial systems to bedrock uplift and to changes in late Quaternary climate too. Three classes of deposition in Figure 1.5 illustrate the care needed in defining sand and gravel as bedrock. The active range-bounding fault controls the behavior of this fluvial system. Stream channel processes normally switch abruptly from net erosion to net deposition after crossing the fault zone. Sandy alluvial-fan deposits of Miocene age have been elevated and now underlie watersheds in this hypothetical mountainous landscape. Most of us would agree that the uplifted Miocene fan deposits, although unconsolidated, should be classed as bedrock in a geomorphic sense. They are mountainous terrain into which drainage basins are carved. The gravelly fill terraces in the valley upstream from the mountain front are the result of climate-change perturbations. Without perturbations the watersheds of tectonically active mountain ranges in the Mojave Desert would have undergone uninter- rupted long-term degradation of their valley floors. But late Quaternary climatic fluctuations significantly affected sediment yield and stream discharge. Climate-change perturbations in arid and humid watersheds can temporarily reverse the tendency for stream-channel downcutting, even in rapidly rising mountain ranges. Climate-change perturbations are dominant because they quickly affect geomorphic processes throughout a drainage basin, whereas uplift on a fault zone is local and the resulting increase in relief progresses upstream relatively slowly. Climate-change induced aggradation events in the Mojave Desert raised valley floors <5 to >50 m. The range is largely due to lithologic controls on weathering and erosion. Aggradation was the result of insufficient stream power to convey bedload supplied from hillslopes whose vegetation changed drastically when the climate changed. Major aggradation events at about 125 ka and 10 ka were times of widespread stripping of hillslope sediment reservoirs that were no longer protected by dense growth of plants. A climatic perturbation at about 60 ka also coincides with a global sea-level highstand and caused an aggra- Age of aggradation event valley fill 111111 125 ka 60 ka 10 ka soi Figure 1.5 Summary of late Quaternary deposition for a typical Mojave Desert, California fluvial system where times and locations of aggradation are controlled by climatic perturbations that overwhelmed the effects of uplift along active fault zones. Hachures show soil profiles that postdate the ends of aggradation events and record brief Intervale of nondepoel-tion on the fanhead. 10 Chapter 1 dation event of smaller magnitude. So the sedimen-tology and thickness of each late Quaternary fill-terrace, and the concurrent increments of alluvial-fan deposition, were different (Table 1.2). Vertical separations between the beveled bedrock beneath several valley fills record stream-channel downcutting induced by uplift along the range-bounding fault zone in the intervals between climate-change induced aggradation events. The potent 10 ka aggradation event might have buried the equally strong 125 ka aggradation event at cross section A-A if there had been no tectonically-induced lowering of the valley floor. More tectonically induced degradation has occurred at cross section B-B' than at A-A because it is closer to the active fault zone. Depositional elevation of the stream terrace tread is a clear-cut example of nontectonic elevation of landscape elements. Such deposits should not be classed as bedrock. How should we regard the area of active alluvial-fan deposition downstream from the range-bounding fault? Surface ruptures on the normal fault create the space for continuing accumulation of basin fill. Such fans are tectonic landforms because nearly constant deposition would not have occurred without continuing uplift. Differential uplift along the fault has been sufficiently rapid to maintain late Quaternary aggradation adjacent to the mountain front (Section 4.2.2). Major Late Quaternary climatic changes caused the rate of fan aggradation to vary and influenced the locations of fan deposition. Minor, brief climatic fluctuations are superimposed on the long-term climatic controls. They caused brief episodes of stream-channel downcutting in the mountains and temporary entrenchment of the fan apex. Brief local cessation of depositional processes allowed incipient soil-profile development on the fan surfaces adjacent at cross section C-C. Each aggradation event was strong enough to backfill the fanhead trench, thus allowing fan deposition to continue to radiate out from an apex at the mountain front. It is debatable as to whether such fan deposits should be regarded as bedrock. Perhaps they should be classed as bedrock because the locations of fan deposition are tectonically controlled. Deposition of a thick fan would not occur here in the absence of active faulting. Alternatively, one might argue that rates of sedimentation vary with late Quaternary climates. Deposition merely tends to partially offset tectonic lowering of basin altitudes in an extensional terrain. Such fans should not be classed as bedrock. Lithospheric rigidity interjects the important element of scale into our perception of what constitutes uplift. Tectonic-uplift variables behave differently at the local scale of a single hillside or small watershed as compared to large chunks of the Earth's crust. For each point in a landscape, tectonic deformation caused by different styles of faulting and folding is superimposed on regional uplift (or subsidence) caused by broad warping of the lithosphere. This is a matter of different wavelengths for different earth-deformation processes. Alluvial geomorphic surface Aggradation age, ka Basis for age estimate Q4 Active washes, riparian trees, no rock varnish on cobbles Q3b ~0 14C dating of plant fossils, lake stratigraphy, rock varnish Q3a -12 14C dating of plant fossils, lake stratigraphy, rock varnish Q2c -60 230Th/234U ages of pedogenic carbonate, uranium-trend date, calibrated fault slip age estimate, cosmogenic 10[3e age estimate Q2b -125 230Th/234U ages of pedogenic carbonate Q2a 240-730 K/A dating of tuff, basalt flow, normal paleomagnetic polarities Q1 >\2D0 K/A dating of basaltic sources dissected into ridges and ravines Table 1.2 Pulses of climate-change induced alluviation in the Mojave Desert of California. Summarized from Tables 2.13 and 2.15 of Bull (1991). Scrunch and Stretch Bedrock Uplift 11 Tectonic uplift = Local uplift + Crustal uplift (1.2) The background regional crustal warping may be slow or fast, but it affects erosion rates of local land-forms as well as those of entire mountain ranges. Local faulting creates topographic anomalies such as rising mountain fronts that attract tectonic geo-morphologists (Chapter 3). We analyze landforms to separate local tectonic deformation from background regional uplift. But separating tectonic from isostatic uplift can be difficult at the watershed spatial scale because not all earth deformation is purely vertical. Scrunch and stretch tectonics plays an important role in deformation of Earth's crust. For example, plate-boundary subduction is a tectonic process, but how much of the resulting bedrock uplift is the result of isostatic uplift caused by thickening of the crust? How much is the result of scrunch induced by concurrent folding and thrust faulting? Conversely, in extensional terrains how much of a decrease in altitude is offset by isostatic adjustment resulting from concurrent erosion of mountain ranges? How much of lowering induced by stretch tectonics is offset by aggradation (Fig. 1.5) in basins that receive the deposits? Let's begin with brief summaries of the contents of the "Pure Uplift" and "Geomorphic Processes" boxes of Figure 1.4 to gain background before delving into "Stretch and Scrunch" box. Many factors affect magnitudes and response times for isostatic uplift. Important slow changes in the crust include accretion, or thinning, of light, buoyant crustal materials. Temperature increase or decrease changes the density of crustal rocks, thus changing their buoyancy. Phase changes in minerals that reflect changing pressures or temperatures alter buoyancy contrasts with adjacent rocks. Change to denser minerals decreases rock volume, which also tends to directly lower land-surface altitudes. Pure strike-slip faulting does not raise or lower a landscape, but major horizontal shifts of mountain ranges and crustal blocks may alter regional distributions of isostatic forces. Many plate-boundary strike-slip faults have cumulative displacements of more than 50 km, so this style of tectonic deformation may change the crustal loads on opposite sides of a fault sufficiently to cause isostatic re-adjustments. This important aspect of strike-slip faulting deserves its separate box within pure uplift tectonics in Figure 1.4. Changes in altitude that occur at bends and sidesteps of strike-slip faults, are included in the local- tectonic-displacement box because they are classified as scrunch and stretch tectonics. Transpressional or transtensional components of most plate-boundary fault zones also are best considered as part of scrunch and stretch tectonics. Tectonic processes and isostatic uplift may increase land-surface altitudes, but landscape altitudes also change because of several geomorphic processes. We have already mentioned the surficial processes of fluvial and volcanic deposition. Another is fluvial erosion, which tends to lower hills and streams. Both sets of processes affect crustal weight, and when sufficient may cause isostatic adjustments. Diagenesis of recently deposited basin fill tends to lower land-surface altitudes. Compaction of saturated clayey, silty beds in a sedimentary basin is analogous to crustal changes that produce denser minerals. It is pure vertical subsidence. Bulk density increases as water is gradually expelled from sediments by the weight of the overlying stratigraphic section, plus several hydrodynamic forces. The resulting decrease in bed thickness lowers the overlying strata and the land surface. Ground water derived from infiltrating rain and snowmelt dissolves minerals. Solution is a greatly different geomorphic process than landsliding because it is not visually conspicuous. It occurs below the land surface and the resulting ions are invisible in emerging clear springs that nourish streamflow. But substantial mass is removed over Quaternary time spans at depths that range from surficial soil profiles to more than 1 km. The net surface uplift resulting from all Figure 1.4 processes is an algebraic sum. Surface uplift = Rock uplift + Geomorphic Processes (1.3) The sum of geomorphic processes has feedback loops to isostatic uplift and tectonic deformation. Stretching and scrunching are important tectonic processes that lower or raise landscape altitudes. Most importantly, they (not isostatic uplift) initiate the creation of mountain ranges. Let us think of these as being tectonic denudation (Fig. 1.6) and tectonic burial (Fig. 1.8). Both are common, and operate at a variety of spatial scales. I'll focus mainly on scrunch processes because local uplift may appear anomalous when it is ten times the expected regional uplift. Also, it seems that tectonic denudation processes are already nicely discussed in the literature of the past two decades. 12 Chapter 1 Figure 1.6 Diagrammatic sketch of extension assc denudation and crustal thinning. Rollover folds increases closer to the normal fault. Frictional re block generates the shear couple responsible for d Stretch and Scrunch Tectonics Tectonic stretching (Fig. 1.4) is important. The resulting tectonic denudation is widely recognized, and generally is thought of as normal faulting that thins the crust (Armstrong, 1972; Davis and Coney, 1979; Shackelford, 1980; Spencer, 1984; Coney and Harms, 1984; Pain, 1985; Wernicke, 1992; Dickinson and Wernicke, 1997; Burbank and Anderson, 2001, p. 149-151). Normal faulting also occurs locally in compressional settings (Molnar and Lyon-Caen, 1988; Gammond, 1994; Eusden et al., 2005a). England and Molnar (1990) combined tectonic denudation and surficial erosion into a single process called "exhumation". Low angle detachment faulting (Lister et al., 1986; Bradshaw and Zoback, ;iated with normal faulting that causes tectonic :orm where gravitational collapse progressively istance during displacement of the hanging-wall ig folds next to the fault. 1988; Lee and Lister, 1992; Dokka and Ross; 1995; Bennett et al., 1999) can efficiently remove large amounts of bedrock, thereby promoting isostatic rebound (Wernicke and Axen, 1988). Normal faulting in the Basin and Range Province of the western United States has resulted in extension of more than 250 km (Wernicke and Snow, 1998), with a crust that has thinned to about 30 km (Jones et al., 1992). The lower crust of the Basin and Range province should behave as a viscous fluid (Bird, 1991; McCarthy and Parsons, 1994), tending to fill voids created by tectonic extension. Stretch tectonics has distinctive features and resulting landforms (Fig. 1.6). The footwall block typically has minimal secondary faulting, but ten- Scrunch and Stretch Bedrock Uplift sional forces create a myriad of antithetic and synthetic faults in the hanging-wall block. These result mainly from removal of vertical support. Complex structures are induced in hanging-wall blocks where normal fault dip becomes less with depth below the surface to create listric faults. This promotes fault-bend folding. Gravitational collapse is greatest near master detachment faults to create rollover folds (Hamblin, 1963, 1965). These tectonic processes lower surface altitudes. Local vertical displacements, such as range-bounding faults, create space that allows deposition of alluvial fans and other basin fill. Such aggradation raises surficial altitudes so deposition of basin fill is a process that partially offsets tectonic lowering, perhaps by a factor of half. Upper crust thinning enhances the potential for upwelling and isostatic uplift. Crustal rigidity extends this iso-static rebound into the footwall block at the left side of Figure 1.6, in an exponentially decreasing manner with increasing distance from the range-bounding fault. Spatially variable isostatic rebound tilts the land surface. Tectonic denudation caused by a variety of stretch processes thins the upper crust. These reduce crustal loading, and together with an increase in geothermal gradient and lithospheric upwelling promote isostatic uplift that partially offsets the stretch-induced subsidence (Bird, 1991). This self-arresting feedback mechanism is opposite of that caused by tectonic scrunching. The style of normal faulting affects the behavior of fluvial systems. The example used here examines stretch-tectonics controls on the thickness of piedmont alluvial fans. Continuing lowering of a valley and/or uplift of the adjacent mountains creates the space for new increments of piedmont deposition. The resulting alluvial fans reflect the style and rate of tectonic deformation. Prolonged displacement on a range-bounding normal fault can result in fan deposits more than 1,000 m thick. Alluvial-fan deposits are thickest where basins quickly drop away from the mountains, such as the high-angle normal faults of the Basin and Range Province of the western USA. Fan deposits are much thinner where tectonic displacements occur on low-angle faults. Examples include where thrust-faulted mountain fronts are shoved up and over adjacent basins along low-angle faults that dip back into the mountains (Section 3.2.3). I use many examples from the Death Valley region of southeastern California in subsequent chapters, so introduce an 13 interesting example of stretch tectonics here. The locale is the western flank of the Panamint Range. Low-angle normal faults have played an important role in both tectonic extension and landscape evolution of the Death Valley region. Style of alluvial-fan deposition varies with type of fault. Debate continues as to how important such detachment faults are as compared to normal faults that dip steeply at 45° to 65° (Wernicke, 1981, 1995; Walker et al., 2005). Cichanski (2000) made a detailed study of the cur-viplanar low-angle normal faults on the west flank of the Panamint Range that were first noted by Noble (1926) and Maxon (1950). As a geomorphologist, I have no doubt that normal faults that dip only 15° to 35° had substantial slip during the late Cenozoic. My premise is based on the idea that changes in the kinematics of faulting change the landscape. The evidence is the contrasting styles of alluvial-fan deposition. One would expect different types of alluvial fans resulting from low-angle and high-angle normal faulting. Adjustments of fluvial systems to movements on 60° and 25° normal faults are much different (Fig. 1.7). Slip on either steep or gentle fault surfaces causes fluvial systems to cut down into the footwall block and to deposit a new increment of detritus on the hanging-wall block. Part of the newly exposed fault plane is subject to the initial stages of dissection by water flowing in rills, and part is quickly buried by the newest increment of alluvial-fan deposition. Fan slope is also a function of magnitude and type of streamflow events, and the amount and particle-size distribution of the entrained sediment (Bull, 1962; Hooke, 1967). Although many alluvial fans in the Basin and Range Province slope less than 10°, steeper fans are common. Most fans along the Lost River fault zone near Borah Peak in Idaho have fanhead slopes of more than 20° (Section 3.3.4). An assumed fan slope of 20° for the ancestral fans along the western flank of the Panamint Range seems reasonable for this discussion. Thicknesses of tectonic alluvial fans are a function of fault dip and fan slope. The combination of a 60° normal fault and a 20° fan surface provides ample space for thick deposits to accumulate adjacent to the footwall block. Fan thickness in the Figure 1.7B example is 40 m, and would be the maximum of 50 m if the range-bounding fault were vertical. Steep faults are sites of thick fans of small areal extent. Extension on high-angle normal faults also favors incision of deep valleys in the footwall block. 14 Chapter 1 Low-angle fault 13 High-angle fault Most recent increment of alluvial-fan deposition N i X X Most recent increment of valley deepening Lx-z=220 m Lx-z=57 m Figure 1.7 Diagrammatic sketches showing how change from low-angle to high-angle normal faulting changes \ar\Aeca^e characteristics. H is slope fall, and L is slope length horizontal distance. Vertical tectonic displacements, Hx-z, total 100 m in both cases as the footwall block slips from X to Z. Horizontal tectonic displacements, Lx-z, of 57 and 220 m are a function of normal fault dip. B is the present threshold-intersection point where erosion changes to deposition, assuming that the increase of relief of the footwall block is distributed evenly between alluvial-fan deposition and valley deepening. A. 25° normal-fault dip and a 20° fan slope. B. 60° normal-fault dip and a 20° fan slope. In contrast, only thin veneers of deposits accumulate on a 20° sloping fanhead in response to movements on a 25° low-angle normal fault. Fan thickness in the Figure 1.7A example is only 12 m, but the width of the newest increment of onlapping fan deposits is 110 m - four times that of the high-angle fault example. Such low-angle faults are sites of thin fans of large areal extent. An emphasis on horizontal instead of vertical displacement also inhibits erosion of deep valleys in the footwall block. These shallow valleys are part of a diagnostic landscape assemblage suggestive of low-angle normal faulting, as are the smooth sloping hillsides that resemble the carapace of a turtle, the "turtlebacks" of Wright et al. (1974). The thinnest deposits near the intersection point (where erosion changes to deposition) are readily removed by fluvial erosion after deposition ceases. Such erosion may have occurred along the west flank of the Panamint Range, and elsewhere in the Death Valley region. Initiation of steep range-bounding faults in the Pleistocene that cut the now inactive low-angle faults (Cichanski, 2000) would stop deposition of the ancestral fans and begin the process of eroding them. The combination of incremental exposure of the plane of a low-angle fault while it is active and subsequent partial stripping of a thin mantle of fan deposits results in spectacular rilled fault planes (Fig. 2.19A). Scrunch deformation is everywhere in hanging-wall blocks of thrust faults. In addition to synthetic and antithetic faulting, scrunch processes include folding, flexural-slip faulting along bedding planes, and shoving of wedges of crumpled, brittle rocks up gently inclined fault planes. Scrunch style tectonics may dominate locally to the extent of raising surface altitudes an order of magnitude faster than regional uplift rates. It makes for pretty messy earth deformation, but adds much variety to rock uplift (Fig. 1.8). The belt of former piedmont terrain between the two thrust-fault zones is called a piedmont foreland, the topic of Section 3.2.3. Bedrock uplift resulting from scrunch tectonic processes increases landscape altitudes and relief of mountains, thus accelerating erosion that partially offsets regional uplift. Scrunch processes may promote lithospheric downwelling opposite in style to the mantle upwell-ing described for tectonic stretching. Deposition in tectonic basins raises altitudes. Scrunching and Scrunch and Stretch Bedrock Uplift 15 Flexural slip folds and faults with displacements along bedding-plane faults Abandoned streamcourse is now a wind gap Deflected stream flows through a water Figure 1.5 Diagrammatic sketch of types of contractions! faulting and folding associated with tectonic shortening that causes burial and crustal thickening. Overturned strata may suggest displacement by a normal fault. Displacements along bedding-plane faults occur where planes between beds are relatively weak; note rock flowage into fold axes. Thrust faulting buries the apex of the piedmont alluvial fan, and a younger fault folds the fan surface. deposition thicken the crust, thereby promoting iso-static subsidence that partially offsets concurrent rock uplift. Geothermal gradients become cooler where the crust is thickened from the surface down, and the relatively cooler rocks have a lesser potential for iso-static uplift. Tectonic burial has not received as much attention in the literature as tectonic exhumation, so I use Figures 1.9-1.16 to illustrate the diversity and importance of scrunching. Creation of a fault zone causes more than just uplift, because thrust faults are not vertical. Horizontal rock displacement is a major consequence of scrunching. The hanging wall block is raised as it is shoved up the incline of a gently dipping thrust fault (Fig. 1.9A). The horizontal component of displacement increases local crustal thickness. Amounts of horizontal displacement are a tangent function of fault-plane dip: 100 m of vertical dis- placement is accompanied by only 27 m of horizontal displacement for a 75° dipping fault. This increases to 100 m for 45° and to 373 m for a fault with a 15° dip. Mass is added to the footwall-block terrain by tectonic conveyance and deposition of sediments eroded from the newly raised block as the fault trace advances in an incremental manner. In the best of all worlds, tectonic geomor-phologists would use planar or conical landforms as time lines passing through tectonically deforming landscapes. Dating of faulted alluvial geomorphic surfaces can provide valuable information about late Quaternary uplift rates. However, estimation of tectonic displacement rates of faulted stream terraces probably is more reliable for stretch than for scrunch tectonics. The fan surface upslope from the scarp crest in Figure 1.9B is no longer linear. Its undulations sug- 16 Chapter 1 Footwall 0 400 m Figure 1.9 Tectonic uplift and burial Induced by thrust faulting. A. Diagram showing components of uplift and burial created by movement along a thrust fault. Both processes thicken the crust and are functions of fault-plane dip. gest complicated tectonic deformation. A common first impression is that scarp height is indicative of the magnitude of tectonic throw, but scarp height exceeds true displacement where sloping alluvial surfaces are ruptured. A closer approximation can be obtained by noting the vertical separation of projections of the tectonically undeformed fan surfaces upslope and downslope from the fault zone. But this Cucamonga Canyon alluvial fan has a slope that decreases down-fan resulting in lack of parallelism of the projected surfaces. A mean apparent throw of 9.3 m based on maximum and minimum displacements is triple the deformation attributable to scrunching. These apparent displacements need to be corrected for the dip of 720" the faults, which is unknown. A complete discussion is deferred until Section 3.3.4, which describes how to estimate throw for normal-fault scarps on alluvial fans. The interpretation shown in Figure 1.9B is that several synthetic thrust faults ruptured the surface, during several Holocene earthquakes (Morton and Matti, 1987). Another possibility is that the hum-mocky terrain is nothing more than piles of debris near the fault tip that have been bulldozed by thrust faulting along a single thrust fault. Third, compression may have folded the surficial materials. Most likely, the scrunched material resulted from several processes. Holocene bedrock uplift varies from point to point, but approximates the sum of the vertical Apparent vertical tectonic displacements (throw) 1 Scarp height 2 Scrunching uplift component 3 Minimum fan surface offset 4 Maximum fan surface offset 5 Mean rock uplift 100 200 Distance in meters Figure 1.9 Tectonic uplift and burial induced by thrust faulting. E3. Inferred thrust faults along cross section based on topographic profile. All estimates of displacements are apparent, and except for scarp height are based on projections of adjacent undeformed alluvial-fan surfaces upslope and downslope from the fault zone. Cucamonga alluvial fan, San Gabriel Mountains, southern California. Figure 1.10 Deformation of the stream terraces of the Waimangarara River caused by recent surface ruptures of the range-bounding Hope thrust fault, Seaward Kaikoura Range, New Zealand. The young T1 stream terrace is strongly backtilted, and is anomalously high when compared with estimates of late Quaternary uplift rates for this mountain front. component of thrust-fault displacement, folding, and other scrunching that results from compressional deformation of the wedge of material above the thrust fault. The magnitude of horizontal displacement determines the amount of tectonic burial. The algebraic sum of these processes equals the changes of surficial altitudes because this young alluvial fan is virtually uneroded. Isostatic adjustments are not likely at this small scale. Tilted stream terraces are sure to catch the attention of the tectonic geomorphologist, especially when alluvium deposited with a 3° downvalley dip now slopes 2° to 5° upvalley (Fig. 1.10). A splay of the Hope fault that bounds the Seaward Kaikoura Range of New Zealand ruptured the Waimangarara River stream terraces. The two oldest, late-Holocene, stream terraces, Tl andT2, have the same backtilt, so the tectonic deformation is younger than the T2 terrace-tread age. Terrace Tl is 5 m above T2. Terrace tread age was estimated with weathering rind analyses, a surface-exposure dating method (Whitehouse and McSaveney, 1983; Whitehouse et al., 1986; Knuepfer, 1988). Analysis of boulders on the T2 tread implies a late Holocene age (Fig. 1.11). This tuffaceous greywacke sandstone does not have nice, sharp weathering rinds, and rind thickness ranges from 1 to 4 mm. I used the McSaveney (1992) procedure. A peak at -2.5 mm dates as 2,200 ± 300 years before present. Even a 4 mm peak would date to only -4,700 years B. P. Terrace T3 is not backtilted but has a fourfold decrease in slope as it approaches the deformed older stream terraces (Fig. 1.10). So it appears that the range-bounding fault ruptured between T2 and T3 time, and again since T3 time. t Rind thickness, mm Figure 1.11 Distribution of Waimangarara River T2 stream terrace tread weathering rinds in cobbles of greywacke sandstone deposited before the older of two recent surface-rupture events. Normal distribution curve has been added. 0.25 mm class interval, n = 40. 18 Chapter 1 Zone of deformed landforms Minor Figure 1.12 Geologic structures, landforms, and tectonically deformed stream terraces at the mouth of the Waimangarara River, Seaward Kaikoura Range, New Zealand as mapped by Jarg Pettinga, University of Canterbury. Scarp height is an impressive 18 m. Vertical offset of 7 m is a minimum value because Tl and T2 have been buried by an alluvial fan downstream from the fault scarp. The prominent graben at the folded scarp crest (Fig. 1.12) can be used to postulate locations of antithetic and synthetic faults above the master thrust fault, which is presumed to dip less than 50° (Van Dissen, 1989). I suspect that neither the large scarp height nor upvalley stream-terrace tilt is indicative of slip rates on this segment of the Hope fault. The Tl fault scarp on the other side of the river is only about 3 to 4 m high, which is a more reasonable offset for two surface-rupture events. Adjacent segments of this mountain front lack high fault scarps that date to the most recent event. Dip and style of faulting may change within short distances, and subsurface exploration techniques are needed here to fully appraise two possible scenarios. The geologic map (Fig. 1.12) portrays a zone of deformation that tapers towards the southwest, seems to be diffuse on the north side, and is abruptly terminated by the range-bounding thrust fault on the south side. A model of imbricate thrust faulting (Fig. 1.13A) can account for the width of the deformation zone, and synthetic and antithetic faults could produce grabens. If one uses the critically tapered wedge model of Davis and Namson, (1994), the scrunching shown in Figure 1.13B reflects a fault-kinematic equilibrium. Wedge shape would influence dip of the basal detachment surface, synthetic and antithetic fault movements, and thickness of scrunched rock and alluvium. So, much of the rock uplift here may be the result of tectonically induced scrunching processes of folding and bulldozing. Brittle fractured greywacke sandstone under low confining pressures may behave like loose boulders. More coherent bedrock slabs may Scrunch and Stretch Bedrock Uplift 19 220 . ^^r-T^K 200 ISO -t- =^---=^""V.:v-ffgL 100 200 300 400 Horizontal distance, m 500 600 150 100 200 300 400 500 Horizontal distance, m 600 Figure 1.13 Models for deformation of the Waimangarara River terraces. A. Fault steepening towards the surface rotates the stream terraces, creating the backtilting of T1 and T2. Grabens at scarp crest record antithetic and synthetic faulting. fail by rupture along secondary faults. The systematic deformation shown in Figures 1.12 and 1.13B could result from folding instead of haphazard bulldozing processes. Scarp-crest grabens would result from tensional stresses at the crest of an anticline in a folding-dominant model. The Waimangarara River has frequent large flow events that deposit bouldery alluvium on the adjacent piedmont. Erosional widening of the bedrock valley floor in the mountain-front reach may have thinned the slab above the range-bounding thrust fault prior to the recent surface rupture events. Reduction of rock mass strength below a critical-tapered-wedge threshold would have favored tectonic scrunching processes in the broad valley floor upstream from the fault trace, but not along the adjacent parts of this steep mountain front. Rock uplift, r , at location 1 in Figure 1.13B is mainly a function of magnitude of slip along the fault plane, D, and dip of the thrust fault, a. ru = sinaD (1.4) Rock uplift at location 2 in Figure 1.13B could be largely bulldozed materials above the plane of the thrust fault where scrunch rock uplift, sr , has occurred at several fault splays. ru = sinaD + sr11(1,2,3,4) (1.5) The longitudinal profile of the Waimangarara River reflects several possible tectonic inputs. The Figure 1.13 Models for deformation of the Waimangarara River terraces. f3. Critical wedge model in which movement along gently dipping thrust fault has bulldozed and/or folded the fractured greywacke sandstone. stream changes its vertical position in the landscape in response to bedrock uplift. However, fluvial adjustments to rock uplift in the longitudinal profile do not distinguish between regional isostatic uplift, slip on thrust faults, folding, and local bulldozer scrunching of fractured greywacke sandstone. I conclude that the Waimangarara River stream terraces are not ideal time lines passing through a tectonically deforming landscape. The deformed stream terrace treads are good reference surfaces for describing the complicated total bedrock uplift, but should not be used for estimating fault slip rates. Thrust-fault displacement has a vertical component, but secondary folding and crushing is largely a function of horizontal displacement. Both contribute to rock uplift. This local increase in crustal loading due to scrunching is too small to overcome lithospheric rigidity, so let us examine an example that is sufficiently weighty to influence isostatic processes. Erosion becomes ever more important with increase in spatial extent and steepness of a landscape, longer time spans, and decrease of rock mass strength. Erosion rates increase exponentially with hillslope steepness (Ahnert, 1970), so relief that is increased by scrunching accelerates the denudational processes that tend to lower a mountain range that is being created by tectonic forces. One impressive example of large scale thrust faulting, and rapid erosion during the past 1.5 My, is the Salt Range in Pakistan (Burbank and Anderson, 2001). Potential tectonic burial by a tectonically translocated mountain-range size block that is 3 km high and 18 km long (Fig. 1.14) never transpired because erosion occurred as rapidly as scrunching raised poorly consolidated fluvial sediments of the Siwalik Formation 20 Chapter 1 North 20 km South Volume of foreland fluvial strata that has been eroded away as fast as soft fluvial sediments have been thrust up the footwall ramp Basin fill deposited as foreland strata are eroded Basin fill alluvial-fan conglomerate Siwalik Formation fluvial molasse deposited in Himalayan foreland basir Strata of Paleozoic to Eocene age Salt of Paleozoic to Eocene age Pre-Cambrian basement rocks Figure 1.14 Potential large scale tectonic burial that has been offset by erosion of the Salt Range, Pakistan. From Burbank and Anderson, 2001, Figure 7.5, and 3urbank and Beck. 1991. up the ramp of the footwall block. Note the lack of a tapered wedge of mountain range relief away from the top of the ramp. The Salt Range tectonic setting appears to represent a case where regional fluvial erosion balances the tendency for uplift to increase relief. Estimated rates of erosion are 2 m/ky over the large area of 1500 km2. Tectonic loading has been largely offset by concurrent erosion, which increases crustal thickness elsewhere in depositional basins. Active folding provides extreme examples of scrunch-induced bedrock uplift even where rates of regional uplift are modest. Spatial contrasts in uplift rates generally are gradual for active folds and abrupt for active faults. Horizontal strata under a constant rate of tectonic shortening are folded upward, but the crest of the resulting anticline does not rise at a uniform rate. Rockwell et al. (1988) show that uplift for a single, simple fold quickly accelerates to a maximum, and then slows to zero despite unabated com- pression (Fig. 1.15). About 36% of potential uplift has occurred after only 4% shortening of a horizontal bed, a situation where a slow rate of horizontal displacement causes remarkably rapid uplift of the fold hinge. But there is a limit to how much uplift can be produced by contraction (shortening) of a single fold. Continued scrunching creates faults and new folds. Anticlines in fold and thrust belts commonly have thrust faults in their cores (Fig. 1.8), which further complicates assessment of bedrock uplift. Folding may be largely replaced by tilting after a thrust fault propagates through to the land surface. We should expect the landforms and geomorphic responses to tectonic deformation to vary along the trace of fold created by a propagating thrust fault (Fig. 1.8). Spatial variations of local tectonic deformation should reflect the cumulative displacements of individual earthquakes. Level-line surveys of recent historical earthquakes nicely show the contrast in Fold hinge 16% shortening 4% shortening 0% shortening Figure 1.15 Tectonic uplift and burial induced by contractional folding. Deceleration of rates of folding induced uplift, using a model of uniform rates of tectonic shortening (from Rockwell etal., 1955). -20 0 20 Distance from fault Figure 1.16 Spatial variations in deformation caused by two magnitude M 7.3 earthquakes. The thrust-fault example is the 1952 Kern County, California earthquake. The normal fault example is the 1953 Borah Peak, Idaho earthquake. Note that both subsidence and uplift occuv with extensional and contractional earthquakes (from Stein, et al., 1955). styles of folding and faulting associated with normal and thrust faulting (Stein et al., 1988). Such work reveals that some local uplift occurs during a normal-fault surface rupture and some subsidence occurs with a thrust-fault rupture event (Fig. 1.16). Rebound uplift of the footwall block was about 20% that of the hanging-wall block tectonic subsidence result- Scrunch and Stretch Bedrock Uplift 21 ing from the Borah Peak earthquake. Minor faulting and folding is concentrated in the hanging-wall block of normal faults, but the footwall and hanging-wall blocks of an active thrust fault may have similar secondary deformation. Not all tectonic deformation occurs at the moment of an earthquake due to the response time needed for subsequent mantle upwelling. Fault creep and gradual folding may also deform the land surface. Modeling done by Freed and Lin (2002) links tectonic deformation to post-seismic relaxation of viscous lower crust and/or upper mantle - a process that continues for decades. Although some folding or warping during a particular earthquake event occurs as post-seismic deformation, few studies have the data to assess both pre- and post-seismic folding rates over time spans of centuries. One such investigation uses stream terraces formed as a result of downcutting induced by folding and faulting of an anticline. Streams flowing across rising mountains incise bedrock and tectoni-cally induced downcutting is proportional to bedrock uplift rates. Nicol and Campbell (2001) estimated uplift rates for an anticline by measuring the heights of terrace treads above the active channel, and by using weathering-rind and radiocarbon methods to date abandoned floodplain remnants. The scene is a young fold-and-thrust belt in the foothills of the Southern Alps of New Zealand. The Waipara River has a watershed area of 950 km2 where it cuts though Doctor's Anticline. The Karetu thrust fault in the core of the anticline has broken through to the surface. Regional tectonically induced downcutting has been subtracted from the total tectonic displacement measured in the anticline reach to produce the graph of Figure 1.17. The highest terrace has been raised 23 m relative to the active channel but is only about 600 years old. Downcutting curves for two reaches of the stream channel show that accelerated downcutting occurred between 0.6 and 0.2 ka. The mean local bedrock uplift rate was an astonishing 52 m/ky! This example of extreme scrunching is 50 times the estimated Holocene uplift rate for this tectonic province. Nicol and Campbell also use the terrace ages and heights (Fig. 1.17) to assess the temporal distribution of uplift before and after an earthquake on the Karetu thrust fault that occurred 350 ± 50 years ago. Maximum uplift occurred in that century-long time span. The steep sections of the tectonically induced downcutting plot between 0.6 ka and 0.35 ka and 40 km 22 Chapter 1 50 £ 40 -SI -SI ^3 30 <33 W £ W U 20 <33 w -SI 5 10 0 -1-1-1-1- Cessation of inferred aseismic folding Thrust fault rupture event Onset of inferred aseismic folding 0 0.2 0.4 0.6 Terrace age 1.0 ka between 0.35 ka and 0.2 ka are inferred to be the result of aseismic folding. These several examples provide interesting food for thought about scrunching that results in vertical tectonic displacements but tell us very little about magnitudes of horizontal earth deformation. Figure 1.17 Variable rates of tectonically induced stream channel downcutting caused by a late Holocene folding event on a thrust-cored anticline in the foothills of the Southern Alps, New Zealand. The magnitude of tectonic deformation shown here is a minimum because stream terraces are not present at the crest of the anticline. From Figure 12r3 of Nicol and Campbell (2001). Marine terraces provide an opportunity to assess horizontal as well as vertical movements of rocks. Pazzaglia and Brandon (2001) provide an elegant example in their discussion of the tectonic land-forms of the Cascadia forarc high (Fig. 1.18). They examine coastline landforms where the Juan de Fuca plate is converging with the North American plate at 3.6 m/ky at a bearing of 54 °. Shore platform-sea cliff landform couplets generally are created only at times of prolonged sea-level highstands, such as the present. The horizontal and vertical distances between modern and ancient inner edges of marine terraces differ as a function of the horizontal and vertical rates of tectonic displacement. Sea level has also varied in a eustatic (world 2000 150 0 1000 500 0 -500 -1000 -1500 -2000 Distance parallel to plate convergence direction (m) Figure 1.1Ö Horizontal and vertical displacement of marine-terrace landforms on a subducting plate boundary. Schematic cross section across the modern shore platform, sea cliff, 122 ka Sangamon shoreline, and a partially burled sea cliff. Mouth of the Queets River, Olympic Mountains of northwestern Washington, U5A. Figure 17 E3 of Pazzaglia and brandon (2001). Scrunch and Stretch Bedrock Uplift 23 41 < 50 0 -50 -100 -150 ßO 160 240 Time Before Present, ky 320 Figure 1.19 Fluctuations of global sea level since 330 ka (from Chappell and Shackleton, 1956). wide) sense (Fig. 1.19) due to changing volumes of glacial ice and ocean temperatures (Shackleton, 1987). The Sangamon sea-level highstand at about 124 ka was about 5 m above the present ocean level. This means that even under tectonically inactive conditions the Sangamon shoreline for this gently sloping coast should be 5 m higher and quite far inland. Marine terraces were created at times of globally synchronous sea-level highstands (Chappell, 1983; Chappell and Shackleton, 1986; Lambeck and Chappell, 2001). The -124 ka terrace has been uranium-series disequilibrium dated using coral from New Guinea, New Hebrides, Barbados, Haiti, the Mediterranean Sea, Hawaii, Japan, and California. Bloom et al. (1974) estimated the altitudes of many sea-level highstands. These were brief time spans of unchanging terminal base levels for rivers, much like the past 6 ky. Remnants of shore platforms created at a variety of sea levels indeed are useful time lines passing through tectonically active landscapes. Rapid sea-level changes between the highstands raised and lowered the mouths of streams but this is not the same type of base-level fall as faulting of a streambed (Sections 2.2.4, 2.6). The Figure 1.18 analysis assumes similar geo-morphic processes and alluvium mass strength (gravels) at Sangamon time as compared to the present. Because of the higher eustatic sea level the Sangamon highstand sea cliff would have formed at 945 ± 145 m inland of the modern sea cliffs. Instead it is located an additional 505 ± 150 m farther inland. This suggests a horizontal tectonic displacement of about 450 m during the past 122 ka or a mean horizontal tectonic velocity of 3.7 ± 1.1 m/ky. The eustatic component is much smaller than the tectonic component for uplift during the same time span and is simply the change in world-wide sea level. 1.3 Landscape Responses to Regional Uplift Streams incise ever deeper as bedrock is raised into the powerful buzz saw of stream-channel downcutting. Amounts and rates of tectonically induced downcutting are functions of vertical tectonic displacement rates, excess unit stream power, and resistance of earth materials to degradation. Downcutting by small streams flowing over resistant welded tuff may be unable to match a bedrock-uplift rate of 0.1 m/ky; such reaches erode continuously. Downcutting by perennial rivers flowing over soft rock easily keeps pace with bedrock uplift of 5 m/ky. But stream-channel downcutting occurs only during appropriate climatic and tectonic conditions. The tendency of streams to cut down to the minimum gradient needed to transport their sediment load has been a long standing fundamental concept in fluvial geomorphology (Powell, 1875; Mackin, 1948; Leopold, Wolman, and Miller, 1964; Leopold and Bull, 1979; Bull, 1991). Headwater reaches of streams in rising mountains tend to stay on the degradational side of the threshold of critical power, but downstream reaches, with their greater unit stream power, are more likely to attain the base level of erosion through the process of tectonically induced downcutting. Gregory and Chase (1994) minimize the influence of base level in a diametrically opposite model. They conclude that Cenozoic canyon cutting in the Front Range of the Rocky Mountains in Colorado resulted entirely from climatic changes that increased stream power. Molnar and England (1990) also favor dominance of climate-change causes of stream-channel downcutting in this same region. The resulting isostatic uplift would promote further canyon down-cutting. Zaprowski et al., (2005) prefer a model where climatic changes would increase the concavity of the longitudinal profiles of rivers crossing the western Great Plains. Greater concavity would require more intense rainfalls and larger, more frequent flood events in the Quaternary than during the Pliocene. However, analysis of gradient changes of rivers flowing east from the Rocky Mountains (Figs. 1.20A, B) reveals that flexural isostatic rebound of the lithosphere due to Cenozoic erosional unloading accounts for only 20% of the concurrent increase of relief (McMillan et al., 2002). Therefore, tectonic uplift is necessary in order to explain the Front Range canyon cutting and the concurrent deepening of val- 24 Chapter 1 leys across the adjacent Great Plains. McMillan et al. conclude that post-depositional changes in slope of the stream channels in the western Great Plains of Wyoming and Nebraska since 18 Ma are the result of broad-wavelength tectonic uplift centered under the Rocky Mountains. Uplift began during deposition of braided-stream gravels of the Miocene Ogallala Formation. Tectonically induced downcutting has continued to lower the active stream channels relative to the strath beneath the basal Ogallala fluvial gravels. It is not easy to discern uplift in landscapes that lack obvious Quaternary faulting and folding. So this book emphasizes tectonic influences on the landscapes of individual watersheds, preferably where tectonic controls are obvious such as active range-bounding faults. McMillan et al. were able to estimate regional tectonic influences on landscape evolution with a combination of paleohydrologic, strati-graphic, and geophysical analyses involving a spatial scale of 250 km and a time span of -15 My. This challenging project produced some interesting results. The post Ogallala time span coincides with the gradual northward extension of the Rio Grande rift from New Mexico; and it seems reasonable that the accompanying regional tectonic uplift also decreased towards the north. Tectonic rock uplift was followed by an episode of erosion-induced isostatic uplift that began when the rivers of the region ceased deposition and began 5 My of fluvial degradation. Leonard (2002) analyzed the larger valleys draining the eastern flank of the Rocky Mountains. Uplift caused tectonically induced downcutting, which promoted isostatic uplift. He assumes that the base of the Ogallala formation was planar and tilted eastward. The Arkansas River valley in southeastern Colorado was eroded to deeper levels than the valley of the North Platte River in southeastern Wyoming. Maximum warping of the Colorado piedmont occurred near the Arkansas River. Leonard's modeling suggests that the isostatic component of rock uplift (Fig. 1.4) accounts for 50% of the total rock uplift with the remainder being tectonic uplift. About 540 m more uplift occurred along the Figure 1.20C transect at the Arkansas River valley than at the valley of the South Platte River. Leonard's results approximate the lesser uplift amounts suggested by McMillan et al., (2002) along the valley of the North Platte River. These thoughts about the diverse character of uplift will be used when we explore uplift of specific mountainous landscapes. Chapters 3 and 4 go into more detail regarding the complications that arise when one attempts to determine how fast the mountains are rising. Figure 1.4 is a rudimentary summary. It hits the main points, but the influences of many of the variables are not easily constrained to the tidy boxes of this simple model. Deep seated crustal flow (Zandt, 2003) is largely ignored. The isostatic component of rock uplift is the algebraic sum of many processes. It is a function of crustal temperature or mineralogy changes, crustal subduc-tion, spreading and flexing, strike-slip fault loading changes, tectonic denudation or burial, faulting and Figure 1.20 Late Cenozolc tectonic and leo-static uplift of the Colorado piedmont east of the Rocky Mountains. A. Trunk channels of the major rivers flowing eastward across the western Great Plains. Rio Grande is in a rift valley that has propagated northward during the Cenozoic. Scrunch and Stretch Bedrock Uplift 25 0 50 100 150 200 Kilometers east of 105° W longitude 250 2,500 to <ü in o TO TO t o in w ts □ p p ^ 1,500 2,000 " < o 1,000 " 1 1 1 South i i North y? ... '*.......... North Canadian Arkansas South Platte Platte Riyer , , River , , River, .River, 1,000 500 0 □ ü P TO P IS! O is] ts _ Q i 3 y = 2.ßxl0-ao5ox : 1 R = 0.99 1 5 r 10 r , 15 Distance upstream from fault, m 0.01 ic y = 0.50 x 10"' R = 0.98 5 10 15 Distance upstream from fan apex, m a. 0 y = 1.9x10"' R = 0.98 5 10 15 Distance downstream from fault, m Figure 2.12 Spatial decay of an uplift perturbation, for a 100-year time span, along a stream that was displaced 3 m vertically in 1557 by rupture along the Pitay-cachi normal fault, 5onora, Mexico. A. Diagrammatic sketch of a small Pitay-cachi fault study area. F3. Decreasing amounts of tectonically induced downcutting in the reach upstream from the fault. C. Decreasing amounts of tectonically induced aggradation in the reach upstream from the apex of an alluvial fan that is 30 m downstream from the fault because of pre-1557 stream-channel downcutting. D. Decreasing amounts of tectonically induced downcutting in reach downstream from the fault. 1887 uplift event increased the relief and sediment yield of the small watershed. Stream power increased because of the steeper gradient, but resisting power increased even more (terms are defined next in Figure 2.13). Aggradation in the fanhead trench reach is the combined result of increased watershed sediment yield and base-level rise caused by alluvial-fan deposition. The resulting post-1887 tectonically induced aggradation was described by measuring thicknesses of modern gray alluvium above reddish brown weathered Pleistocene deposits along the 14 m of the partially backfilled entrenched stream channel. Spatial changes in thickness of aggradation are systematic (Fig. 2.12C). The Ku value is 0.124. Channel depth decreases systematically in the entrenched stream channel (Fig. 2.12D); Ku is 0.224. We continue by introducing the topics of stream power, resisting power, and thresholds in geo-morphic processes. These concepts provide a base to examine topographic profiles of stream channels that have attained equilibrium configurations. 2.3 Threshold of Critical Power in Streams Altitude and topographic relief increase where bedrock uplift exceeds surficial erosion (equation 1.3). Bedrock uplift increases valley-floor slope, and thereby the stream power available to transport bed- 40 Chap load and to erode valley floors. Bedrock uplift is an independent variable of fluvial systems where it is not affected by other geomorphic variables. Slopes of hillsides and streambeds are dependent variables because they are affected not only by tectonic processes, but also by independent variables such as climate and rock type, and by dependent variables such as soils, plants, and erosional processes. Perhaps we should use the less restrictive term "controlling" instead of "independent" for time spans of more than 1 My. For example, uplift of the Himalayas during the past 35 My has controlled climate on a global scale (Raymo and Ruddiman, 1992). Lithologic control of erosional landforms is associated with spatial variations in structure, fabric, and mineralogy of rocks (Weissel, and Seidl, 1997). The time span being considered influences how we classify variables (Schumm and Lichty, ter 2 1965). Climate should be considered as an independent variable for short time spans, but over long time spans orographic influences of bedrock uplift make watershed climate dependent on bedrock uplift. Climatic control of fluvial systems is so pervasive that a companion book is devoted to the fascinating subject of Geomorphic Responses to Climatic Change (Bull, 1991). Changes to plant communities are central to climate-change impacts (Vandenberghe, 2003; Flenley and Bush, 2006). Late Quaternary climatic changes commonly overwhelmed the effects of concurrent bedrock uplift by abruptly changing the amounts of water and sediment supplied to streams. As a tectonic geomorphologist, you need to separate tectonic from climatic influences on landscape evolution. Hemisphere-scale climate variations affect both styles and rates of erosion of mountain ranges, such as the Andes (Montgomery, 2002). INCREASING RESISTING POWER Sediment eize INCREASING STREAM POWER Stream gradient 500 Coarse T 0 01 Fine 0.0G ftpuuN_O.LV fimpath Hydraulic Rouqhnese Stream discharge 3 I < LLl &Ü (J z Figure 2.13 Schematic balance between modes of aggradation and degradation in streams; zero is the threshold of critical power. Increases or decreases of one or more variables may cause the mode of stream operation to depart markedly from a threshold condition. (Originally from notes of E.W. Lane; modified from Chorley et al., 1954 and Bull, 1991.) Concepts for Studies Streamflow behavior may be regarded as a delicate balance between several controlling factors -stream slope and discharge, sediment size and amount, and hydraulic roughness (Fig. 2.13). The threshold of critical power separates two disequilibrium modes of operation in streams, degradation and aggradation. This threshold is defined as a ratio (Equation 2.4) where the numerator consists of those variables that if increased tilt the balance in favor of degradation, the mode of operation that lowers the altitude of a reach of a stream by fluvial erosion. The denominator consists of those variables that if increased tilt the balance in favor of aggradation. Sustained aggradation raises the altitude of a reach of an active stream channel by selective deposition of bedload. Stream Power (driving factors) Resisting Power (prohibiting factors) = 1.0 (2.4) The available stream power to transport bedload (Ferguson, 2005) as defined by Bagnold (1977) is Q = jQS (2.5) where Q is the total kinetic power, or in terms of power per unit area of streambed, YQS (Ü=-Lyy- = yQSV = T(Ö (2.6) where J is the specific weight of sediment-water fluid, Q is stream discharge, 5 is the energy slope, w is streambed width, d is streamflow depth, V is mean flow velocity, and ICO is the shear stress exerted on the streambed (Baker and Costa, 1987). Unit stream power is a measure of the power available to do work in a reach of stream. The denominator of equation 2.4, resisting power, becomes greater with increases in hydraulic roughness, and the amount and size of bedload. When resisting power exceeds stream power, aggradation of bedload occurs as the suspended load continues to be washed downstream. The deposits of most stream terraces and many small alluvial fans consist largely of gravelly bedload materials, as compared to the silty suspended-load overbank deposits of large rivers. Both bedrock uplift and climatic change profoundly affect the components of the threshold of critical power (Starkel, 2003). Bedrock uplift increases watershed sediment yields by increasing landscape of Rising Mountains 41 relief and steepening valley floors. Climatic changes during the Pleistocene and Holocene changed discharges of both water and sediment. Climate change may increase bedload production so much that aggradation prevails along nearly all streams of a region, including reaches with tectonically active faults and folds. Watersheds in each climatic setting have different styles and magnitudes of geomorphic responses to changes from glacial to interglacial climates (Bull, 1991). Aggradation events in New Zealand occurred during full-glacial times. In the Mojave Desert of California they occurred during the transition to interglacial climates. Use of the threshold of critical power model emphasizes rates of change of the variables affecting a geomorphic process, thus encouraging you to consider the relative importance of many interrelated factors. 2.3.1 Relative Strengths of Stream Power and Resisting Power Relative strengths of stream power and resisting power vary spatially in drainage basins (Graf, 1982, 1983). Consider the hypothetical longitudinal profile for the tectonically active arid watershed depicted in Figure 2.14. Stream power continually exceeds resisting power in the headwater reaches. Slope remains excessively steep, especially where resistant rock types and small streamflows near drainage divides favor slow degradation of valley floors. Bedrock uplift of the mountain block occurs as displacements along the range-bounding normal fault zone. This tectonic perturbation to the entire fluvial system occurs in a narrow linear zone crossing the canyon mouth. Episodic bedrock uplift increases the slope component of stream power. Increases of orographically-induced precipitation also occur in rising mountains over time spans of 100 ky to 10 my, so local climate and daily weather patterns commonly are linked to long-term tectonic controls. In marked contrast, late Quaternary climatic changes over time spans of 0.1 to 3 ky were relatively abrupt. They greatly influenced both stream power and resisting power and simultaneously affected all the watersheds of a mountain range. Numerical modeling suggests vastly longer watershed response times to surface ruptures (Allen and Densmore, 2000). The situation depicted in the aggradational reach of Figure 2.14 is opposite to that of the headwaters. Deposition has been continuous because resist- 42 Chap Mountains Piedmont Figure 2.14 Relative strengths of stream power (5F) and resisting power (RP) along a hypothetical fluvial system in an arid closed basin. Tectonic perturbations are initiated by ruptures on a normal fault in reach T and climatic perturbations are initiated in reach C (all the drainage-basin hill-slopes). TIF is threshold-intersection point. ing power has persistently exceeded stream power. Such reaches are particularly obvious in basins of internal drainage where accumulation of playa and associated alluvial-fan deposits constitute a base-level rise that gradually tends to decrease overall piedmont slope and thereby stream power. Tectonic subsidence of a basin - stretch tectonics - tends to offset the base-level rise caused by aggradation. Examples of depositional basins include Great Salt Lake in Utah, the San Joaquin and Death Valleys in California, and lake basins in North Africa and the Middle East. The mountain-front reach of Figure 2.14 will be the most likely to change its mode of operation as a consequence of climatically induced changes in hill-slope water and sediment yield. This is the most sensitive part of a fluvial system to climatic and tectonic perturbations because it is where stream power may exceed, equal, or be less than resisting power for a given streamflow event. Relative discharges of water and bedload are especially important because moderate changes in resisting power may initiate episodes of aggradation or degradation. 2.3.2 Threshold-Intersection Points Threshold-intersection points are spatial crossings of the threshold of critical power along longitudi- :er 2 nal profiles of valleys (Fig 2.3). Aggradation in a mountain-front reach that formerly was downcut-ting into bedrock moves the threshold-intersection point upstream. Renewed stream-channel downcut-ting through the new valley fill and fan deposits then shifts the threshold-intersection point downstream leaving the recently elevated valley floor as the tread of a fill terrace (Figure 2.14). Threshold-intersection points may shift in a great variety of time scales. More than 2 ky may be needed for a shift such as that depicted in Figure 2.14. In contrast, sediment discharge commonly peaks before water discharge during a flood event. This causes the threshold-intersection point to migrate upstream and then downstream as the ratio of stream power to resisting power changes during the flow event. Threshold-intersection points shift more rapidly along the valleys of streams in humid regions than in arid regions, because wet climates provide greater annual stream power to do the work of transporting the sediment load imposed by weathering and erosion of hillslopes. 2.4 Equilibrium in Streams Those who seek tectonic information from landscapes need to know how tectonic perturbations change the behavior of fluvial systems. How do landscapes evolve when perturbed by vertical and horizontal earth deformation? Do they tend towards unchanging configurations? Do some landforms achieve such a steady state (equilibrium) sooner than others? Vital skills include being able to assess how far removed a landform is from a steady-state condition, and how quickly part of a fluvial system can achieve a new equilibrium. Depositional landforms, such as growing deltas and alluvial fans, do not even tend toward steady-state conditions (Bull, 1977a, 1991, Section 1.6.2). 2.4.1 Classification of Stream Terraces Some former valley floors are equilibrium reference surfaces that record tectonic deformation. Let us classify types of stream terraces, noting their suitability for tectonic studies. Two types of equilibrium, and the crossing of the threshold of critical power, can be recognized by the presence of distinctive types of stream terraces. We are particularly interested in stream terraces resulting from tectonic and climatic perturbations, but acknowledge that common minor Concepts for Studies terraces (called autocyclic) form because of responses to fluctuations of geomorphic processes within a fluvial system. Alternating aggradation and degradation events are common in fluvial systems that are sensitive to climatic changes. Termination of aggradation and degradation events creates key landforms. A degradation event is an interval of net lowering of a valley floor by fluvial erosion. Minor stream terraces form during pauses in the downcutting process. A degradation event commonly is terminated by a prolonged episode of lateral beveling when the stream stays at the same low position in the landscape. Different climatic controls promote an aggradation event, which tends to backfill valleys (Barnard et al., 2006) and commonly is terminated by deposition of a broad alluvial surface. Aggradation tends to bury fluvial landforms created by the preceding degradation event. Such burial is continuous as shown by lack of buried soil profiles in the stratigraphic section of an aggradation event. Fortunately, net stream-channel downcutting occurs in tectonically active landscapes. Watersheds with infrequent aggradation events are less likely to conceal landforms of the pre- A Broad surface that marks the end of an aggradation event is : fundamental climate-change reference surface of streams^ ^ Internal-adjustment / i of Rising Mountains 43 Broad strath beveled at the base level of erosion is fundamental tectonic reference surface of streams Climatic aggradation surfaces nternal-adjustment terraces Internal-adjustment fill terrace zlT. Tectonic straths cut in bedrock Internal adjustment minor strath terrace Charwell River at the ase level of erosion ceding degradation event when tectonically induced downcutting lowers the active channel below the former valley floors represented by stream terraces. First let us get acquainted with standard stream-terrace terminology (Fig. 2.15A). Terraces may be paired or unpaired. Unpaired-terraces (Davis, 1902) typically occur on the insides of meander bends along a stream that is steadily down-cutting, but sometimes occur where a stream temporarily erodes laterally into a bedrock hill. A paired terrace consists of remnants of a former floodplain that when connected describe a former longitudinal profile of the stream. They occur on both sides of a valley where not removed by erosion. An important characteristic of a paired terrace is that it has continuity along a valley. Flights of stream terraces can be likened to flights of stairs; both consist of a sequence of alternat- Figure 2.15 Tectonic, climatic, and internal-adjustment terraces of the Charwell River, New Zealand. A. Sketch of basic definitions. A strath beveled at the lowest possible position in a landscape is the tectonic reference \andform. A climate-change induced aggradation event buries the tectonic strath. The tread of the resulting fill terrace records a time of change from aggradation to renewed degradation. Pauses in the subsequent stream-channel downcutting that generate stream terraces may be due mainly to adjustments caused by variable hillslope water and sediment yield. B. Multiple strath and aggradation surfaces. Geomorphic responses to late Quaternary climatic changes and tectonic perturbations, and Internal adjustments of the fluvial system, are recorded as a flight of stream terraces in this rising landscape. The Flax Hills and Stone Jug aggradation surfaces are fill-terrace treads that record terminations of climate-change Induced aggradation events. Internal adjustments are preserved as minor strath, fill-cut, and fill terraces that record pauses in degradation from the aggradation surface to the younger broad tectonic strath terrace at the base level of erosion. 44 Chap ing treads and risers. A terrace tread is remnant of a former valley floor that has been abandoned by the stream as a result of stream-channel incision. The corresponding terrace riser is a scarp created by fluvial lateral erosion above the tread. Former levels of streams may be classed as fill, fill-cut and strath terraces (Leopold & Miller, 1954; Howard, 1959; Bull, 1990). A fill terrace is formed by aggradation of a valley floor and subsequent channel incision into the alluvium that leaves remnants of the former active channel as the tread of a paired fill terrace. The tread represents a time of crossing of the critical power threshold when the mode of operation switches from aggradation to degradation. Longitudinal profiles of fill terraces generally do not represent attainment of equilibrium, because this threshold crossing tends to be abrupt and may occur at different times along a valley. A strath terrace is genetically the same as a fill-cut terrace: straths are surfaces beveled in bedrock and fill-cut surfaces are beveled in alluvium. This generally brief equilibrium (pauses in a degradation event) is followed by renewed stream-channel downcutting that preserves remnants of the beveled surface, and gravel being transported at the time of formation, as the terrace tread. In this case, the riser (or streambank) was formed shortly before initiation of renewed stream-channel downcutting. Strath and fill-cut terraces differ from fill terraces in that only thin layers of stream gravel cap their erosion surfaces. These thin deposits may be regarded as lag deposits or gravelly cutting tools that most likely would have been entrained and redeposited by the next large flood. Erosional lowering of a valley floor does not proceed at a constant rate. Stream-channel downcutting rates vary with unit stream power and bedload transport rate. Pauses in valley-floor degradation allow fill-cut and strath terraces to form. Brief reversals in degradation of valleys caused by deposition of bedload result in minor fill terraces. Typically, both of these brief, minor, types of events create internal-adjustment terraces. In contrast to tectonic and climatic stream terraces, they result from temporal variations of dependent variables in a fluvial system. Such stream terraces have been studied in experimental (sandbox) models. They are called "complex responses" (Schumm, 1973; Parker, 1977; Schumm et al., 1987) and are referred to as "auto-cyclic" landforms by Hasbargen and Paola (2000). These authors studied internal-adjustment terraces ter 2 in sandbox models by initiating upstream migration of knickpoints that locally increased sediment yield from hillslopes. Although internal-adjustment terraces are important for understanding adjustments in fluvial systems, they do not record the start or end of climate-change induced aggradation, nor can they be used to assess rates of valley floor downcutting that are related to bedrock uplift. Many other processes cause perturbations of water and sediment discharge that favor formation of internal-adjustment terraces; fires and landslides are examples. A kinematic wave of bedload increase sweeps down trunk stream channels only to be followed by stream-channel incision to produce an internal-adjustment fill terrace. Aggradation may last only a few years or decades. Dating flights of internal-adjustment terraces in adjacent drainage basins usually reveals a lack of synchroneity (Bull, 1991). Each watershed behaves differently. In contrast, tectonic and climatic stream terraces tend to be synchronous throughout much of a mountain range. Examples of synchronous terraces range from a brief episode of valley backfilling caused by an earthquake-induced sediment-yield increase to regional aggradation of longer duration caused by stripping of the hillslope sediment reservoir during the Pleistocene—Holocene climatic change. Important terraces form at the end of major aggradation and degradation events. Thick alluvial fill deposited in response to regional climatic change raises valley floors and alluvial-fan surfaces. The amount and rate that valley floors are raised by such deposition are mainly a function of the time span of the climatic perturbation and magnitude of departure from the threshold of critical power. A fill-terrace tread represents the maximum altitude attained by backfilling a valley floor, and records the end of persistent aggradation that commonly lasts 1 to 15 ky. This type of aggradation surface is the fundamental climatic stream-terrace landform. The tread is the only landform created that has temporal significance if the riser above the tread is merely a valley side. Aggradation event surfaces in adjacent watersheds are synchronous where response times to climatic perturbations are similar. There are limits to the depth of erosion of any valley floor, and the termination of a degradation event sets the stage for creation of an important type of strath terrace. Lateral erosion that bevels bedrock and widens an active channel at the end of a degradation event Concepts for Studies may persist for 1 to 10 ky. It creates a strath that is the fundamental tectonic stream-terrace landform (Fig. 2.4). This alluvium-bedrock interface records the lowest possible longitudinal profile for a particular tectonic setting. Strath age is defined as the end of an interval of strath formation. This occurs when renewed, persistent deposition raises the streambed and buries the beveled bedrock surface. Termination usually requires a climatic change. Alternatively, valley floor bedrock beveling can be terminated by base-level fall induced by a large surface rupture in a downstream reach. This increases stream power sufficiently that the threshold of critical power is crossed to create a tectonic strath terrace landform. Multiple late Quaternary tectonic strath terraces record important times of attainment of equilibrium along the streams of rising mountains. In contrast to the generally minor internal-adjustment strath terraces, tectonic straths tend to be broader, and are more likely to be correlated between adjacent reaches. Reaches in adjacent drainage basins with similar bedrock resistance, discharge, and bedload transport rates generally complete a degradation event at about the same time. If so, time spans of tectonic strath formation may be synchronous for a mountain range. Adjustments of reaches of streams that achieve such equilibrium conditions are discussed in the next two sections. 2.4.2 Feedback Mechanisms The threshold approach to geomorphic investigations emphasizes how far removed a system is from stable conditions. Thresholds are essential for studying interactions between geomorphic variables that do not tend towards an unchanging "steady-state" condition. So, we need to consider such relations in a bit more detail. Interactions between geomorphic variables may tend to create a landform, or landscape, that does not change with the passage of time. Alternatively, such interactions may result in an opposite tendency away from equilibrium conditions. Key to understanding both modes of landscape evolution are feedback mechanisms between dependent variables that drive the system toward or away from steady-state conditions. Studies of both self-enhancing and self-arresting feedback mechanisms are encouraged by the threshold approach to geomorphology, whereas studies only of self-arresting feedback mechanisms are encouraged by the equi- of Rising Mountains 45 librium approach to geomorphology. The behavior of streams is one of the better examples of how self-arresting feedback mechanisms promote equilibrium. Landslides represent the opposite situation where self-enhancing feedback mechanisms drive the process towards its culmination rather than toward a balance between variables. Progressive accumulation or erosion of hill-slope colluvium are examples of self-enhancing mechanisms. Colluvium deposited on bare rock increases infiltration capacity, thereby providing both water and soil to support vegetation. The vegetation traps additional colluvial materials from upslope sources, thereby furthering the tendency for accumulation of more colluvium. For a reversal in the mode of operation to occur, a threshold must be passed that separates tendencies for progressive accumulation from those of progressive erosion of colluvium. Climate-change perturbations cause this system to alternate between two modes of operation and both landforms and landscape evolution cannot tend towards unchanging conditions. Such nonequilibrium modes of operation may involve time spans of decades or millions years. Progressive increase in areas of massive granitic outcrops in the Sierra Nevada of California over several million years may be considered an irreversible change. The self-enhancing feedback mechanism of rapid runoff from bare rock continues to erode the soil produced by weathering of granitic rocks igruss) at the margins of hillslope outcrops. Clyde Wahrhaftig (1965) made a classic study of such a landscape evolution. He sought to learn more about the formation of stepped topography developed in the massive granitic rocks of the western slopes of the Sierra Nevada. Bedrock exposed by streams expanded in area to become the dominant hillslope landform. Such topographic inversion — valley floors eventually become ridgecrests - requires that the drainage net, as well as the hillslopes, undergo progressive change. Thus Wahrhaftig's study provides a nice example of the importance of self-enhancing feedback mechanisms. The steady-state model is inappropriate because a key independent variable - erodibility of surficial materials - changes in both time and space. A different example of self-enhancing feedback mechanisms is the progressive development and ultimate collapse of a slump rotation block in massive sandstone cliffs (Schumm and Chorley, 1964). A seemingly insignificant tension crack at the top of the cliff signals initiation of a process that may require 46 Chap thousands of years. Insufficient lateral support for the cliff face allows the minute crack to gradually widen at an exponentially faster rate, thereby enhancing several processes that culminate in failure of the sandstone monolith when it collapses into a pile of rock-fall blocks. Water and ice accumulate in the ever widening crack, rotation shifts the center of gravity of the monolith, and block rotation fractures the lower portion of the sandstone pillar thereby causing a progressive decrease in rock mass strength. Now, let us examine the opposite type of feedback mechanism. The adjustment of a stream to an increase of gravel eroded from the hillslopes of its fluvial system is an example of self-regulation by self-arresting feedback mechanisms. Large increases of bedload change all hydraulic variables in meandering stream channels, which in turn results in a decrease in channel sinuosity, so that more bedload can be conveyed. In braided channels, increases in bedload may cause aggradation of the valley floor with maximum alluviation in the upstream reaches. This increase in streamflow gradient is the principal way stream power is increased. In both meandering and braided streams, adjustments of the hydraulic variables will continue as long as the stream is supplied an excess of bedload (Paola and Mohrig, 1996). Changes in flow characteristics or stream-channel characteristics that are greater than needed to transport the bedload will result in counterbalancing adjustments that will decrease the transporting capacity and competence of the stream. A key tectonic landform is the longitudinal profile of a stream channel (Merritts et al., 1994). We consider present and past longitudinal profiles as potential reference datums passing through tec-tonically deforming landscapes. So, it is worth our while to consider several types of equilibrium in such streams, and the landscape characteristics associated with each type. 2.43 Dynamic and Static Equilibrium The base level of erosion concept describes reaches of streams that have achieved one of two types of equilibrium-static equilibrium and dynamic equilibrium (which can be further classified as type 1 or type 2). We define and discuss these terms here in order to clarify how climatic controls and internal adjustments in fluvial systems lead to static equilibrium, and how tectonic controls create situations of dynamic equilibrium in landscape evolution. Unless ter 2 classed as static or dynamic, the general term "equilibrium" refers to reaches that have attained the base level of erosion. Static equilibrium is attained briefly when bedload derived from the watershed hillslopes is transported through a reach with neither aggradation nor degradation of the streambed (Leopold and Bull, 1979). One example is a gradual, instead of abrupt, switch from aggradation to degradation at the conclusion of an aggradation event. Bull (1991, Section 5.4.4.1) describes field identification of an aggradation surface that records a period of static equilibrium. A stream that pauses briefly during a degradation event creates minor fill-cut or strath terraces that also record temporary static equilibrium. John Hack used the ideas of G.K. Gilbert when he applied the concept of dynamic equilibrium (steady state) to open fluvial systems of mountains. Hack believed that a steady-state landscape configuration would develop between processes that tend to raise and denude mountains. All elements of a landscape in dynamic equilibrium are mutually adjusted to each other. The resulting landscape assemblage downwastes at a uniform rate, and its configuration does not change (Gilbert, 1877; Hack, I960, 1965; Carson, 1971). Thus, dynamic equilibrium is independent of time and acknowledges that climatic and tectonic energy is continually entering and leaving the system. Adjustments to perturbations restore a dynamic steady state without significant delays. The steady-state model of landscape evolution continues to be highly regarded by many geomorphologists. The Gilbert-Hack model is heuristic, because we have yet to obtain field evidence for attainment of true steady state for an entire landscape (Bull, 1977a). Important tectonic and climatic perturbations occur frequently during time spans of 0.1 to 100 ky and affect denudation rates that vary with both time and space within drainage basins. In my opinion, response times of geomorphic processes are sufficiently long for hillslopes to prevent attainment of steady state for either watersheds or mountain ranges. Models using non-steady state assumptions are preferable for studies of landscape evolution. Solyom and Tucker (2004) nicely describe obvious examples of why steady state is inappropriate for modeling short-term behavior of entire watersheds by emphasizing interactions between the variables of storm duration, basin shape, and peak discharge. Central to their argument is that precipitation-runoff inputs are not uniformly distributed in Concepts for Studies a spatial sense, creating disparities that increase with drainage-basin area and with increasing watershed aridity. Nonuniform inputs of precipitation typically create situations of partial-area contributions of runoff from a watershed (Dunne and Black, 1970; Yair et al., 1978), which lead to lesser concavity (rate at which a longitudinal profile becomes flatter downstream). Lack of a vigorous proof for dynamic equilibrium for entire watersheds does not detract from the value of the base level of erosion concept, and the tendency of specific landscape assemblages toward time-independent shapes. One nice application of the dynamic equilibrium model is the way in which powerful streams respond to bedrock uplift. Two categories of dynamic equilibrium in streams may be defined in terms of length of attainment of the base level of erosion. Hack (1973, 1982) defined steady-state adjustment between variables as those longitudinal profiles that plot as straight lines on semi-logarithmic graphs because stream discharge increases logarithmically downstream (Wolman and Gerson, 1978). Two types of dynamic equilibrium can be recognized in the field. Diagnostic landforms for attainment of type 1 dynamic equilibrium include a strath beneath the active channel and a valley floor that is sufficiently wide for preservation of strath terraces. Type 1 dynamic equilibrium is present when stream down-cutting and bedrock uplift rates are the same. An analogy would be the constant shape and position of a rotating circular saw (the stream) as a log (the mountains) is raised into it. The resulting sequence of longitudinal profiles represents an infinite number of tectonic base levels of erosion as the stream reestablishes equilibrium conditions after each tectonic perturbation in the rising landscape. Climatic perturbations in real world fluvial systems do not allow perpetual uniform downcutting in response to bedrock uplift. Instead, climatic factors modulate the discharge of water and bedload so as to vary the rate of valley-floor downcutting. The result is a series of closely spaced minor strath terraces, each created briefly at an appropriate time as dictated by changing climatic constraints. Type 2 dynamic equilibrium is present in streams with a strong tendency toward, but lack of sustained attainment of, the base level of erosion. Diagnostic landforms include narrow valley floors whose longitudinal profiles (like those of type 1 stream reaches) plot as concave lines on arithme- of Rising Mountains 47 tic graphs and as straight lines on semi-logarithmic graphs. Straths and remnants of strath terraces generally are not present. Recent studies of bedrock stream-channel longitudinal profiles and stream-channel characteristics are a worthwhile departure from the previous emphasis on less confined, easily eroded alluvial streams (Baker and Kochel, 1988; Seidl and Dietrich, 1992: Wohl, 1994; Montgomery et al., 1996; Baker and Kale, 1998; Hancock et al., 1998; Pazzaglia et al., 1998; Sklar and Dietrich 1998; Tinkler and Wohl, 1998a, b; Massong and Montgomery, 2000; Wohl and Merritt, 2001, Whipple, 2004). Irregular longitudinal profiles are characteristic of disequilibrium streams upstream from most type 2 equilibrium reaches where interactions between variables have yet to approximate the base level of erosion. Landforms in disequilibrium reaches include convex valley sideslopes plunging into V-shaped canyons with waterfalls and rapids, and convex longitudinal profiles, even when plotted on logarithmic graphs. Stream-channel widths for degrading streams tend to be narrower than for streams in equilibrium. Consider the case where renewed degradation converts an equilibrium reach characterized by strath cutting (active-channel width is not confined) to confined streamflow that is typical of downcutting reaches. Streams degrading into bedrock may establish a self-enhancing feedback mechanism that accelerates downcutting until equilibrium conditions are approached. Narrowing of streamflow tends to increase unit stream power (equation 2.6). Millions of years may be needed for ephemeral streams flowing over resistant rocks to approach their base levels of erosion. Some response time is needed for large rivers to adjust to a pulse of uplift even in humid mountain ranges. The time spans needed for landscape adjustments typically are longer than intervals between uplift. Upstream reaches and adjacent hillslopes will have longer response times. So a model where interacting landscape elements continue to adjust makes more sense than proclaiming still another example of steady-state conditions. One alternative is the allometric change conceptual model (Bull, 1975a; 1991, Section 1.9) that describes orderly interactions between dependent variables in a changing landscape before or after the base level of erosion is attained. Allometry in biology describes relative systematic changes in different parts of growing organ- 48 Chap isms (Vacher, 1999). Allometric change in geomor-phology describes orderly behavior in nonsteady state fluvial systems. Hydraulic geometry of stream channels (Leopold and Maddock, 1953) is a good example. Increase of flow width with increasing discharge at a gauging station is static (at a single location in a watershed) allometric change. Increase of flow width with downstream increase of discharge (a sequence of locations) is dynamic allometric change. The allometric change model emphasizes the degree of inter-connectivity of different geomorphic processes and landscape characteristics. Allometric change emphasizes system behavior instead of attainment of steady state and is mentioned here to alert readers that most regressions between geomorphic variables in this book imply nonsteady-state assumptions. Perennial streams of humid regions flowing over soft or highly fractured rock are better suited for studies of responses of streams to bedrock uplift during the past 40 ky. One such stream is the Charwell River, whose main branch drains 30 km2 of the Seaward Kaikoura Range of New Zealand. Times of aggradation events were estimated by radiocarbon dating of wood in the fill-terrace deposits of the region. The Charwell River is the type area for the Stone Jug, Flax Hills, and Dillondale climate-change induced aggradation events, which have synchronous counterparts elsewhere in New Zealand (Barrell et al., 2005; Litchfield and Berryman, 2005). Terrace tread ages were estimated with weathering rind analyses (Knuepfer, 1988). Strath ages were estimated by radiocarbon dating of tree trunks, in growth position, just above the straths. These age estimates indicate termination of intervals of major strath beveling at about 40 and 29 ka. Tectonic strath terraces are those that record times of attainment of the base level of erosion (type 1 dynamic equilibrium) in rivers that are being tectonically elevated. The Charwell River has both internal-adjustment and tectonic strath terraces. The piedmont reach extends 12 km downstream from the highly active range-bounding Hope fault, has local faulting and folding, and has experienced episodes of climate-change induced aggradation. Bedrock uplift ranges from 0.2 to 1.3 m/ky. Successive tectonic strath terraces are at lower positions in the landscape because the Charwell River lowers its channel in response to continued bedrock uplift (Fig. 2.15B). Each broad tectonic strath was created at a time of interstadial or interglacial climatic conditions, such as the present Holocene climate. Each episode of tee- ter 2 tonic strath formation ended at the time of initiation of an aggradation event. The long-term tendency of rivers to cut down into a rising landmass is temporarily reversed during aggradation events, which in humid, mesic to frigid climates coincide with times of full-glacial climates (Bull, 1991, Chapter 5; Wegmann and Pazzaglia, 2002). Rivers then switch to degradation and may catch up to, and re-establish, type 1 dynamic equilibrium during times of interglacial climates. This is when streams have excess stream power, relative to resisting power. Rapid stream-channel downcutting (Bull, 1991, Fig. 5.18) began at 14 to 16 ka after termination of the latest Pleistocene aggradation event, and ended at about 4 ka. By then, the Charwell River had attained its tectonic base level of erosion downstream from the range-bounding Hope fault. Bedrock uplift has continued to raise the stream during the past 4 ky, and the active channel has cut down into soft piedmont-reach lithologies at a rate that matches the local bedrock-uplift rate. Fluvial erosion switched from largely vertical to mainly horizontal and the stream beveled a strath as wide as 200 m (discussed and shown later in Figure 2.26). Degradation stream terraces also record times of attainment of the base level of erosion. The flight of Charwell River stream terraces provides examples of how self-arresting feedback mechanisms result in temporary static equilibrium. Latest Pleistocene aggradation buried the 29 ka tectonic strath and raised the streambed altitude 40 m. Then rapid degradation of the bouldery valley fill occurred. The stream paused a dozen times as it cut down before it caught up to its long-term (tectonic) base level of erosion (Fig. 2.16). Tectonic and climatic controls on other Marlborough streams follow a similar pattern but magnitudes of change in positions of valley floors are functions of the amounts of latest Quaternary aggradation and uplift. Mean rates of post-glacial incision for the Saxton River on the Awatere fault are 1.4 m/ky compared to 5.1 m/ky for the Charwell River (Fig. 2.16). The shapes of the two incision curves are the same, and both streams have returned recently to the base level of erosion where the rate of stream-channel downcutting matches the uplift rate. Each pause was a time of attainment of static equilibrium that was recorded by a minor terrace. Stream-channel entrenchment into poorly sorted sandy gravel promoted selective entrainment and transport of the finer portion of the valley fill. Accumulation of residual boulders on the streambed Concepts for Studies Types of stream terraces Tectonic I Internal adjustment |Climatic| 0 4 8 12 16 Age, ka Figure 2.16 Tectonic, climatic, and internal-adjustment terraces of the Charwell River: variations of stream-channel downcutting rate since 16 ka. Uncertainty cross estimates are for terrace ages and distances below the aggradation surface reference level. From Figure 6 of Bull and Knuepfer (1957). Gray plot is for Saxton River on the Awatere fault. From Figure 13 of Mason et al. (2006). created a lag deposit that armored and protected the streambed from further degradation by 1) increasing the shear stresses needed to initiate movement of streambed gravel, and 2) by increasing the hydraulic roughness. Riparian vegetation grows beside streams with armored beds, further increasing hydraulic roughness. Neither net aggradation nor net degradation occurs, even though the active channel is still far above its tectonic base level of erosion. These brief episodes of static equilibrium end abruptly when a 1 -ky flood event entrains and smashes streambed-armor boulders, and destroys riparian vegetation. Both changes reduce hydraulic roughness and the shear stresses needed to entrain bedload in the normal range of stream discharges. Renewed streambed degradation and winnowing then create a new lag gravel at a lower altitude. Remnants of the former streambed, with the characteristic stream-bed armor preserved as a capping layer of boulders, remain as treads of fill-cut or minor strath terraces. Thus without invoking either climate-change or tectonic perturbations, a self-arresting feedback of Rising Mountains 49 mechanism can occur repeatedly during the formation of a flight of degradation terraces. These are regarded as internal-adjustment terraces because they are not the result of changes in the independent variables of this fluvial system. Each records a pause in valley-floor degradation caused by temporary increases of resisting power. Different geomorphic responses to bedrock uplift in the reach upstream from the Hope fault have resulted in narrow V-shaped canyons (Bull, 1991, Fig. 5.7), convex footslopes, and waterfalls. The fractured greywacke sandstone is more resistant to erosion than the soft Cenozoic sediments downstream from the fault. Bedrock-uplift rate is about 3.8 ± 0.2 m/ky based on altitudinal spacing analysis of uplifted marine-terrace remnants. Degradation by small headwater streams is unable to keep pace with bedrock uplift, but some mid-basin reaches of the Charwell River have sufficient unit stream power to approximate type 2 dynamic equilibrium. The reach just upstream from the Hope fault has remnants of two fill terraces, perhaps reflecting repeated episodes of base-level rise caused by rapid deposition of 30-50 m of latest Pleistocene valley fill in the adjacent reach downstream from the Hope fault. Many possible interactions between variables can produce equilibrium in streams. Increased slope and reduced stream-channel width may be important in achieving equilibrium after vertical offset of a streambed by an earthquake surface rupture. Changes in hydraulic roughness and stream-channel pattern may occur after a landslide increases bed-load. Powerful streams may maintain equilibrium by adjusting interactions between variables, within a limited range, as short-term climatic change influences unit stream power and bedload transport rate. Perturbations that force a reach of a stream beyond its capacity to maintain equilibrium conditions initiate aggradation or degradation events. Sections 2.6 and 2.7 provide several examples of how strath terraces can be used by tectonic geo-morphologists. But first, let us outline the characteristics of response times of fluvial systems to climatic and tectonic perturbations. 2.5 Time Lags of Response Several concepts, aggradation and degradation events, attainment of the base level of erosion, and time lags of response can be summarized with a simple diagram (Fig. 2.17) Reaction time is the delay before Chapter 2 Figure 2.17 Anatomy of an aggradation-degradation event illustrating the concepts of reaction time, relaxation time, and response time after a climate-change perturbation causes an episode of alluviation of a valley floor. Stream at base level of erosion streambed aggradation begins: for example, the interval when hillslope plant cover decreases thus increasing sediment yield sufficiently that the stream can no longer maintain equilibrium conditions. Relaxation time is the time span needed to complete the aggradation event. Response time is shown as the total elapsed time from the time of a climatic perturbation to end of the aggradation event. It is the sum of reaction and relaxation times (Allen, 1974; Thornes and Brunsden, 1977; Brunsden and Thornes, 1979; Brunsden, 1980). The ensuing degradation event returns the stream to the same base level of erosion in this case. This is indicative of a lack of uplift during the time span represented by the aggradation-degradation event. Fluvial systems also respond to being elevated. Persistence time is the time span during which fluvial system behavior is constant, which here is a condition of type 1 dynamic equilibrium. 2.5.1 Responses to Pulses of Uplift The concept of response time in a fluvial system to tectonic inputs can be illustrated with a threshold-equilibrium plot (Fig. 2.18). Reactions of a stream to two hypothetical perturbations are illustrated for a reach that is 2 km upstream from the mouth of a watershed and initially in type 1 dynamic equilibrium. The first perturbation is a 2 m surface rupture on the range-bounding normal fault. Tectonic lowering of the reach immediately downstream from the fault, relative to the mountain block, creates a knickpoint. This short tectonically steepened reach Reactior time -R. Relaxation time -R, Time Figure 2.15 Hypothetical thresh old-equilibrium plot showing the components of response time, R, for a reach 2 km upstream from a normal fault. Delayed responses are depicted for two perturbations; a 2 m surface rupture on the range-bounding fault, and a landslide upstream from the study reach. Response time is the sum of the reaction time, K , and relaxation time, R . F is the time of X s persistence of new equilibrium condition, and T and E are the times of threshold and equilibrium conditions respectively Concepts for Studies migrates upstream. A knickpoint may be a waterfall initially but commonly becomes rapids - a knick-zone — where rock mass strength is low. Processes of upstream migration are different for these two types of stream-channel perturbation. Knickpoints retreat mainly by undermining and collapse and knickzones by streambed abrasion and plucking by fast moving saltating bedload. Both may be significant local departures from the typically concave longitudinal profile indicative of a reach in equilibrium. As such they can inhibit continuity of fluvial systems (Section 2.5.2) to the extent that waterfalls separate relict landscapes in a dramatic fashion (Crosby and Whipple, 2006). Rates of incision into bedrock are considered to be proportional to shear stress exerted by streamflow (Howard and Kerby, 1983; Tucker and Slingerland, 1994; Snyder et al., 2000, 2003). Some geomorphologists prefer to emphasize total stream power (Seidl and Dietrich 1992), bedload transport rates relative to the threshold of critical power (Kooi and Beaumont, 1994), bedload abrasion potential (Sklar and Dietrich, 1998), or unit stream power (Whipple and Tucker, 1999). The interval is quite short for the first persistence time of Figure 2.17. Migration of the hypothetical knickzone (Figure 2.17) still further upstream undercuts a hillside and triggers a landslide - the second perturbation. The result is an increase in bedload input to the stream that causes minor aggradation in the study reach, followed by a new equilibrium condition. Note that equilibrium is regarded as an interval with no change of streambed altitude, whereas the threshold is depicted as a point in time when the system reversed modes of operation (for example, from degradation to aggradation). The first two equilibrium intervals are examples of type 1 and type 2 dynamic equilibrium. The third is an example of static equilibrium, because the stream is above its long-term (tectonic) base level of erosion. Reaction time is a measure of the sensitivity of a fluvial system to a perturbation, and relaxation time is a measure of how efficiently a geomorphic system adjusts to a perturbation. The reach immediately upstream from a surface rupture reacts quickly as waterfalls and rapids are created. Hillslopes in the headwaters of the same stream react slowly, and flat summits and plateaus have such exceedingly long reaction times that they are essentially isolated from downstream perturbations. Resistance of rocks may be the same throughout a drainage basin, but stream power decreases exponentially in the upstream direc- of Rising Mountains 51 tion. Spatial variations in response time inhibit attainment of steady-state conditions for entire drainage basins. Outputs of fluvial systems, such as sediment yield, integrate response times from all parts of a drainage basin. Renewed degradation of a valley floor in response to a range-front uplift event provides a new source of sediment as streamflow downcuts into reaches previously at equilibrium. Stream-channel entrenchment steepens adjacent hillsides, which promotes mass movements; both processes increase sediment yield. Tectonically induced increases of watershed sediment yield increase resisting power and tend to accelerate deposition in reaches downstream from the active fault. Chapter 4 has many examples of the abrupt initiation of deposition of alluvial fans that coincide nicely with this tectonic base-level fall. An aggradation-rate increase on the alluvial fan reach of a fluvial system alluvial fan occurs as quickly as upstream channel downcutting is initiated, but maximum rate increase may be delayed substantially until tributary streams and their large areas of hillslopes are affected by an episode of range-front faulting. So reaction time is short for the depositional reach, but relaxation time may be extended until the effects of a tectonic perturbation have spread throughout much of the drainage basin. The response-time model described above applies only to those fluvial systems with continuity. Both erosional and depositional reaches of some streams can have characteristics that inhibit continuity of fluvial-system behavior. 2.5.2 Perturbations that Limit Continuity of Fluvial Systems Waterfalls are dramatic landforms that typically isolate upstream reaches from the effects of surface rupture on downstream fault zones. These streambed cliffs effectively decouple upstream from downstream reaches. Decoupling isolates the reaches in the headwaters of a watershed, which may show minimal consequences of incremental increases in watershed relief emanating from an active range-bounding fault. Decoupling inhibits stream-channel downcutting in a trunk valley from migrating up tributary valleys. Instead, such upstream reaches are graded to the top of a waterfall, or to a rapid with a drop sufficient to dissipate energy in a hydraulic jump. Rapids and falls are local base levels (Fig. 2.2B). 52 Chap Geomorphic responses to climatic changes may control the magnitude, time of formation, and position in a watershed of a waterfall (Sections 2.6.2, 2.7). Episodes of uplift along a range-bounding fault increase the height of a bedrock fault scarp even when deep beneath the gravels of a climate-change induced aggradation event. Age controls for the Charwell River landscape evolution allow estimation of process rates (Bull and Knuepfer, 1987). The period between about 26 and 14 ka ago was a time of valley floor backfilling. Episodic vertical movements along the Hope Fault during this time span continued to displace both bedrock and valley fill. Streamflows would quickly eradicate scarps formed in unconsolidated fill, but a sub-alluvial bedrock fault scarp would have become progressively larger until Holocene degradation was sufficient to expose bedrock once again at about 9 ka. The Hope Fault separates the mountain and piedmont reaches; both are rising and relative uplift is about 2.5 m/ky. Thus, late Quaternary cli- Figure 2.19 Decoupled hillslopes, piedmonts, and stream channels. A. Slowly eroding upland in the Kelly Range, Southern Alps, New Zealand changes abruptly to cliffy headwater reaches of rivers that are downcutting rapidly in response to bedrock uplift of 5 m/ky (Bull and Cooper, \9&&). Uplands are shore platform(s) of uplifted marine terraces (SP). Degraded former sea cliff at SC. The foreground cliffs decouple the upland terrain from the ad\acent downstream reaches of these fluvial systems. Hut at H is 110 m below shore platform at SP. ter 2 matic perturbation modulated tectonic processes to create and then exhume a prominent bedrock waterfall about 40 m high. After exposure it would retreat upstream as a knickpoint or steeper reach. The present anomalously steep reach 1300 m upstream from the Hope Fault departs from a smooth profile by about 40 m. Cliffs prevent continuity of geomorphic processes on hillslopes too. One example is the free face on a young fault scarp. This is why we can't do diffusion-equation modeling for a fault scarp that still has a free face. Another common example of lack of fluvial system continuity is the precipitous drop below a tabletop upland or mesa. Flux of sediment and water through cliffy landscapes is interrupted in a system that otherwise would behave systematically enough to be described by a single set of equations. Summit uplands are subject to such different processes that they may be eroding at a miniscule rate relative to the valley floors and hillslopes downvalley from the cliffy terrain that decouples the headwaters from the rest of the fluvial system. Remnants of shore platform-sea cliff landscapes of raised marine terraces (Fig. 2.19A) are another example of decoupled terrain. Reaches of active deposition may decouple fluvial systems as efficiently as waterfalls. Consider the case of a downcutting stream that emerges from a rising mountain range, crosses an aggrading alluvial fan, and joins a trunk river. A base-level fall on the trunk river will migrate upstream as a headcut but cannot be transmitted through the alluvial-fan reach as long as it remains on the aggradational side of the threshold of critical power. Fans decouple fluvial systems by not allowing base-level falls further downstream to influence stream channels and hillslopes in their source areas. The Carrizo Plains adjacent to the Temblor Range of California have ephemeral streams that respond to tectonic and climatic perturbations (Ouchi, 2005). Sieh and Wallace (1987) note that the fanhead trench of Wallace Creek has been incised for about 3.7 ky and records 128 m of right-lateral displacement by the San Andreas fault zone during this time span (Fig. 2.19B). Alluvial-fan deposition has subsequently occurred on the fanhead of the unnamed creek, whose young surface has undergone minimal stream-channel offset. Both streams have undergone arroyo cutting (Bull, 1964a, b) during the past 200 years. Erosional knickpoints during this climate-change induced stream-channel entrenchment can easily move up the continuous channel of Concepts for Studies of Rising Mountains 53 Wallace Creek 0.5 Unnamed eneek 1.0 km 5 Axrt carUmr hnUrn»\ Wallace Creek. The 400-m long aggradational reach stops knickpoint migrations in the unnamed creek. The fluvial system of the unnamed creek will remain decoupled until all threshold intersection points are eliminated so all reaches are on the erosional side of the threshold of critical power. Even big rivers can be decoupled. The Grand Canyon reach of the powerful Colorado River illustrates a third style of fluvial decoupling. This reach lacks the smooth concave longitudinal profile indicative of attainment of equilibrium conditions that one would expect for a big river. Instead, the profile is convex with numerous abrupt steps (Fig. 2.19C). Steps coincide with tributaries from cliffy 900 &00 - zs < 700 600 - 500 50 100 150 200 250 Distance downstream from Lees Ferry (km) Figure 2.19 Decoupled hillslopes, piedmonts, and stream channels. B. Topographic map of Wallace Creek crossing the San Andreas fault that shows piedmont stream channels that are decoupled from their source areas by intervening reaches of alluvial-fan deposition. Wallace Creek is the only stream channel with continuity that permits upstream migration of knickpoints. Trace of the San Andreas fault is at base of the escarpment. From Figure 3 of Sieh and Wallace (19S7). watersheds whose frequent debris flows deliver 1 to >3 m boulders to small fans (Fig. 2.19D). The river quickly removes the sand fraction from such debris fans, and can shift some of the smaller boulders a short distance downstream during flood discharges (Webb et al., 1999, Figure 41). Repeat photography (Webb et al., 1999, Figure 14) shows the same large boulders in the Lava Falls rapids after a moderate flow of 6,200 m3/s (220,000 ftVs). Encroachment by debris fans controls the longitudinal profile of the river and the hydraulic behavior of the rapids (Dolan et al., 1978; Howard and Dolan, 1981; Kieffer, 1985; Webb et al., 1989; Melis and Webb, 1993; Webb, 1996; Griffiths et al., 1996, 2004). The Grand Canyon reach does not have the typical pool-riffle sequence characteristics of bedload streams (Webb et al., 1996, p. 152) because of the size and flux rate of boulders from tributaries. The result is constricted, steep reaches with near-critical streamflow characteristics and upstream migrating hydraulic jumps. Half the drop in altitude in the Grand Canyon reach occurs in short rapids (Leopold, 1969). Each bouldery obstruc- Figure 2.19 Decoupled hillslopes, piedmonts, and stream channels. C. Longitudinal profile and gradients of the Grand Canyon reach of the Colorado River, Arizona between 50 and 250 km downstream from Lees Ferry. Steep reaches coincide with rapids (only a few are named here) resulting from bouldery debris flows derived from cliffy tributary streams. 1927 data set supplied courtesy of Robert H. Webb, U. S. Geological Survey. 54 Figure 2.19 Decoupled hillslopes, piedmonts, and stream channels. D. Diagrammatic Grand Canyon debris fan and rapids. 1 is debris-flow fan from tributary watershed. 2 is constricted river flow plunging down a rapid with large immobile boulders. 3 is debris bar of cobbles and small boulders derived from debris fan. 4 is secondary rapid caused by debris bar. From Griffiths et al., 1996 as modified from Hamblin and Rigby, 1965. Chapter 2 As elsewhere in the American Southwest (Bull 1991, Chapter 2) the change to Holocene climates resulted in much more frequent debris flows. Summer monsoon-type rainstorms and winter cyclonic storms, such as cutoff lows off the coast of Southern California and Mexico, that are sufficiently warm to produce thunderstorms were scarce or absent during times of cooler full-glacial climates. So, the influence of debris flows on the Colorado River longitudinal profile may not have been important before the change to a Holocene monsoonal climate. Pazzaglia (2004, p. 268) concludes "There remains no good single explanation for why dramatically steep slopes on the Great Escarpments of passive margins erode at slow rates approaching 5 m/My whereas equally steep slopes in tectonically active settings may erode at rates three orders of magnitude faster approaching 5000 m/My." The concept of impediments to continuity in fluvial systems may explain this paradox. Disconnected landscape elements cannot transmit base-level falls in a manner resembling watersheds where the trunk stream channel is the connecting link between different parts of an integrated fluvial system. Numerical analyses that fail to recognize the presence of two separate, adjacent, landscape systems might use a process-response model that does not represent the real world. Conceptually, the ancient Great Escarpments of South Africa resemble the sandstone spires of Monument Valley, Arizona. Both are spectacular weathering-limited cliffs whose rates of denudation have little relation to base-level changes in the streambeds of nearby fluvial systems. tion acts as a local base level for the upstream pool reach in much the same manner as first described by John Wesley Powell (1875, p. 203-204). Continued influx of debris-flow boulders is sufficient to maintain rapids as significant local base levels. Lava that flowed into the canyon was a local bedrock base level that was removed by the river. Bouldery debris fans are a different type of perturbation. They are renewable base-level controls that define the character of the Grand Canyon reach. Human impacts are increasing the importance of this perturbation. Heights of some rapids are increasing since dam construction eliminated large annual floods (Graf, 1980; Melis and Webb, 1993; Webb et al, 1996). 2.5.3 Lithologic and Climatic Controls of Relaxation Times The independent variables of lithology and structure, and of climate, largely determine the types of processes that create hillslope landforms, and the time needed for streams to achieve equilibrium conditions. Joints and fractures and petrologic fabric beneath a watershed change little with time, whereas climate change is ubiquitous. Climate also varies between north- and south-facing slopes, and with altitude. Lithology also varies with space, and even monolithologic watersheds - those drainage basins underlain by a single rock type such as quartz mon-zonite or greywacke sandstone - typically have a highly variable density of joints, fractures, and shears. Concepts for Studies of Rising Mountains 55 Precipitation Temperature Mean Annual (mm) Mean Annual (°C) Extremely arid Arid Semiarid Semihumid Humid Extremely humid < 50 50-250 250 - 500 500- 1,000 1,000-2,000 > 2,000 Pergelic Frigid Mesic Thermic Hyperthermic > 0 0-8 8-15 15-22 >22 Seasonality Index (Sp)* Seasonality Index (St)51 Nonseasonal 1-1.6 Weakly seasonal 1.6 - 2.5 Moderately seasonal 2.5 - 10 Strongly seasonal > 10 Nonseasonal < 2 Weakly seasonal 2-5 Moderately seasonal 5-15 Strongly seasonal > 15 * Precipitation seasonality index (Sp) is the ratio of the average total precipitation for the three wettest consecutive months (Piv) divided by the average total precipitation for the three consecutive driest months (Pd). Sp = Pw/Pd * Temperature seasonality index (St) is the mean temperature of the hottest month (Th) minus the mean temperature of the coldest month (Tc) in °C. St = Th-Tc Table 2.1 Classification of climates. Relaxation times after fault ruptures of streambeds are short for large rivers flowing on soft rocks, and long for ephemeral streams flowing on hard rocks. Locally massive, hard rocks greatly increase the lifespan of a waterfall that decouples upstream from downstream reaches. Multiple surface-ruptures produce knickpoints that may migrate upstream only to increase height and permanence of waterfalls. Table 2.1 defines the climatic terms used in this book. Each category, including extremely humid and extremely arid, is characterized by major differences in geomorphic and pedogenic processes. The temperature terms are from Soil Taxonomy (Soil Survey Staff, 1975). Mean annual air temperature at a site approximates soil temperature at a soil depth of 50 cm. Climatic constraints affect the time needed for fluvial processes to shape a given landform by at least an order of magnitude (a ten-fold variation). Consider the triangular facets shown in Figure 2.20A. Weak to moderately resistant rocks and an arid, thermic, strongly seasonal climate are responsible for Saline Valley triangular facets with minimal dissection. The lower portion of the rilled facet approximates the plane of an exhumed 35° to 40° range-bounding normal fault such as those described by Cichanski (2000). One might conclude that bedrock uplift must be rapid to form such dramatic triangular facets. The bedrock-uplift rate probably is typical of other rapidly rising mountain fronts in the Basin and Range Province, most likely being 0.3 to 1.0 m/ky. The minimal degradation seen here is in large part the result of the arid, thermic to hyperthermic climatic setting. A much different climate influences the triangular facets of Figure 2.20B. The northwest front of the Southern Alps of New Zealand is being raised along the oblique-reverse range-bounding Alpine fault that dips under the range. Quartz-biotite schist offers little resistance to erosion after being weathered in the extremely humid, mesic, weakly seasonal climate. Deep valleys dissect the triangular facets. One might erroneously conclude that this landscape is indicative of a slow bedrock-uplift rate. Instead, this is one of the fastest rising major mountain fronts in the world - rock uplift and ridgecrest uplift is about 5 to 8 m/ky 56 Chapter 2 Figure 2.20 Lithologic and climatic control of tectonic landforms Illustrated by a comparison of triangular facets. A. Mountain front along the southwest side of arid Saline Valley in southeastern California. The mountain-piedmont junction coincides with a normal fault. The slightly rilled lower surface, just above the mountain-piedmont junction, has a homogeneous abearance because it is fault gouge. Contrasting litholo-gies are obvious higher on the slope where the thin layer of gouge has been removed by erosion. Local patches of colluvium and alluvium cling to the fault plane such as at the top of the waterfall at the left side of the view. (Bull and Cooper, 1986; Yetton and Nobes, 1998). Valley-floor surface uplift is < 1.0 m/ky because these big rivers with large annual stream power have impressive stream-channel downcutting rates. Increases of bedload size and amount may have contributed to modest long-term increases in stream-channel gradient in reaches upstream from the range-bounding Alpine fault. The main divide of the Southern Alps is high partly because sustained rapid erosion of deep valleys promotes isostatic compensation that further increases the altitudes of peaks and ridgecrests. The factors that influence surface uplift (Fig. 1.4) can be used to elaborate on the usefulness of tectonic landforms such as triangular facets. Surface uplift more closely approximates bedrock uplift in Saline Valley because erosion is mini- F3. Mountain front along the northwest side of extremely humid southern Alps of New Zealand. The dense rain forest provides little protection against rapid erosion of schist. The mountain-piedmont junction coincides with the oblique-reverse Alpine fault. mal. This favors preservation of triangular facets as a tectonic landform. However, rapid erosion of weathered rocks creates tectonic landforms suggestive of relatively less bedrock uplift in the Southern Alps. The magnitude of climate-controlled erosion is large enough to affect styles of crustal faulting (Koons, 1989; Norris and Cooper, 1995). Average surface uplift for Southern Alps watersheds surely is reduced by rapid erosion of the landscape, but still exceeds that of less tectonically active Saline Valley. The importance of climatic control on landscape evolution demonstrated by this comparison underscores the difficulty of using landforms for quantitative estimates of bedrock or surface uplift rates. Alternatively, one can rank qualitative classes of surface uplift based on assemblages of tectonic land-forms within a given climatic province (Chapter 4). The concept of relaxation time also applies to the consequences of Pleistocene-Holocene climatic change. Reaction times typically are brief when protective plant cover is changed and hillslope soils undergo net erosion instead of net accumulation. The pulse of valley-floor alluviation caused Concepts for Studies by stripping of the hillslope sediment reservoir has a relaxation time of only a brief 1 to 3 ky in hot deserts (Bull, 1991), but is much longer for vegetated hillslopes of humid regions. Density of hill-slope plant cover is not changed as much, and the volume of soil and colluvium is an order of magnitude greater. The relaxation time of Japanese watersheds to the Pleistocene-Holocene climatic change exceeds 10 ky and may be a factor in the present high watershed sediment yields (Oguchi, 1996). 2.5.4 Time Spans Needed to Erode Landforms Tectonic geomorphology studies focus mainly on the past 10 to 100 ky in areas of accelerated landscape evolution (rapid bedrock uplift, soft rocks, and extremely humid climate) and on more than 10 My in slowly changing pedimented landscapes of some arid regions. Hills and streams continue to change after tectonic uplift of mountains has virtually ceased. The time needed for erosion to create landforms indicative of tectonic stages of landscape evolution ranges from less than 1 ky to more than 1,000 ky. The time span needed for each landform noted on the left side of Figure 2.21 is a function of uplift, of Rising Mountains 57 rock resistance, and volume of material to be eroded after cessation of uplift. Only a short time is needed for the concentrated power of a stream to remove a small volume of unconsolidated alluvium to create a fanhead trench. Immense time spans (>10 My) are needed to consume the last vestiges of an uplifted planar surface. Such escarpment retreat is accomplished by gradual weathering of bedrock and slow erosion of hillslopes, and the volume of rock to be removed is huge. Isotopic ages allow rough estimates of the times needed to erode landforms in the Mojave Desert and the Coast Ranges of California. Potassium-argon ages for volcanic materials in mountains and basin fill range from 0.5 to more than 5 Ma. Granitic and metamorphic rocks predominate in the arid Mojave Desert, and soft mudstone and sandstone predominate in the semiarid to subhumid Coast Ranges. Erosion rates vary with climatic setting by at least two orders of magnitude (Fig. 2.20). The Mojave Desert and Coast Range plots on Figure 2.21 are separated by approximately an order of magnitude. The sheared granitic and metamorphic rocks of the semiarid to subhumid Transverse Ranges occupy an intermediate position. Estimates of denudation rates based on amounts of sediment trapped in 450 Eliminate planar uplands, form circular drainage basins Form pediment with inselbergs Eliminate triangular facets Embay mountain front Erode U-shaped valley Entrench alluvial fan 104 105 W6 107 Sequence of erosional stages Years since cessation of uplift Figure 2.21 Diagram comparing estimated times needed for changes in landforms after cessation of active uplift for different climates and rock types in 10 km2 fluvial systems. The Coast Range and Mojave Desert stages are spaced on the ordinate so that most points approximate straight lines. Plots without control points have less dating control. A. Sheared and fractured greywacke sandstone in humid New Zealand. F3. Soft mudstone and shale in the semiarid central Coast ranges of central California. C. Sheared and altered granitic and metamorphic rocks of the subhumid San Gabriel Mountains of southern California. Quartz monzonite in the arid Mojave Desert of California. E. Gneissic and granitic rocks in the extremely arid Sinai Peninsula of Egypt. After Bull (1955). 58 Chapter 2 debris basins (Scott and Williams, 1978; Brown and Taylor, 1982) suggest rapid denudation of the San Gabriel Mountains at about 1.5 m/ky. Extremes of rates of landscape evolution are represented by the easily eroded fractured greywacke sandstone and schist of the extremely humid Southern Alps of New Zealand and by the extremely arid Sinai Peninsula. A wide range of spatial and temporal scales of investigation is needed for the overdue incorporation of landscape analyses as an integral component of the plate tectonic paradigm. Process-oriented studies emphasize small spaces and time spans as short as the elapsed time since a recent earthquake (Arrowsmith and Rhodes, 1994). At the other extreme, spaces can be as large as mountain ranges, or entire tectonic provinces, and time spans may exceed 10 My (Davies and Williams, 1978; Oilier, 1982). 2.6 Tectonically Induced Downcutting Streams incise ever deeper as bedrock is raised into the powerful buzz saw of stream-channel downcutting. Amounts and rates of tectonically induced downcutting are functions of vertical tectonic displacement rates, excess unit stream power (equation 2.6), and resistance of earth materials to degradation. Downcutting by small ephemeral streams flowing over resistant welded tuff may be unable to match a bedrock-uplift rate of 0.1 m/ky; such reaches remain on the erosional side of the threshold of critical power. Downcutting by perennial rivers flowing over soft rock easily keeps pace with bedrock uplift of 5 m/ky. But stream-channel downcutting occurs only during appropriate climatic and tectonic conditions. The tendency of streams to cut down to the minimum gradient needed to transport their sediment load has been a long standing fundamental Figure 2.22 Sketch of two longitudinal stream profiles graded to similar sea-level highstands. Rivers erode down Into rising mountains and then widen their valley floors by beveling strath surfaces when they are not able to downcut further. In this case tectonically induced downcutting between times A and E3 has left the downstream reaches of the longitudinal profile as a strath terrace passing through the rising landscape. concept in fluvial geomorphology (Powell, 1875; Mackin, 1948; Leopold, Wolman, and Miller, 1964; Leopold and Bull, 1979; Bull, 1991). Headwater reaches of streams in rising mountains tend to stay on the degradational side of the threshold of critical power, but downstream reaches, with their greater unit stream power, are more likely to attain the base level of erosion through the process of tectonically induced downcutting (Fig. 2.22). 2.6.1 Straths, Stream-Gradient Indices, and Strath Terraces Many streams return to the base level of erosion after tectonically induced downcutting is interrupted by aggradation events that temporarily raise the streambed. The Charwell River, New Zealand fluvial system (Figs. 2.23, 2.24) is sensitive to both tectonic and climatic perturbations; it has frequent climate-change induced aggradation events, numerous internal-adjustment terraces, and occasional times when the stream bevels its valley floor to create a tectonic landform - a major strath. Prior piedmont valleys, with their flights of Pleistocene stream terraces, have been preserved. Their rich history of landscape evolution has been set to one side as a result of rapid right-lateral displacement of the watershed by the Hope fault, which is at the mountain front-piedmont boundary. The following discussion focuses on the present-day valley and its flight of terraces, whose creation was modulated by several late Quaternary global climatic changes. The mere presence of either marine or strath terraces has tectonic significance. Only one sea-level highstand was higher than the present high stand during the past 350 ka. It occurred at about 125 ± 5ka Concepts for Studies 4 km Figure 2.23 Drainage networks of the Main and Right Branches of the Charwell River, New Zealand, k is knickpoint migration from Hope fault since ~9 ka. K is where several knickpoints have accumulated to create a large step in the streambed. Numbers are for Strahler (1952, 1964) stream ordere for a third order tributary. From Infomap 260 031, New Zealand Department of Survey and land Information. when the oceans were about 5 to 6 m above present sea level (Chappell, 1983, 2001; Chappell et al., 1996; Israelson and Wohlfarth, 1999). So just the presence of coastal shore platforms higher than 6 m shows that the land is rising relative to the sea-level datum. Similarly, the presence of flights of paired strath terraces shows that the terrain is rising, relative to the long-term base level of erosion of the stream. Dating the times of formation, and measuring heights, of either strath or marine terraces provides estimates of bedrock-uplift rates. Reaches of the Charwell River upstream and downstream from the range-bounding Hope fault have different styles of response to uplift. This part of the Seaward Kaikoura Range is rising three of Rising Mountains 59 times faster than the adjacent piedmont reach. Rock mass strength also is much greater in the mountains, where unit-stream power becomes progressively less farther upstream. The longitudinal profile in the mountains has the characteristics of a disequilibrium stream, whereas the river flowing down the piedmont easily achieves type 1 dynamic equilibrium. Tectonic strath terraces of the Charwell-River reach downstream from the front of the Seaward Kaikoura Range illustrate the importance of this landform to tectonic geomorphologists. The longitudinal profile is much more concave upstream from the range-bounding Hope fault and the average gradient is fivefold less downstream from the fault (Figs. 2.23, 2.24). The marked change in concavity of the two reaches mainly reflects rapid uplift of more resistant mountain bedrock, and pronounced overall widening of active-channel streamflow in the piedmont reach. Downstream increases of discharge and decrease in size of bedload are only moderate in this short distance, so may be less important than litho-logic and tectonic controls. Slower uplift and softer rocks in the piedmont reach favor prolonged attainment of the base level of erosion at the conclusion of degradation events that follow pulses of aggradation. Valley-floor portions of fault zones were buried beneath thick alluvium during aggradation events. The highly irregular longitudinal profile upstream from the Hope fault in part reflects cumulative surface ruptures as much as 40 m that were not able to migrate upstream until bedrock beneath episodic valley fill was exposed to erosion. We need ways to quantify both the irregularities in the longitudinal profile upstream from the Hope fault and degree of smoothness downstream from the fault zone. Stream-gradient indices are introduced as a valuable concept here. John Hack used characteristics of large rivers in the humid Appalachian Mountains of the eastern United States to define a stream-gradient index that describes influences of many variables that influence the longitudinal profiles of stream channels (Hack, 1957, 1973, 1982). Equilibrium adjustments, termed hydraulic geometry, assume orderly interactions between streamflow variables. Hydraulic geometry of stream channels is based on stream-gauging data, and typically has an order of magnitude scatter of data when used in logarithmic regressions of discharge and streamflow characteristics (Leopold and Maddock, 1953; Leopold, Wolman, and Miller, 1964). It defines statistical relationships between streamflow 60 Chapter 2 \200 < 800 400 Piedmont reach Stream length, km Figure 2.24 Longitudinal profile of the Charwell River from the headwaters to the junction with the Conway River, South Island, New Zealand. SL is stream-gradient index, k is a knickpoint that has migrated upstream from the Hope fault where it originated as a fault scarp beneath alluvium between 26 and 9 ka. K is larger multiple-event knickpoint. From Figure 4 of Bull and Knuepfer (1957). parameters and channel morphologies. Using the approximate relations provided by hydraulic geometry dispenses with having to measure streamflows in virtually inaccessible localities. Discharge (Q) from a watershed increases as a power function with drainage-basin area {AJ: Q ca; (2.7) Many studies have compared length down a stream channel, L, from the main divide with drainage basin area, A , and have found that L increases at least as rapidly as Ad, (exponent is >0.5), bA?£ o (2.8) where the units for L are miles and for A are square miles. This exponential function is now revered as "Hacks law" and has been the subject of many re-evaluations (Smart and Surkan 1967; Mueller, 1972; Seidl and Dietrich, 1992; Montgomery and Dietrich, 1992; Rigon et al., 1996). Experimental watershed studies by Lague et al. (2005) found that power functions of mainstream length increase almost linearly with drainage area, and that Hack's law is not significantly dependent on uplift rate. The systematic decrease in slope as described by concave longitudinal profiles of stream channels is nicely described by a power function between the slope of a reach of a stream, 5 , and A,. S = kA, (2.9) Hack used length, L, as a reasonable proxy for discharge, Q. He tested whether streams had achieved equilibrium by analyzing spatial variations in the product of slope of a reach, AH/ALr (change of altitude/length of reach) and the horizontal length to the midpoint of the reach from the watershed divide, Lsc. This is the "stream gradient index", or SL index, where SL is defined as: SL = A H A L (2.10 Verification of Hack's SL model was achieved when he showed that reaches of Appalachian rivers had fairly constant values of SL for consecutive reaches. The Appomattox River has remarkably constant SL values over a distance of 150 km (Hack, 1982). See the Figure 2.30 example discussed in Concepts for Studies Section 2.7. This implies that 1) larger "bankfull" flow events were cumulative responses of the entire watershed, 2) streambed hydraulic roughness is constant downstream, and 3) bedload size and transport rate remained about the same downstream. Analyses of stream-gradient index should be classed as narrow or inclusive. Narrow pertains to an anomalous stream-gradient index. It typically is only a short reach that describes only one or several contour intervals. It is useful for describing magnitudes of longitudinal profile abnormalities caused by locally high rock mass strength and/or knickzones that have migrated upstream from a source of base-level fall perturbations. Inclusive pertains to long reaches of a longitudinal profile that have a constant rate of longitudinal-profile decrease in gradient associated with progressively larger streamflows from a headwater divide source area. Adjacent reaches with dissimilar inclusive gradient-indices describe variations in longitudinal profile caused by factors such as adjustments to spatially variable uplift rates (Keller and Rockwell, 1984), changes in rate of downstream increase of stream discharge, change in the direction of a valley, and change in median particle size of gravelly stream-beds. Narrow gradient-indices describe disequilibrium reaches of streams. Inclusive gradient-indices can be used to describe situations of type 1 or type 2 dynamic equilibrium. Narrow SL values of 1300, 97, and 1900 for adjacent reaches upstream from the Hope fault (Fig. 2.24) record the inability of the Charwell River to smooth out some irregularities in the longitudinal profile. These anomalies result from frequent large tectonic displacements of the streambed and variable rock mass strength of greywacke sandstone. The heights and present positions of these knickpoints are also a function of late Quaternary climate change. Stream-gradient index analyses may not apply equally well to all streams. I suspect that this index describing the behavior of an erosional fluvial system should not be used where streams are aggrading. Of course, a brief pulse of deposition that has uniform thickness would not change analysis results because erosional processes prior to deposition created the form of the longitudinal profile. Application to ephemeral streams should proceed with caution, especially where most convective-storm rainfalls generate flash floods over only part of a watershed. Infiltration of streamflow into a dry streambed results in progressive decrease of Rising Mountains 61 in discharge, which is opposite to the trend of the large perennial rivers where Hack defined the stream-gradient index. Hack's model assumes that longitudinal-profile concavity results from ever-increasing stream power in the downstream direction. Concentration of sediment load concurrent with decreasing discharge of water can move a degrading ephemeral streamflow closer to, or across to the depositional side of, the threshold of critical power. For such reasons, ephemeral streams typically have longitudinal profiles that are much less concave than for humid region watersheds that generate bankfull discharges of similar size (Wolman and Gerson, 1978). We can expect stream-gradient indices to increase downstream, even in equilibrium reaches, where ephemeral streamflow behavior has constant or even decreasing stream power in consecutive downstream reaches. Hack also noted that long equilibrium reaches of perennial rivers plotted as a straight line on semilogarithmic regressions of altitude H and In of Lsc. Each long reach can be numerically described by an inclusive gradient-index. Such linear relations are described by H = C-k(lnL) (2.11) C is a constant and k is the inclusive gradient-index (slope of the regression). The derivative of equation 2.1Ö with respect to L is streambed slope, S: c, _dl± d(kinL) k dL~ dL ~ L (2.12) The inclusive gradient-index can be estimated by regression analysis or by using data points from the longitudinal profile: {inL-lnL) (2.13) where H and H. are the altitude and distance from the watershed divide for an upstream point on the stream channel, and H. and L. are for a downstream point on the longitudinal profile. Examples from the Charwell River are introduced here. The Right Branch of the Charwell River is presently beveling a strath as it flows from the Hope fault to its junction with the Main Branch. 62 Chapter 2 =s -P < 500 450 400 350 0.65 8.75 8.85 8.95 9.05 In (Horizontal distance (L) along valley floor from watershed divide) 450 350 =s P P < 250 I—i—i—i—|—i—i—i—|—i—i—i—|—i—n—r Junction with Right Branch Crossover point sed for both plots r0 150............................. 8.8 9 9.2 9.4 9.6 9.8 10 10.2 In (Horizontal distance (L) along valley floor from watershed divide) Figure 2.25 Inclusive stream-gradient indices for piedmont reaches of the Charwell River. A. Right Branch from Hope fault to junction with the Main Branch. F3. Main Branch from Hope fault to the junction with the Conway River. The crossover point is used in both regressions. A semi-log regression of altitude and distance from the headwater divide (Fig. 2.25A) indeed plots as a straight line. The perfect correlation coefficient is in part due to soft mudstone bedrock beneath the stream channel, and the lack of topographic obstacles. Of course correlation coefficients tend to be high where cumulative altitude is regressed against cumulative distance. The inclusive gradient-index for this fairly small stream is 260. A similar analysis for the Main Branch also demonstrates attainment of equilibrium conditions. The complete dataset has a correlation coefficient of 0.994 and an inclusive gradient-index of 214, just what one would expect for a stream whose watershed is three times larger than that of the Right Branch. Alternatively, the Main Branch can be modeled as two reaches with different characteristics. This improves the correlation coefficients slightly. The trend of consecutive points for the reach upstream from the crossover point does not reveal where the Right Branch enters the Main Branch or where the Main Branch narrows where it flows through a gorge cut in massive sandstone. The upstream reach flows straight down the piedmont at a bearing of 170°. The river impinges on Flax Hills at the location of the crossover point. This topographic obstacle deflects the course of the river by 50°, changing the direction of the valley to a bearing of 230°. The steeper regression trend for the downstream reach has an inclusive gradient-index of Concepts for Studies 245. The contrast between gradient indices of 184 and 245 supports treating these as two datasets, thus counteracting the initial impression. Evaluation of the equilibrium stream channels of the Right Branch and Main Branch for reaches just downstream from the Hope fault is more appropriate for comparison of gradient-indices for these humid region streams, 184 for the Main Branch and 260 for the Right Branch. Straths are beveled along reaches of streams at equilibrium - where the inclusive gradient-index remains constant from reach to reach. It is useful to viewstrath formation in a context of a strath-formation threshold defined as the effective stream power needed to mobilize streambed materials, and above which the stream can do the work of beveling a strath. This threshold is reached more often in downstream reaches of a stream as is suggested by the general lack of straths in the upstream half of most watersheds. Climatic and lithologic variables play critical roles in determining the wide range of conditions affecting the strath-formation threshold for a specific drainage basin. Rainfall-runoff magnitudes and rates greatly affect peaks and durations of streamflows and amount and size range of suspended and saltating sediment. Such interactions between the fluctuating variables over the long term affect the numerical values for the inclusive gradient-index. A broad modern strath in the reach immediately downstream from the Hope fault has been beveled across soft sedimentary rocks (Fig. 2.26) and is indicative of prolonged attainment of type 1 dynamic equilibrium. This late Holocene valley floor is a nice example of a steady-state landform. Peter Knuepfer (1988) did the weathering-rind dating of exposed of Rising Mountains 63 boulders on the treads of the flight of degradation terraces (Fig. 2.16). The lower strath terraces of the flight are shown here. The scarp of the oblique right-lateral Hope fault is the bushy riser at the far end of the pasture. This is a nice example of attainment of equilibrium for a specific reach of the fluvial system (Fig. 2.25B). The Charwell River quickly re-established the base level of erosion here many times after brief departures during the past 4 ka. These minor variations in streambed altitude are merely the product of the normal fluctuations in the spectrum of discharge of water and sediment. Strath terrace heights in the Flax Hills reach, 7 to 8 km downstream from the Hope fault, were surveyed using the modern strath as a reference level (Bull, 1991, Fig. 5.19). It is 2.5 ± 0.5 m below the active channel of the Charwell River. Seven radiocarbon ages on fossil wood collected from basal aggradation gravels just above several older straths reveal that tectonically induced downcutting in this reach is 0.37 ± 0.03 m/ky. Assuming that this bedrock-uplift rate was uniform during the past 200 ka, one can estimate strath ages by dividing strath height by tectonically induced downcutting rate. For example, the strath presently at 30.3 ± 0.5 m above the modern strath is estimated to have formed at about 82 ka. 30.3 m 30.3 m/ky' 82±8ka (2.14) Times of tectonic strath formation occurred at approximately 0, 40, 54, 62, 72, 82, and 114 ka in the Flax Hills reach. Times of strath terrace forma- iHope Faultl Figure 2.26 Reach of the Charwell River downstream from the Hope fault has been at the base level of erosion. The resulting steady-state landform is a 400 m wide strath cut in soft Ceno-zoic basin fill that is capped with a veneer of stream gravel. This surface of detrital transport since 4 ka continues to be lowered at a rate equal to the rock-uplift rate. 64 Chapter 2 Strath height, m Marine terrace age, ka Inferred strath age, ka 0 0 0 * 29 # 14.9 40 40 44 # 20.1 53 54 23.1 59 62 26.6 72 72 30.3 81 82 96 # 100 # 42.1 118 114 124 53.7 No match for 54 m strath 145 62.0 176 168 Table 2.2 Relations between technically induced valley-floor down-cutting and inferred ages of major straths of terraces along the Flax Hills reach of the Charwell River, South Island, New Zealand. Marine-terrace ages are from Chappell and Shackleton (1956); and Shackleton (1957). * River cut a strath near the mountains, but only incised part way through the Flax Hills aggradation-event alluvium in the study reach. # Tectonic strath of this age is present in another reach of the River tion at about 29, 44, 96, and 100 ka were observed in other reaches, but not in the Flax Hills reach because 1) erosion has removed the strath, 2) the strath was not exposed at the time of my survey, 3) or insufficient vertical separation to distinguish between adjacent straths because of the locally low bedrock-uplift rate. Strath terrace age estimates for the Charwell River (Table 2.2) coincide with the ages of dated global marine highstands of sea level (Chappell and Shackleton, 1986; Gallup et al., 1994). The coincidence between the inferred ages of Charwell River straths and the isotopic ages of global marine terraces is the result of similar timing of climate-change modulation of marine coastal and fluvial geomorphic processes during the late Quaternary. Both the Charwell fluvial system and the coastal marine system are controlled by global climatic changes that fluctuate between full-glacial and interglacial extremes. Times of rapid aggradation of New Zealand valleys occurred at times of maximum accumulation of ice on the continents and lowstands of glacio-eustatic sea levels. Downstream reaches of these streams easily attain type 1 dynamic equilibrium during interglacial climates at times that coincide with the times of maximum melting of continental ice masses and attainment of sea-level highstand. Hillslope plant cover and geomorphic processes in the Charwell River watershed were greatly different for these two regimes (Bull, 1991, Chapter 5). The 29, 40, and 54 ka tectonic straths of the Charwell and nearby rivers can be identified by radio- carbon and luminescence dating of the adjacent overlying deposits. Together with the modern (0 ka) strath, they provide readily accessible time lines to assess local rates of tectonically induced downcutting of the valley floor. For example, the 0 and 29 ka straths can be used to estimate the bedrock-uplift rate for a reach of the Charwell River that is 1 km downstream from the Hope fault. Analyses of weathering-rind thicknesses on surficial greywacke cobbles by Knuepfer (1984, 1988) were the basis of the 10.8 ± 1.9 ka age estimate for the tread of the fill-cut degradation terrace shown in Figure 2.27. The lack of paleosols or beds of loess in the basal 23 m of uniformly massive sandy gravels is suggestive of a single pulse of aggradation of Stone Jug gravels. The process of returning to the Charwell River base level of erosion involves stream-channel downcutting through the aggradation event gravels and then through a thickness of bedrock equal to the total rock uplift since the time of the pre-aggradation event strath. Holocene degradation of an additional 39 m below the 29 ka strath occurred in this reach. The rate of tectonically induced downcutting is 39 m 29ky = 1.3 ± O.lm/ky (2.15) Two additional points are illustrated by Figures 2.26 and 2.27. Even the small 30 km2 main branch of the Charwell River may easily attain the base level of erosion in downstream reaches for sufficiently long time spans to bevel tectonic straths (type Concepts for Studies of Rising Mountains 65 climate change in much of the South Island of New Zealand are nicely synchronous because of fairly similar topography, lithology, and humid climate. Brief regional aggradation events can also have nontectonic origins such as regional coseismic landslides (Hancox et al., 2005). Termination of periods of strath formation is much more likely to be synchronous. Studies of the Greenland ice cores give us a better appreciation of the rapid onset of major climate changes (Alley, 2000; Peteet, 2000). The age uncertainties for isoto-pically dated times of global marine terraces are less than the ±5 to ±10 ky uncertainties for strath ages that are a function of surveying and uplift-rate-calibration errors (see equation 2.14 for an example). 2.6.2 Modulation of Stream- Terrace Formation Figure 2.27 View of 11 ka Charwell River fill-cut terrace. About 23 m of gravel lie on a tectonic strath that formed at about 29 ka. The 39 m between the buried strath and the present tectonic strath reflects the amount of tectonically induced river downcutting since 29 ka: the basis for estimating an uplift rate of 1.3 ±0.1 m/ky. 1 dynamic equilibrium). Second, it is not necessary to preserve a complete section of aggradation gravels in order to identify the aggradation event that buries a tectonic strath. The excellent agreement between ages of global marine terraces and local tectonic stream terraces ties the Charwell River terrace chronology to a global climatic chronology. Global climatic changes result from variations in the Earth's orbital parameters - the astronomical clock - (Berger, 1988). The similarity of the Table 2.2 pairs of ages may permit assignment of ages for straths with less radiocarbon dating control. Potential dating uncertainties for straths older than 40 to 50 ka include violation of the assumption that both systems have similar response times to global climatic perturbations. This can happen for aggradation events because watershed characteristics are sufficiently variable to result in crudely synchronous, or even diachronous, aggradation surfaces for a suite of adjacent watersheds. Watershed responses to by Pleistocene—Holocene Climatic Changes Times of formation of tectonic landforms commonly reflect other important variables of fluvial systems such as annual unit stream power (a measure of a stream's capacity to do work). Times of tectonic strath formation along the Charwell River were largely controlled by the rather overwhelming influence of late Quaternary climatic changes. Climatic-change impacts of watershed geomorphic processes raise and lower the streambed at rates faster than the concurrent bedrock uplift caused by the sum of tectonic forces and isostatic adjustments (Fig. 2.28). The piedmont reach of the Charwell River was either aggrading or was catching up to new base levels of erosion. This reach was raised by a combination of uniform rapid bedrock uplift and intermittent valley-floor backfilling of 30-60 m. Aggradation events were the dominant process during the Pleistocene, whereas the Holocene has been characterized by degradation. In order to occasionally catch up to a new tectonic base level of erosion the stream had to degrade through the most recently deposited valley fill, and then through a thickness of bedrock equal to the amount of bedrock uplift since the last time the stream attained type 1 dynamic equilibrium. The Charwell River barely had enough time to bevel a new tectonic strath after attaining the base level of erosion, before the onset of the next aggradation event. These brief episodes of attainment of equilibrium allow comparison between the times of strath cutting with the times of solar insolation maxima and sea-level rise. The agreement 66 Chapter 2 Degradation \owere the valley floor Stream cute down 'and remains at new base level of erosion -1-1-1- Valley floor raised by climate-change induced aggradation ■ 120 Times of tectonic strath terrace formation^ z so i 40 0 o f SI in O 20 30 Time, ka would not be nearly as nice (Table 2.2 and Figure 2.28) for larger streams that may remain at the base level of erosion for 60-90% of the time. One reason for distinguishing between type 1 and type 2 dynamic equilibrium is that tectoni-cally induced downcutting can be used to estimate uplift rates only when comparing situations of type 1 dynamic equilibrium. These streams have parallel longitudinal profiles of stream terraces that indicate return to similar combinations of variables for reaches where channel width is less than valley-floor width. This is not the case for type 2 dynamic equilibrium streams incising into bedrock. Longitudinal profiles may be concave and exponential, but unfortunately we can only examine the present assemblages of land-forms because type 2 streams do not create suitable landforms that are preserved. Consider the Grand Canyon reach of the Colorado River in northern Arizona. Active normal faulting at the western end of the Canyon during the Quaternary caused tectonically induced downcutting of roughly 0.4 m/ky but this decreased to 0.2 m/ky at 100 km upstream in the eastern reach of the Canyon (where strath terraces are more likely). It appears that this 100 km long reach has been steepened 400 m in the past 2 My. Steeper gradient and narrower channel width are the obvious consequences, but changes in hydraulic roughness may be just as profound. These several alterations do not let us use changes in the altitude of the longitudinal profile to estimate uplift rate. Furthermore, influx of large boulders from cliffy tributary streams does not allow the river to behave Figure 2.25 Changes in the streambed altitude of the Charwell River, New Zealand reflect the combined influence of tectonic and climatic controls during the past 45 ka. Tectonic strath terraces are created only during brief time spans that follow climate-change modulated episodes of tectonically induced down-cutting. Simplified from Figure 5.24 of Bull (1991). as a system of interrelated reaches (Figs. 2.19C, D). Waterfalls in the Charwell River upstream from the Hope fault also limit use of fluvial landforms to estimate uplift rate. 2.7 Nontectonic Base-Level Fall and Strath Terrace Formation Not all strath terraces represent time lines in tectonically deforming landscapes. So let us clarify other aspects of this valuable landform with examples of the few exceptions to what might have seemed a general rule in the preceding discussions. The most obvious nontectonic strath is an unpaired terrace resulting from local lateral migration of a stream into a bedrock hillslope. Such a nontectonic strath could even form while a valley floor is being slowly raised during the terminal stages of an aggradation event. Pauses in a degradation event may temporarily allow a stream to bevel either fill-cut surfaces in alluvium or strath surfaces in bedrock before the stream has downcut sufficiently to return to a new base level of erosion. These common erosion surfaces are nontectonic internal-adjustment terraces (Charwell River at - 14 to 4 ka (Fig. 2.16) for example). Base-level falls can be induced by climatic perturbations to fluvial systems as well as by uplift. A good example is from the piedmont along the mid-Atlantic coast of the eastern United States. Isostatic uplift continues at a very slow rate in response to Concepts for Studies gradual erosion of the Appalachians and tectonic uplift generally is so minor as to be trivial. So, this would seem to be an improbable region to observe large amounts of stream-channel downcutting below prominent strath levels. But, beautiful, prominent strath surfaces occur along the lower reaches of large rivers that have cut spectacular bedrock gorges just above their terminal tidewater reaches. The Great Falls are in the terminal reach of the Potomac River west of Washington, DC (Fig. 2.29). The prominent strath surface was beveled across the highly resistant late Proterozoic sandstone and schist of the Mather Gorge Formation. Cosmogenic 10Be and 26Al dating of 18 samples collected from Mather Gorge downstream from the Great Falls of the Potomac River (Bierman et al., 2004; Reusser et al., 2004) indicate rapid incision of the strath began at about 30-32 ka. This strath was beveled during the preceding 50 ky and perhaps during an even longer time span. The 22 m height of the strath terrace shown in Figure 2.29 surely cannot represent the consequences of either tectonic or isostatic uplift in this region in such a short time span. A detailed study by Bierman et al. (2004) concludes that gorge incision coincides with a 50 m decline in sea level at the beginning of the most recent ice age. Sea level continued to fall to more than 100 m below the present level (Fig. 1.19) - a major climatic perturbation. The regional nature of the perturbation is suggested by synchronous similar straths in other coastal plain rivers such as the lower Susquehanna, Rappahanock, and James Rivers. Processes of incision by large and small rivers into bedrock intrigue geomorphologists (Tinkler and Wohl, 1998a, b) because the valley floor is the base level for all adjacent hillslopes. River incision rates control the rate of landscape response to changes in rock uplift rate and Quaternary climate change (Howard, 1994). Seidl and Dietrich (1992) assume that incision rate is proportional to stream power, but this model may not explain some variations in incision rates (Stock and Montgomery, 1999) and longitudinal profile shapes (Sklar and Dietrich, 1998). The mix of variables surely has to include rock mass strength (Selby, 1982b, Moon, 1984) and bedload size and amount (Sklar and Dietrich, 2001). I would add that information about all of the above variables is of little use unless one knows how far removed a reach of a river is from the base level of erosion. Gilbert (1877) knew that streams close to the base level of erosion or on the aggradational side of Rising Mountains 67 of the threshold of critical power would be limited in their capacity to incise into bedrock - regardless of unit stream power, rock mass strength, or bedload transport rate - because the valley floor would be mantled with protective alluvium much of the time. However, a streambed far to the degradational side of the threshold will be exposed to valley-floor degradation. The magnitude of departure from the threshold of critical power is the crucial factor and is controlled by both climatic and tectonic factors. Application of such stream power/resisting power ratios for adjacent aggrading and degrading reaches is illustrated in Figure 6 of Bull (1997). Field-work based modeling needs to focus more on the locations and time spans where excess stream power is available to incise valley floors. The Potomac and other rivers extended downstream into newly exposed reaches of the continental shelf that were steeper than the prior tidal reaches. Terminal reaches either dropped over the edge of the continental shelf, or were steepened by dropping into the head of a submarine canyon (Tailing, 1998, Fig. 3). Assuming no change in discharge of either water or sediment, such rivers would then have terminal reaches strongly on the erosional side of the threshold of critical power. The potential for erosion would have been still greater if sea-level fall coincided with either a decrease of bedload transport rate or an increase in stream discharge. Figure 2.29 Broad strath terrace approximately 22 m above the present floor of the Potomac River at Great Falls national Park, Maryland. Strath incision of this magnitude is anomalous in this tectonically inactive setting. 68 Chapter 2 200 < 100 Local base level K 750 Shaly siltstone Limestone Sandstone Predominant rock type in reach Quartzite Gneiss 300 400 500 km Figure 2.30 Stream-gradient indices for the Potomac Piver reach upstream from tidal Mather Gorge. Inclusive gradient-Indices of the longitudinal profile (low SL values) approximate straight lines on this semi-logarithmic graph have achieved type 1 or type 2 dynamic equilibrium. Stream-gradient index analysis uses a logarithmic scale of distance, in this case from the headwaters of the South Branch of the Votomac River, k is the 30 ka knickpoint of the Great Falls of the Votomac. I\ is reach with exceptionally resistant lithology that tends to arrest and accumulate upstream migrating knickpoints. Both are narrow gradient-Indices. Graph and rock types from Flack (1973). A second, much different type of climatic perturbation might have played an important role in strath-terrace formation. Reusser et al. (2004) point out that massive loading by nearby continental glaciers would have the effect of depressing the crust beneath the ice, which would tend to create a belt of rock uplift in front of the continental glacier. This creates an ephemeral crustal upwarping - a foreb-ulge. Such raising of the land surface in the terminal reaches of large coastal rivers in front of the ice would initiate stream-channel downcutting that has continued to the present. Timing of strath incision could also be modulated by local presence of resistant valley floor rocks that could delay the onset of accelerated downcutting. Potomac River incision appears to have migrated rapidly upstream; Bierman et al. (2004) conclude that the initial knickpoint migration required only a few thousand years. The river continues to deepen Mather Gorge at a uniform rate of about 0.8 m/ky since 35 ka. Similar climatic perturbations may have played a role in forming the classic Schooley and Somerville erosion surfaces of Davis (1890) in the coastal Appalachian region. Stanford et al. (2001, 2002) use the post late-Miocene decline in sea level to account for two episodes of valley-floor incision and strath-terrace formation. Stepwise glacioeustatic events dominated landscape evolution on this low-relief passive margin. Incision occurred when global ice mass increased, thus causing sea-level decline. Times of formation of broad straths coincided with long periods of stable or rising sea level. The resulting flexural isostatic uplift of the area was the product of both mountain-range denudation and shifting of the resulting sediment to coastal basins of deposition. Estimated regional long-term uplift rates are minis-cule, being only about 0.02 m/ky, and in contrast to rapid gorge incision of Mather Gorge between 37 to 13 ka of about 0.8 m/ky (Reusser et al., 2004). The downstream reaches of the Potomac River have SL indices that clearly show either attainment of equilibrium conditions, or have pronounced knickpoints indicative of obvious disequilibrium (Fig. 2.30). The Great Falls is a late Quaternary feature but the resistant quartzites of Blue Ridge may act as a long-term local base level that accumulates upstream-migrating knickpoints in much the same way as noted for the Charwell River of New Zealand (Fig. 2.24). This lithologic control is a long-term impediment to the continuity of the Potomac fluvial system. Combination of Cenozoic flexural isostatic uplift and upstream knickpoint migration can tern- Concepts for Studies porarily accelerate stream-channel downcutting, thus facilitatingpreservation of remnants of old erosion surfaces (Pazzaglia and Gardner, 1993, 1994, Pazzaglia et al., 1998; Zaprowski et al., 2001). Strath terraces have formed in tectonically inactive regions such as Australia (Goldrick and Bishop, 1995) and in southern Arizona; both areas have minimal connection to sea-level fluctuations. Such landforms have great antiquity because they are the result of valley floor downcutting in response to isostatic uplift resulting from gradual erosion over millions of years. Bedrock uplift in this case is purely isostatic. Times of strath incision, as for tectonically active settings, occur at times when climatic controls favor strong departure from the threshold of critical power by altering discharge of water and sediment. They also influence the concavity of river longitudinal profiles (Zaprowski et al., 2005). Tectonically inactive Appalachian streams have more concave longitudinal profiles where peak annual discharge is greatest, suggesting that this factor may be more important than watershed size in determining rates of valley floor incision. Location within a fluvial system also is important, so we next turn our attention to a dimensionless way of defining locations within watersheds. 2.8 Hydraulic Coordinates This concluding section of Chapter 2 introduces a way of locating hydraulic positions within smaller elements of fluvial systems - hillslopes and drainage basins. Use of hydraulic coordinates is a dimension-less format that facilitates comparisons between large and smaller drainage basins. A good way to relate tectonics to topography is with models that quantify spatial and altitudinal positions of water and sediment flux in fluvial systems. I do this with dimensionless hydraulic coordinates for points on local hillslopes and for watershed locations relative to the trunk valley floor of a drainage net. Active range-bounding faults mark abrupt transitions between the erosional and depositional domains of many tectonically active fluvial systems. Tectonic geomorphologists use hydraulic coordinates to relate the position of an active fault or fold to nearby or distant parts of a fluvial system, and to compare the morphologies of hills and streams in different tectonic and climatic settings. Two planimetric coordinates are described in terms of ratios of horizontal distances of flow direction down hills and streams. The hillslope-posi- of Rising Mountains 69 tion coordinate, Hpc, is the planimetric length from a ridgecrest divide to a point on the hillslope, L , divided by the total length from the divide to the edge of the valley floor, L{. H = -± pc L . (2.16) Hillslope-position coordinates range from 0.00 at the ridgecrest to 1.00 at the base of the footslope. The basin-position coordinate, Bpc, is the planimetric length, L , from the headwaters divide of a drainage basin along the trunk valley to a stream-bed point divided by the total length of the valley, L . from the headwaters divide to the mouth of the vt drainage basin. The point in the valley floor should be directly down the flow line from a hillslope point of interest, which commonly is directly opposite the stream-channel point. (2.17) Total length is not measured along a sinuous stream channel, because this is a landform that changes too quickly for our longer-term perspective, nor along a chord between two endpoints. Instead, the basin-position coordinate describes distance along the trend of a valley. Interpretation of the numerical values is fairly straightforward. For example, 0.50R/0.67 describes a point half way down the right side hillslope for a point whose flux of sediment and water is two-thirds of the distance through a drainage basin. Right (R) and left (L) sides are when looking downstream. The aerial photos used in Figure 2.31 illustrate descriptions of hillslope- and basin-position coordinates. Locations of points along trunk streams are easy to define in low-order basins (Fig. 2.31A); flow proceeds from points 1 to 2 to 3 whose hydraulic coordinates are listed in Table 2.3. Point 2 marks the location where fluxes of water and sediment from a small tributary valley enter the trunk stream. Point 4 is a third of the way down the right-hand side hillslope opposite a basin-position coordinate of 0.48. Note that I have made a subjective decision to regard rills as part of this hillslope instead of as lower order tributary valleys. This type of subjective decision is always present and is largely a matter of scale and mode of depicting drainage nets, and the purposes 70 Chapter 2 Figure 2.31 Aerial photographs of drainage basins eroded into soft rocks of the eastern Rodman Mountains, centrai Mojave Desert, southern California. Hydraulic coordinates for each numbered point are listed in Tables 2.3 and 2.4. A. Low-order watershed with points to illustrate basic concepts of hydraulic coordinates. F3. Complex watershed with points to illustrate numbering for nested basin-position coordinates. of your investigation. The rills, which would not be apparent on conventional topographic maps, are useful. In this case, they alert us to measure distance along flow lines that are not at right angles to the trunk valley floor in much of this watershed. The left-side position of point 5 is denoted with an L, and this slope also drains directly into the trunk channel. Basin-position coordinates can also be used to describe points on hillslopes within multiple orders of nested tributary valleys. The first step in defining hydraulic coordinates in more complex watersheds is to identify the trunk channel of the longest sub-basin, from the headwater divide to the mouth of the drainage basin. We then identify the junctions at which water and sediment from tributaries join this trunk stream. By working our way upstream we go from the mouth of the master stream to the sources of sediment and water. Two large sub-basins are shown in Figure 2.3IB. The longest flow path is through the southern sub-basin, which in this case has a larger watershed area than the northern sub-basin. Flow begins at point 1 and leaves the watershed at point Concepts for Studies of Rising Mountains 71 Hydraulic coordinate number daein-poeition coordinate Hillelope-poeition coordinate 1 0.00 0.00 2 0.27 1.00 3 1.00 1.00 4 0.4Ö 0.30R 5 0.70 0.72L Table 2.3 Basin-position and hillslope-position coordinates for points in the iow-order drainage basin shown in Figure 2.31A. 6. Points 2, 3, 4, and 5 are the locations of several stream junctions whose tributary valleys contain the hillslope points of Table 2.4. Each stream junction marks an abrupt increase of discharge from tributary watersheds that here are progressively larger in the downstream direction. Hillslope-position coordinates are the same as for the simple watershed of Figure 2.31 A, but basin-position coordinates should include the location of the tributary in the nested hierarchy of stream orders. For example, points 7 through 12 are all located in the tributary that joins the trunk channel at a basin-position coordinate of 0.85, so (0.85) is placed before the basin-position coordinate in the tributary. Point 10 is on a left-side hillslope of a small basin that has a third-order relation to the trunk valley floor. Its location would be described by [0.85, 0.68) 0.70; 0.27L]. The number 0.85 tells us where flow from the large second-order tributary joins the third-order trunk stream. The number 0.68 refers to the fact that this first-order stream joins a second-order stream at a position that is 68% of the distance from the headwater divide of the second-order basin. Flow passing through hillslope-position coordinate 0.27L passes directly to first-order channel at the local basin-position coordinate of 0.70. Similarly, point 13 is 40% down a left side hillslope and 52% down a first-order basin that drains to a second-order basin at 90% of its length, which in turn drains to the third-order stream at 57% of its length: [0.57, 0.90) 0.52; 0.40L]. Assignment of relative planimetric positions is flexible, depending on map and image scales and the needs of the geomorphologist. Dimensionless numbers can also be used to describe relative relief positions. The hillslope-fall ratio, Hf, is the decrease in altitude from a ridgecrest divide to a point on the hillslope, H, divided by the total decrease in altitude from the divide to the base of the footslope, H . _ H Hf = -H (2.18) th The basin-fall ratio, Bf, is the decrease in altitude from the headwaters divide along the trunk valley to a point in the valley floor, H , divided by the total watershed relief (decrease in altitude from the headwater divide to the mouth of the drainage basin) R. H (2.19) Hillslope and basin-fall coordinates for a sequence of nested drainage basins can be described in much the same way as for hillslope- and basin-position coordinates by using percentages of total relief in each sub-basin draining to the trunk stream. Dimensionless hydraulic coordinates and ratios minimize the factor of size in comparisons of fluvial landscapes. The basin-position coordinate can be used to describe knickpoint migration in adjacent watersheds of different sizes and stream powers. In Chapters 2 and 3 it will be recommended that a standard basin-position coordinate of 0.9 be used to evaluate valley floor width-valley height ratios in relation to range-bounding faults. Hillslope-position coordinates can be used to standardize survey procedures. Uses include descriptions of features of the relative locations of change from convex to concave slopes, landslide head scarps, and where rills start to incise with respect to their source ridgecrests. I use hydraulic coordinates to examine tectonic signatures in hillslope morphology. I also prefer to use the basin-fall ratio for dimensionless analyses of the stream-gradient index (Hack, 1973) and for semi-logarithmic longitudinal profiles of streams. The longitudinal-profile of the trunk stream of the Right Branch of the Charwell River (Fig. 2.32) 72 Chapter 2 Hydraulic coordinate number Baein-poeition coordinate Hillslope-position coordinate 1 0.00 0.00 2 0.33 wo 3 0.46 wo 4 0.57 wo 5 0.65 wo 6 1.00 wo 7 (0.65) 0.00 0.00 6 (0.85) 0.31 0.67R 9 (0.85) 0.44 0.51 L 10 (0.85, 0.68) 0.70 0.27L 11 (0.85) 0.68 1.00 12 (0.85) 0.70 0.81R 13 (0.57 0.90) 0.53 0.40L 14 (0.57) 0.28 0.52L 15 (0.46) 0.87 0.39R 16 (0.33) 0.45 0.36R Table 2.3 Basin-position and hillslope-position coordinates for points In the more complex drainage basin shown In Figure 2.31B. summarizes many Chapter 2 concepts. This is an example of how climate change modulates the timing of an episode of retreat of a knickpoint created by displacements on the range-bounding Hope fault. Frequent vertical displacements on the range-bounding Hope fault at a basin-position coordinate of 0.71 (1.00 for watershed analyses upstream from the mountain front) are sufficiently strong perturbations to reduce continuity (Section 2.5.2) of mountain and piedmont reaches. The fractured greywacke sandstone of this 10 km2 watershed is so sheared and fractured that resistant ledges for prominent waterfalls generally are absent along a very narrow valley floor. Rates of knickpoint retreat and stream channel incision are strongly tied to unit stream power - the mean bed shear stress of Whipple (2004). Nearby smaller and larger watersheds display much different responses to similar Hope fault tectonic perturbations, even though the rock mass strength is similar. Stream power thresholds have to be crossed before significant knickpoint retreat can occur, and total annual stream power becomes progressively less with smaller basin-position coordinates of a knickpoint that is migrating upstream. Only a few floods in the broad spectrum of stream-discharge events exceed the threshold unit stream power required to incise the bedrock trunk stream channel of the small Right Branch drainage basin. The result is a long response time to the 9 ka Hope fault tectonic perturbation. Sources and rates of introduction of bedload from the hillslope are not variable in these brecciated, fractured greywackes. Such stream-channel abrasive tools are poised to be dumped into the stream channel in all parts of the watershed. The response of the Right Branch to incision is partly a function of being an elongate drainage basin with no major tributaries (Fig. 2.23). In contrast, the Main Branch of the Charwell River is circular with many abrupt increases in discharge with the addition of major tributary inputs. Knickpoint retreat may slow dramatically after reaching a major tributary junction. Bull and Knuepfer (1987) note "The waterfall . . . occurs just upstream of a junction with a major tributary where the stream also crosses a ridge of more massive sandstone." Thus, lower unit stream power and increased rock mass strength appear to have created knickpoints that have become cumulative in a short reach of a fluvial system. Linear trends on semi-logarithmic plots for reach B (Fig. 2.32) and for reach D (also see Fig. 2.25B) attest to attainment of stream channel equilibrium. Reach D flows on soft Cenozoic rocks for 2 km before joining the main branch of the Charwell Concepts for Studies 20 ■ 40 Q ^3 60 CD &0 \00 70 Si! .^.-i!^s? *jfc.sr?\JS *l>^.,^V? «y 5? j I ....... 100 10 20 Basin position coordinate Figure 2.32 Dimensionless semi-logarithmic profile of the Right Branch of the Charwell River, New Zealand. From Figure 4 of Bull and Knuepfer, 1957. River (basin-position coordinates of 0.71 to 1.00). It is presently at its base level of erosion and beveling a tectonic strath since 4 ka. Reach D has achieved type 1 dynamic equilibrium as described by the Figure 2.25 equation. Reach B is still actively downcut-ting, but plots as a straight line on this semi-log plot, even though the stream flows through steep, rugged landscape. The central section of reach B, between altitudes of 800 and 600 m, has an inclusive gradient-index of 250 (same as reach D). So, lacking straths, reach B is a nice example of type 2 dynamic equilibrium between basin-position coordinates of about 0.27 and 0.52. Convex reach C (0.52 to 0.71) is in disequilibrium because of 40 m of uplift along the Hope fault. Post-26 ka reaction time (-17 ky) to surface ruptures is abnormally long. The stream could not degrade into bedrock as long as latest Pleistocene climate-change induced stripping of the hillslope sediment reservoir kept the system strongly to the aggradational side of the threshold of critical power. Exhumation of the sub-alluvial fault scarp that started at about 9 ka initiated knickzone retreat, which has progressed only a few hundred meters upstream. Long-term rock uplift favors maintenance of type 2 dynamic equilibrium conditions in reach B. Response times may be so long that the next climate-change aggradation event may occur before the 40 m of cumulative Hope fault displacements, that occurred between 26 to 9 ka, extends upstream to a basin-position coordinate of 30. Lack of fill terrace remnants in reach B suggests that it was sufficiently steep to remain on the degradational side of the threshold of critical power. Continued stream-chan- of Rising Mountains 73 nel downcutting might reduce or eliminate convex reach C. Climate-change induced aggradation seems to occur only in reach D, a base-level rise that spreads upstream far enough to bury the Hope fault for the duration of an aggradation event. Headwater reach A (basin-position coordinates of 0.00 to -0.27) is persistently degrading and unable to achieve equilibrium because of low annual unit stream power relative to the rock mass strength of materials beneath the trunk stream channel. Surprisingly the Right Branch fluvial system is neither sensitive (long reaction time) nor efficient (long relaxation time) in its response to a large tectonic perturbation emanating from the mountain front. System adjustments here are strongly modulated by climatic and lithologic controls. Chapter 2 tectonic concepts should be applied in the context of watershed climatic and lithologic controls on geomorphic processes. This helps us better understand the significance of external factors such as late Quaternary global climate change, sea-level fluctuations, and vertical tectonic deformation. The base level of erosion is the reference datum for studies of tectonics and fluvial topography. The threshold of critical power separates degradation and aggradation modes of operation of fluvial systems. It is purposely defined as a multi-variable ratio to remind us not to overemphasize a single variable, such as streamflow gradient, when trying to comprehend fluvial-system behavior. Time lags of response help us focus on the frequencies and magnitudes of tectonic and climatic perturbations, their locations within a fluvial system, and the magnitudes and time spans of departures from equilibrium conditions that such perturbations usually cause. Bedrock uplift has a major influence on geomorphic processes and landscape evolution. Relief orographically controls precipitation and temperature, and defines potential energy of flowing water even where tectonic elevation of mountains ceased long ago. Increases of fluvial-landscape relief emanate from active geologic structures through the process of tectonically induced downcutting. Streams act as connecting links that transmit tectonic perturbations to upstream reaches. Active faults and folds separate fluvial reaches with vastly different processes and landforms. Degradation changes to piedmont aggradation where a stream crosses an active range-bounding fault. Let us explore how bedrock uplift affects mountain fronts in the next two chapters. Chapter 3 3.1 Introduction Mountain-front escarpments have caught the attention of humankind for centuries. Settlers of the American west viewed distant mountains as a change of scene and as an impending challenge. Geologists wonder if active faults and folds separate mountains from lowlands. Active geologic structures in topographic escarpments are zones of concentrated tectonic base-level fall for fluvial systems. One challenge is to discern which mountain fronts have fault zones that are sufficiently active to generate damaging earthquakes. Progressive urban encroachment onto mountainous escarpments occurs after gently sloping land is occupied or when homeowners seek impressive views from their residences. Residential construction on tectonically active escarpments, such as in Los Angeles, California, Salt Lake City, Utah, and Wellington, New Zealand, increases earthquake and landslide risks. Steep, high mountain fronts can be menacing but the next surface rupture may occur along the more subtle low fronts and scarps. In order to assess potential earthquake-related hazards the paleoseis-mologist needs to identify and date key tectonic land-forms and apply her or his knowledge of landscape evolution. Tectonic geomorphologists are faced with diverse questions when viewing a sea of suburban development that laps onto foothills of lofty mountains (Fig. 3.1). How old is the escarpment, and what are the past and present rates of uplift? How seismogenic are the pulses of mountain-building uplift? Has a steady-state balance been achieved between uplift and denudation of the mountain slopes? (Lave and Bur-bank, 2004), or is a model of continuously changing landscape more appropriate? Earth scientists, engineers, and planners benefit from geomorphic tectonic assessment of whether or not range-bounding fault zones are active or inactive. How long has it been since the most recent surface rupture, and when will the next one occur? What advice should be given to those seeking to bulldoze low piedmont fault scarps in order to build new housing subdivisions, or to those already admiring Image showing the proximity of the Los Angeles metropolitan area (lower-right coastal plain) of southern California to the imposing mountain front of the rugged San Gabriel Mountains which are rising >2 m/ky. Pacific Ocean in foreground. Mt. Baden Powell at the right side rises to 2,866 m. Shuttle Radar Topographic Mission perspective view with Landsat overlay; image PIA02779 courtesy of the Jet Propulsion Laboratory and NASA. 76 Chapter 3 their views from homes built on the crests of high fault scarps? Clearly there is a need to do more than merely describe the locations and types of faults present. With each passing year Quaternary earth scientists are better able to define the locations and magnitudes of future surface ruptures, and to estimate the rates of uplift along faults associated with low and high escarpments. The San Gabriel Mountains and the thrust faults along its south flank are associated with a bend in the strike-slip San Andreas fault. Right-lateral movements along this restraining bend cause local crustal shortening, so the thrust and strike-slip styles of faulting are intimately related. It is logical for pa-leoseismologists to ask "do synchronous surface ruptures of the San Andreas and thrust faults occur as a single mega-earthquake event?" Alternatively, thrust-fault earthquakes occur independently. The hazard implications for the Los Angeles metropolitan region are profound. The San Fernando (U.S. Geological Survey, 1971), Whittier Narrows (Hauksson et al., 1988), and Northridge (Hudnut et al., 1996) earthquakes demonstrate the seismically active nature of the mountain front and adjacent basin (Dolan et al., 1995). We can expect more damaging earthquakes Figure 3.1 Urban development encroaching onto a thrwet-faulted mountainous escarpment east of Cucamonqa Canyon, 5an Gabriel Mountains, southern California. Both the high mountains and the lower structural bench are being raised along thrust faults. Less obvious active faults rupture the urbanized piedmont alluvial fans. especially if the major range-bounding Sierra Madre-Cucamonga fault (Fig. 3.3) ruptures. Will 40-90 km of this fault rupture synchronously with the next surface rupture of the San Andreas fault? The result would be a Mw magnitude >8.0 earthquake. Mw is earthquake moment magnitude (Hanks and Kanamori, 1979). Thrust faults in the Elkhorn Hills (Sieh, 1978a), and perhaps elsewhere, ruptured during the 300 km surface rupture of the San Andreas fault in 1857. An appraisal of the Mw magnitude 8.3 Gobi-Altay, Mongolia 1957 earthquake (Bayarsayhan et al., 1996; Kurushin et al. 1997) serves as a useful prototype. It's surface-rupture length was 250 km. The spatial arrangement of thrust and strike-slip faults is remarkably similar to the Cucamonga and San Andreas faults. They conclude that the probability of such an event is speculative, but "the similarities are too great for the possibility of such an event to be ignored". Although the rapid contractional strain rate between the coast and the San Gabriel Mountains dictates big thrust-fault earthquakes (Dolan et al., 1995), it is rather unlikely that they will occur concurrently with the next San Andreas strike-slip earthquake (Hough, 1996). Mountain Fronts 77 Figure 3.2 Image showing the tectonic setting ofthe San Gabriel Mountains, which appear in the lower right part of the view as a lens of raised terrain caught between the San Andreas fault (prominent diagonal slash at right side) and the range-bounding thrust faults on the Los Angeles side of the mountain range. The Garlock fault at the top right of the view bounds the north side of a wedge of technically quiet terrain in the western Mojave Desert. Shuttle Radar Topographic Mission perspective view with Landsat overlay; image PIA03376 courtesy of the Jet Propulsion Laboratory and NASA. The purpose of Chapter 3 is to review several ways to assess the hazard potential of tectonically active mountain-front landscapes. Mountain fronts are created by diverse styles of faulting and folding, so the overall theme is landscapes that respond to tectonic base-level fall. I apply the conceptual models of Chapter 2 and use fluvial landforms to better resolve several specific problems. Stream channels, terraces, and faulted alluvial fans are used to determine which thrust faults are capable of producing the next earthquake, and to measure the true throw of normal-fault surface ruptures. Passive margin escarpments fall outside of the paleoseismology emphasis of this book. Erosion-in- Los Angeles Mountain front Major fault zone Stream I Watershed Figure 3.3 Location map for place names and illustrations pertaining to the San Gabriel Mountains. 110° 00' _I_ 78 Chap duced isostatic uplift of tectonically inactive escarpments is fascinating. Examples include Drakensburg in South Africa (Gilchrist et al., 1994; Brown et al., 2002; van der Beek, 2002), Blue ridge in the eastern United States (Spotila et al., 2004), and Western Gnats bordering the western side of the Deccan Plateau in India (Ramasamy, 1989). The term "mountain front" pertains to more than the topographic junction between the mountains and the adjacent piedmont. A mountain front is a topographic transition zone between mountains and plains. This landscape assemblage includes the escarpment, the streams that dissect it, and the adjacent piedmont landforms. Our discussion starts with the diagnostic land-forms of triangular facets, mountain-piedmont junctions, and piedmont forelands of an active mountain range bordering the Los Angeles metropolitan area in southern California (Figs. 3.2, 3.3). Triangular facets evolve during a million years of erosion and episodic uplift of resistant rocks of arid regions (Fig. 2.20A). Formation of mountain-piedmont junctions may be likened to a contest between the relative strengths of uplift along a range-bounding fault zone and fluvial dissection. Locations of active thrust faulting typically shift from mountains into adjacent basins. Piedmont forelands are newly raised and deformed blocks between the new and old thrust faults. These low piedmont scarps are easy to overlook, but may eventually rise to become impressive escarpments with the passage of geologic time. The Gurvan Bogd mountains of the Gobi Altay, Mongolia were formed by a system of strike-slip faults with a reverse component. The magnitude Mw-8.0 earthquake of 1957 has attracted paleoseis-mologists from around the world to study the marvelous scrunch tectonics of this remote arid region. Low ridges rise through broad piedmonts of >3,000 m high mountains and roughly parallel the older mountain fronts. Florensov and Solonenko (1963) used the term 'foreberg' for these hills created by the complexities of scrunch thrust faulting (Kurushin et al., 1997). Such folds, antithetic and synthetic faults, and elongated backtilted ridges result from the shortening component on a broad active intracontinental fault zone. They have a common function, which is to broaden the deforming zone by creating new structures that accommodate both strike-slip and scrunch shortening components of tectonic deformation. The landscape further suggests that these new structures evolve by lateral propagation, increase in 3 Figure 3.4 Scrunch tectonics of piedmont forelands and forebergs. Illustration and caption are from Figure 5 of Bayasgalan et al. (1999). A Cartoon of a transpressional "flower structure" adapted from Sylvester (1955). B. Cartoon of the internal deformation within a foreberg, based on observations at Gurvan Bulag, Mongolia. Note the flattening of the underlying thrust at very shallow depths, which Is probably responsible for the collapse of the thrust "nose" by normal faulting (Kurushin et al., 1997) and the left-stepping back-thrusts and right-stepping normal faults, which suggest a component of left-lateral slip. C. Cartoon showing migration of active faulting away from the main range front, leaving uplifted and dissected fans in the hanging wall of the new fault and older, abandoned faults and shear fabrics within the uplifting mountain range. Mountain Fronts 79 amplitude, and may eventually merge and form new fault zones of considerable length. The evolution described here is thus peculiar to strike-slip faults with a reverse component, and can form many of the features of the "flower structures" that are often described in such oblique-shortening zones (Fig. 3.4A). The interplay between the rates of sedimentation and erosion allows some elevated fans between the foreberg ridges and the mountains to be much less dissected than would otherwise be expected, because the rising foreberg is a base-level rise (Fig. 3.4B). Few arid-region streams have sufficient annual unit stream power to accomplish the tectoni-cally induced downcutting needed to cross the rising landform (Owen et al., 1997). The streams of the highly seasonal semiarid San Gabriel Mountains dissect such piedmont forelands easily. Range-bounding faults of the lofty mountain range become much less active as tectonic deformation is transferred to the newest outermost fault zone (Fig. 3.3C). Piedmont foreland and foreberg shapes result from changes in thrust-fault dips of the underlying thrust faults in the uppermost 200 m and also at depths or more than 1 km. Add features like fault-bend and fault-propagation folds and it is easy to see why each structural geologist devises a different tectonic scenario for a study region. I use the model of Ikeda and Yonekura (1979) and Ikeda (1983) for the San Gabriel Mountains, where characteristic suites of fluvial landforms document shifts in the locations of scrunch tectonics. Chapter 3 also explores normal-fault landscapes. I evaluate a conceptual model for segmented surface-rupture behavior of active faults, and then apply the fault segmentation model to a normal fault in Idaho. The best way to test the characteristic earthquake model is to make measurements of historical and prehistorical surface ruptures with sufficient precision to define surface rupture behavior in the boundaries between fault segments. Such responses of hills and streams to episodic surface ruptures are then used in Chapter 4 to identify active range-bounding faults and to discern regional patterns of uplift rates of mountain fronts. 3.2 Tectonically Active Escarpments Hills record long-term interactions between uplift and landscapes. Compared to streams, hills respond slowly to the cumulative effects of many small increments of uplift along active geologic structures. Long response times to uplift are due primarily to huge volumes of rock that have to be weathered into erod-ible-size fragments before tectonic landforms such as triangular facets can be created (Menges, 1987, 1990a, b). Mountainous topography is the consequence of fluvial erosion initiated by the first pulse of uplift. Mountains continue to evolve for millions of years after tectonic uplift has ceased. However, isostatic uplift (Fig. 1.4) continues in response to denudational unloading. The resulting landscape assemblages record the rates and magnitudes of rock uplift and concurrent fluvial erosion; both processes increase relief. The topic of mountainous escarpments is part of the much broader subject of hillslope development whose erosion is initiated by base-level fall. Many of the early papers are worth reading. Hill-slope processes and forms are reviewed by Young (1972), Carson and Kirkby (1972), and Cooke and Warren (1973); important papers about hillslopes include those by Gilbert (1877), Horton (1945), Strahler (1950, 1957), Schumm (1956), Leopold and Langbein (1962), Hack (1965), Abrahams (1994), and Anderson and Brooks (1996). Significant papers about the effects of uplift on mountainous escarpments include those by Davis (1903), Louderback (1904), Blackwelder (1934), Gilbert (1928), King (1942, 1968), and Wallace (1977, 1978). 3.2.1 Faceted Spur Ridges The splendid triangular facets of the Wasatch Range escarpment in north-central Utah have been a classic example of a tectonic landform since the time of William Morris Davis (1903). Blackwelder (1934), Hamblin (1976), and Wallace (1978) describe triangular facets as being fault planes that have been modified by erosion, an explanation that seems appropriate for mountains bounded by normal faults. Triangular facets result from base-level fall, and occur in a variety of tectonic settings. Erosion of facets at the truncated ends of spur ridges may be associated with normal faults (Fig. 3.5A), anticlines (Fig. 3.5B), thrust faults (Fig. 3.5C), and even along escarpments formed by erosional base-level fall (Fig. 3.5D). The overall similarity of the facets shown in Figure 3.5 is suggestive of a more general relation than erosional notching of normal-fault planes. The faceted ends of the spur ridges are steep hillslopes that reflect recent cumulative range-front uplift. Sharp-crested spur ridges divide an escarp- 80 Chapter 3 igure 3.5 Triangular facets of different tectonic environments. A. Spur ridges truncated by a normal fault on the east side of the Toiyabe Range, central Nevada. merit into drainage basins. Each spur ridge terminates at the range front in a characteristic triangular outline. Initial development of faceted spurs is similar, even where uplift is along a reverse fault that dips into the mountains (Fig. 3.5C). An early stage consists of crudely planar 20° to 40° hillslopes. Keller and Pinter (2002, p. 10) nicely depict the key topographic and stratigraphic features for many normal faults (Fig. 3.6). Uplift of range fronts in west-central Nevada proceeds as 1 to 3 m surface ruptures every 5 to 10 ky (Wallace, 1978). These may seem small and infrequent, but the resulting topography is spectacular and the landforms are truly indicative of the relative rates of rock uplift and erosion. Wallace's block diagrams depict the evolution of a fault-generated mountainous escarpment (Fig. 3.7) that reflects the long-term (>10 My) interactions between uplift and denudation. Initial faulting (stage A) creates a linear scarp crest that migrates away from the range boundary. A range crest is created by merging of scarp crests that migrate from the range-bounding faults on opposite sides of the rising block. Valley floors notched into the rising block (stage B) are zones of most rapid tectonically induced downcut-ting. The intervening spur ridges and the range crest are gently sloping. These landscape elements have the slowest rates of tectonically induced degradation (stages C and D). The mountain-piedmont junction continues to be straight and the valley floors narrow during continuing rapid uplift, even where rocks are Figure 3.5 Triangular facets of different tectonic environments. F3. Triangular facets on the north side of the Wheeler Ridge anticline, south edge of the San Joaquin Valley, California. The topographic benches may be the result of mass movement processes (Bielecki and Mueller, 2002). soft. An aerial view of the Tobin Range (Fig. 3.8) reveals the simplicity of the terrain from which Wallace developed his concepts of landscape evolution. Rugged faceted spur ridges owe their substantial heights to sustained tectonic base-level fall at the range-front landscape boundary, and to the profound initial difference between valley-floor and ridgecrest rates of denudation. Increase of hillslope steepness and relief also increases the rate of hillslope erosion. Landslide processes become more important as the ever steeper valley side slopes become progressively more unstable (Pain and Bowler, 1973; Pearce and Watson, 1986; Keefer, 1994; Hovius et al., 2000; Dadson et al., 2003, 2004). The style of landscape evolution reverses after cessation of rapid uplift, but hillslopes never attain a steady-state condition. Hillslope denudation rates exceed uplift rates, so the mountain-piedmont junction becomes sinuous as it retreats from the position of the range-bounding fault to create a pediment (stage E of Fig. 3.7). Non-steady state denudation brings ridgecrests closer to the valley floors, which remain at type 1 dynamic equilibrium with valley floors that become progressively wider. Dissection of range front triangular facets proceeds independently of the stream-channel down-cutting of the adjacent trunk valley floors. Consequent drainage nets on young triangular facets initially consist of parallel rills. Capture of flow from adjacent rills occurs. Small elongate watersheds form on these planar surfaces and become more circu- Mountain Fronts Figure 3.5 Triangular facets of different tectonic environments. CI. Aerial view of a set of triangular facets that terminates the spur ridge of a thrust-faulted mountain front of the 'dan Gabriel Mountains, Southern California. C2. Facet-dissection stages as described in Table 3.1. Younger facets 1 and 2 are nested Inside older higher stage 3 and stage 4 facets that are deeply incised by small stream channels. DF3, basins to catch debris swept off recently burned steep hillslopes during winter storms. lar through the processes first described by Horton (1945). Development of progressively larger drainage nets at the ends of spur ridges concentrates available stream power, promoting efficient erosional destruction of the triangular facets. Planar facets with numerous closely spaced parallel rills (Fig. 2.20A) are eventually transformed into a deeply dissected ridge-and-ravine topography (Fig. 2.20B) in which the triangular shape of the facet is less obvious (Fig. 3.9). Erosional dissection of faceted spurs can be described as six stages (Table 3.1). The stages are easy to distinguish in the field, on aerial photographs, or on detailed topographic maps. The time needed to achieve a given stage is a function of two compensating processes. Uplift increases facet height, and fluvial erosion deepens valleys. With the passage of time, faceted spurs adjacent to an active fault become higher and more dissected. The lowest, most recently created, part of a facet is less dissected (Fig. 3.5C) because it has been exposed to erosion for time. These qualitative descriptions of facet dissection are used to help define mountain front tectonic activity classes in Chapter 4. Figure 3.5 Triangular facets of different tectonic environments. D. Triangular facets along the edge of a fanhead embaymerrt at Cucamonga Canyon, San Gabriel Mountains, southern California. These nontectonlc facets were created by lateral erosion induced base-level fall caused by streamflow. 82 Chapter 3 V-shaped canyon and remnants of stream terraces record tectonically Induced downcuttlnq Coarse-grained alluvial-fan deposits merge with fine-grained basin fill Primary and secondary faults and scarps Figure 3.6 Diagrammatic sketch of the topographic expression of an active normal-fault system. Uplift on a range-bounding normal fault creates a base-level fall that causes deep valleys to be eroded In the mountain block. This tectonic displacement favors accumulation of thick alluvial-fan deposits downstream from the normal-fault zone. From Keller and Pinter (2002, Fig. 1.7C). Figure 3.7 Block diagrams showing the sequential development of a fault-generated mountainous escarpment. A. Initial faulting creates a linear scarp. B. 'bcarp crest migrates away from the rising range boundary to form range crest. C. Valleys are notched Into the rising block; their floors are the locations of rapid tectonically induced downcutting by streams. The crests of spur ridges are the locations of slow tectonically induced degradation. D. episodic displacement along the range-bounding fault maintains a steep, straight mountain-piedmont junction. Main and spur ridge divides continue to rise faster than degradation can lower them. E. The mountain-piedmont junction becomes sinuous and the valley floors become wider after cessation of uplift. Relief becomes less as degradation lowers the ridgecrests. Figure 4 of Wallace (1977). Mountain Fronts 83 Figure 3.5 View of the west side of the normal-faulted terrain of the Tobin Range, west-central Nevada showing the basic landscape elements of a tectonically rising landscape as described by Wallace (1977). These include the range crest, spur ridges extending to triangular facets at a straight range front, and a deep valley with a narrow valley-floor width. The irregular dark line at the mountain front is the surface rupture of the 1915 earthquake. The planimetric trace of this topographic transition between mountains and piedmont is useful for assessing whether or not the mountain front coincides with an active range-bounding fault zone. The sinuosity of the mountain-piedmont junction represents a balance between 1) the tendency of uplift to maintain a sinuosity as low as that of the range-bounding fault or fold, and 2) the tendency of streams to erode an irregular junction between the mountains and the plains. Straight mountain-piedmont junctions generally indicate the presence of an active fault. Embayed, pedimented mountain-piedmont junctions suggest tectonic quiescence. Downstream Facet class Erosional landforms 1 Flanar surface with only rills. Includes scarps that have yet to be carved Into facets by streams flowing across the scarp. 2 Y\anar surface with shallow valleys extending a short distance into the facet. 3 Valleys extend more than 0.7 the horizontal distance between the base and top of the facet. 4 Deep valleys extend more than 0.7 the horizontal distance. 5 Greatly dissected but the general form is still obvious. 6 So dissected that the general form of a facet is not obvious. 7 Triangular facets are not present because they have been removed by erosion, or they never existed. Table 3.1 Stages of dissection of triangular facets. 3.2.2 Mountain—piedmont junctions Transitions between mountainous escarpments and adjacent basins typically are abrupt. Steep hills give way to gentle piedmont slopes in both tectonically active and tectonically inactive landscapes. Piedmonts may consist of either the depositional environment of coalescing alluvial fans, or the erosional environment of pedimented terrain. Fans and pediments may be smooth in the arid realm. They tend to be dissected and less obvious in humid regions where floodplains are a common piedmont landform and forests may cloak subtle features of the landscape. 84 Chap Figure 3.9 Aerial view of triangular facets near La Canada, 5an Gabriel Mountains, Southern California. Stage 4 of Table 3.1. increase in stream power maximizes the potential for downcutting and lateral erosion where streams leave the mountains. The result is a highly sinuous mountain-piedmont junction, even in homogenous rocks, but only under tectonically inactive conditions. Small structures, such as joints, foliation, and bedding planes, also influence sinuosity of mountain-piedmont junctions. Tectonically inactive mountain fronts with structures that parallel the range front may have an anomalously straight mountain-piedmont junction and well defined triangular facets. Sinuosity of the mountain-piedmont junction also is a function of the width of a mountain range (Parsons and Abrahams, 1984; Mayer, 1986). Wide mountain ranges have large drainage basins that are more likely to have sufficient stream power to quickly attain the base level of erosion and create pediment embayments after uplift has ceased. Range width decreases with erosional retreat of range fronts. Drainage-basin size, and stream power, become less. Mountain-piedmont junction sinuosity may become lower as the mountain landscapes are progressively replaced by the beveled bedrock of pediments. The constraint of drainage-basin size on sinuosity of the mountain-piedmont junction is illustrated by the McCoy Mountains of southeastern California. Geophysical studies by Rotstein et al. (1976) suggest that the faults initially bounding the McCoy Mountains structural block now are 1 to 2 km from the present range front. The mountains are only half of their original width. Average drainage-basin length 3 Figure 3.10 Tectonically inactive mountain fronts of the McCoy Mountains in southeastern California. A. Map of the mountain-piedmont junction and watershed divides. L and H are low and high sinuosity mountain fronts. 1-5 are study watersheds. for fronts at L (Fig. 3.1 OA) is only about 1 km and the sinuosity of the mountain-piedmont junction is moderate. Mountain fronts are highly sinuous at H where drainage basins are twice as large. Granitic rocks weather slowly and streamflow is ephemeral in this arid, hyperthermic, moderately seasonal climate (Table 2.1). Unit stream power is large during flash floods during infrequent incursions of tropical depressions into the southeastern Mojave Desert or during some wintertime cyclonic storms. The stream channels of the McCoy Mountains degrade by abrasion and plucking during floods, but long-term weathering of the granitic rocks into small particles plays a much larger role than in humid settings such as the Potomac River (Fig. 2.29). McCoy Mountain Fronts 85 Figure 3.10 Tectonically inactive mountain fronts of the McCoy Mountains, California. F3. Topographic map of watershed 4. Contour Interval is 50 m. Gray is mountain bedrock. Pattern is piedmont alluvium whose extent defines the present mountain-piedmont junction. Heavy dashed line is present trace of ends of spur ridges. J ( near 3) is stream junction. 1, 2, 3, 4 are defined In Figure 3.10C. fluvial systems have had >10 My to greatly modify this landscape. There has been ample time for erosion of bedrock stream channels to achieve equilibrium relationships, in marked contrast to a rapidly rising mountain front such as the Charwell River (Figs. 2.16, 2.24-2.28). Planimetric and longitudinal-profile aspects of fluvial-system equilibrium in a tectonically inactive watershed are shown in Figures 3.10 B, C. The 2 km wide strath at the mountain front is a pediment embayment indicative of ample time and stream power to achieve prolonged type 1 dynamic equilibrium. Diminishing watershed area resulting from 2 km of mountain front retreat to its present position is offset by the prolonged time span for the ephemeral stream to do this work. Pediments are not a special landform when viewed in terms of processes. They form where stream(s) at the base level of erosion bevel straths that coalesce to form the beveled bed- rock piedmont landform that we call a pediment. It takes millions of years for this process to remove spur ridges between adjacent drainage basins. The smooth slightly concave longitudinal profile in the reach between locations 3 and 4 records attainment of type 1 dynamic equilibrium conditions. The upstream narrowing of the pediment embayment (Fig. 3.1 OB) reflects the importance of concomitant spatial decrease of unit stream power and the relative increase of the importance of rock mass strength of materials beneath the stream channel. Stream power prevails at the mountain front but eventually a threshold is crossed where unit stream power is insufficient to overcome rock mass strength. So the upstream end of the pediment embayment coincides with the junction (J) of the two largest streams in this drainage basin. Stream power upstream from this junction is insufficient to bevel broad valley floors in this rock type. 700- £ 600-<ü Z5 ^ 400- 200- i- -r 2 3 Distance, km 4 Figure 3.10 Tectonically inactive mountain fronts of the McCoy Mountains. C. Longitudinal profile of trunk stream channel of watershed 4 of figure 3.10A. 1 is watershed divide. 1-2 is disequilbrium reach. 2-3 is type 2 dynamic equilibrium reach. 3-4 is type 1 dynamic equilibrium reach. Vertical exaggeration is 4.0. 86 Chapter 3 Valley-floor widening of the strath extends only a short distance further to location 3 (Figs. 3.1 OB and 3.IOC). The longitudinal profile becomes steeper upstream from 3, but still has a form indicative of attainment of equilibrium. But, in contrast with the downstream reach (inclusive gradient-index is 129), the valley floor is narrow (inclusive gradient-index is 200). So, reach 2-3 is best regarded as being type 2 dynamic equilibrium. Disequilibrium conditions prevail in the headwater's reach, 1-2. Unit stream power is minis-cule relative to rock mass strength. Rates of stream-channel downcutting are so slow that disequilibrium prevails even after 10 My of tectonic quiescence. The watershed area needed to generate sufficient stream power to achieve type 1 dynamic equilibrium in the McCoy Mountains is partly a function of drainage net configuration (Fig. 3.10D). Headwaters reaches, such as 1-2 or 1-3, occupy only a small portion of total watershed area in large drainage basins, so equilibrium conditions are achieved at a basin-position coordinate (Section 2.8) of only 0.25. An example is watershed 1 of Figure 3.10A. The location at which type 1 dynamic equilibrium is attained is farther downstream where headwater's source areas are smaller. 3.2.3 Piedmont Forelands Normal, reverse, and strike-slip faults associated with tectonically active landscapes may be classed geomor- 6 to £4 to m * 2 TO 1 ' Y = 7.57 - 11.65 X - Kz = .605 - ♦ i 0.2 0.6 0.3 OA 0.5 Basin-position coordinate Figure 3.10 Tectonically inactive mountain fronts of the McCoy Mountains. D. Regression of drainage-basin areas upstream from type 1 dynamic equilibrium stream channels and basin-position coordinates. phically as internal, bounding, or external. Internal faults rupture mountain bedrock. Internal strike-slip and thrust faults may have associated normal faults. Prolonged uplift along range bounding faults can create escarpments rising above adjacent plains. External faults occur in the depositional reaches of fluvial systems, and may be incipient mountain fronts. They include: 1) Piedmont fault scarps created by the newest splays of a thrust-fault system, 2) Strike-slip faults crossing the basins between mountain ranges, or 3) Inactive former range-bounding faults that no longer coincide with the mountain-piedmont junction because erosion has caused the mountain front to retreat after cessation of uplift. This section describes landscapes of a compressional tectonic setting that have all three geomorphic fault classes. The topography associated with active thrust faults (Figs. 3.4, 3.11, 3.12) is much different than that for normal faults (Fig. 3.6). Uplift along a range-bounding thrust fault can create an imposing escarpment (Fig. 3.1), but primary surface-rupture locations shift when new splays of a propagating thrust-fault zone encroach into the adjacent basin (Yielding et al., 1981). The former range-bounding fault becomes an internal mountain front that is less tectonically active, or becomes inactive. The new fault at the edge of rapidly rising hills is the latest in a series of range-bounding faults. Tectonic deformation ruptures and folds this piedmont foreland, which is a bedrock block, capped by remnants of piedmont alluvium. New streams dissect the former depositional slopes as a consequence of piedmont terrain being incorporated into ever-broadening mountain range. Paleoseismology of piedmont forelands is challenging to the tectonic geomorphologist. Each structural block has a different tectonic history, and recently created external faults pose a deceptively high earthquake hazard that is largely hidden from view. Locations and magnitudes of earthquakes have changed during the late Quaternary and blocks between thrust faults are deformed by folding and antithetic faulting. After an earthquake, the effects of synchronous vertical displacements on several fault zones may overlap where they migrate upstream as knickpoints (Section 2.5.1). Each perturbation diminishes in magnitude as it moves upstream. A good example of active tectonics of suites of thrust faults is the area of the San Fernando earth- Mountain Fronts 87 Figure 3.11 Diagrammatic sketch of the topographic expression of an active thrust-fault system. Prolonged uplift on an internal fault has created a high scarp. Uplift may be rapid on the range-bounding fault but insufficient time has elapsed for this scarp to become an Impressive landscape element. From seller and Vinter (2002, Flg. 1.7B). Internal thin alluvial-fan deposits quake of 1971. Geologists have made extensive and diverse studies in southern California (Crook et al., 1987; Southern California Earthquake Center Group C, 2001 ). However, the San Fernando fault zone was not included on the detailed geologic map by Oakeshott (1958). This range-bounding fault became obvious after the Mw 6.4 earthquake ruptured 12 km of a range-bounding piedmont foreland (Fig. 3.12). The Mw 6.7 Northridge earthquake of 1994 occurred on a nearby blind (buried and hidden from view) thrust fault where rupture did not reach the surface. Although moderate in size both earthquakes caused major damage and loss of life. Landscapes associated with thrust-fault migration have strikingly similar diagnostic features. The piedmont foreland bounded by the San Fernando fault zone (Fig. 3.12) resembles those of the Mahiru and Misaka Ranges of Japan (Figs. 3.13A, B). A 2 to 5 km wide piedmont foreland is present between two thrust-faulted mountain fronts. The front created by displacement on an external fault, although low, has been the more active of the two during the late Quaternary. Tectonic landforms include stage 1 and 2 triangular facets (Figure 3.12) and a straight mountain front (not embayed by erosion). Only rapid recent uplift can maintain such tectonic Figure 3.12 Aerial view of the piedmont foreland near Little Tujunga Canyon, San Gabriel Mountains, southern California. The Sunland fault zone bounds the internal front, and the Lakeview fault zone bounds the range front in the foreground; it ruptured in 1971. Chapter 3 signatures in a landscape underlain by soft sandstone, mudstone, and weakly cemented conglomerate. The smooth surface areas of the piedmont foreland are capped by stream gravel, which was laid down when it was the depositional part of the fluvial system. Active tectonism has converted a former depositional reach of the fluvial system into hilly terrain. The Mahiru Range of northeastern Japan has excellent examples of piedmont forelands created by thrust-fault migration (Fig. 3.13A). The Kawaguchi fault bounds an impressive internal front and separates Miocene volcanic rocks, mudstone, and sandstone from the Pliocene sandstone and mudstone of the Senya formation (Ikeda, 1983). The younger Senya fault is 3 km basinward of the Kawaguchi fault and bounds the 120 m high Senya Hills. It was the site of the 1896 Mw magnitude 7.2 Rikuu earthquake. Ikeda did not find evidence for an 1896 surface rupture along the internal Kawaguchi fault. Despite the rugged nature of the Mahiru Range, erosional embayments extend far into valleys whose broad floors suggest attainment of the base level of erosion. Bear in mind that tectonic landforms in soft rocks erode quickly in a humid region (Fig. 2.20B). Mahiru Range streams may degrade fast, and/or Ho-locene uplift rates have been negligible or slow. The strikingly similar piedmont foreland of the Sone Hills (Figs. 3.13B, C) bordering the Misaka Range is also capped by several remnants of a thin blanket of fluvial deposits. Ikedas model (Fig. 3.14) describes encroachment into piedmonts by active thrust faults (Fig. 3.15). Although the older internal front has more topographic relief, the range-bounding fault is presently more active. Many fault splays and folds are present in the hanging wall of either thrust fault. Earth deformation may tilt part of the piedmont foreland back toward the main range. Backtilting, synclinal folding, and antithetic faulting may result in local subsidence or an uplift rate that is slower than that of the tectonic bulge. Thrust-fault surface ruptures of the 1896 Rikuu event in northeastern Japan and the 1971 San Fernando event in southern California have the same geomorphic and tectonic features. Figure 3.14 describes this common tectonic style by assessing relative rates of uplift for the "basin fill", "piedmont foreland", and "mountain range" structural blocks. Only local components of surface uplift are evaluated here - faulting, folding, erosion, and deposition. Total tectonic uplift (equation 1.2) would include broader wavelength styles of uplift. The sum of local and more regional components of uplift probably would result in all three structural blocks of Figure 3.14 ris- Figure 3.13 Characteristics of thrust faulted mountain fronts in Japan. A. Contour map (interval 10 m) of the 5enya Hills piedmont foreland of the Mahiru Range, in northern Honshu Island draining to the Sea of Japan. The Kawaguchi thrust fault (dashed line) bounds the internal front. The dotted line along the range-bounding Senya thrust fault indicates the surface rupture of the 1S96 Rikuu earthquake. From Figure 4 of Ikeda, 1953. Mountain Fronts 89 Figure 3.13 Characteristics of thrust faulted mountain fronts In japan. F3. Contour map {Interval 20 m) of the 'done Hills piedmont foreland of the Misaka Range. Active faults are shown by lines, active anticlinal axes with small dotted lines, and inactive Internal fault zone with large dotted lines. Arrows show terrace-tread tilt directions. Figure 7 of Ikeda (1953). ing during the late Pleistocene and at the present. The present internal fault was the range-bounding fault of a rugged, fast-rising mountain range during the late Pleistocene. Fractured ground along the trace of this fault caused by the 1971 San Fernando earthquake suggests that it has not become totally inactive, but much of the slip now occurs on the splay of the fault system that terminates as the new range-bounding fault. The present rate of rock uplift may be less than the surface denudation rate, hence the interpretation of slight net local surface subsidence of the "mountain range" block at present. Erosional processes also reduce the present rate of surface uplift of the tectonic bulge. Note the volume of material eroded below the former piedmont surface during the past -0.4 My (Figure 3.12). Present basin-fill surface uplift at A is the sum of rock uplift due to tectonic scrunching Mudflow deposit Internal SO ka pumice faultVff ' ■Alluvial-fan gravel "--^vl^^ Pleistocene Sone group 0 2 3 km Figure 3.13 Characteristics of thrust faulted mountain fronts In Japan. C. Geologic cross section A-A' of Figure 3.13F3 across the done Hills showing tectonic setting of typical piedmont foreland. From Figure of Ikeda (1953). processes and aggradation of alluvial-fan gravels, as was the similar tectonic setting at A' before inception of the range-bounding fault. The model used here has tectonic basin subsidence being a bit more than surface uplift caused by alluvial-fan deposition. Ikedas model nicely explains the general topography of the Little Tujunga Canyon landscape. Many structural complications surely are present. Sharp (1975) notes that the fault dip is steeper than 20°. Yeats et al. (1997) summarize evidence for lateral components of slip. Correspondence with Bob Yeats points out the likelihood of bedding-plane faults, and Hiro Tsutsumi of the Kyoto Institute of Geophysics considers that the 1971 fault may be a well-expressed flexural-slip fault on the side of the Merrick syncline (Tsutsumi and Yeats, 1999; Tsutsumi et al., 2001). Structural geologists have yet to connect the 1971 fault with a large-displacement fault at depth. I proceed with a geomorphic appraisal using Ikedas modeling as one example, realizing that improved structural models will be forthcoming. Many earth scientists have studied the geometry of the faults and the displacements associated with the 1971 San Fernando earthquake and the more recent adjacent Northridge earthquake of 1994. The Figure 3.14 model can be evaluated with geophysical data (Allen et al., 1971; Sharp, 1975; Ikeda and Yonekura, 1979; Ikeda, 1983), by crustal-movement surveys (Castle et al., 1975; Savage et al., 1986), and by geologic information (Kamb et al., 1971; Proctor et al., 1972; U.S. Geological Survey, 1996). Ikedas model illustrates some likely complexities of Quaternary landscape change associated with multiple thrust faults, each having a different 90 Chapter 3 Figure 3.14 Tectonic elements of a piedmont foreland created by Quaternary migration of thrust faults. Adapted from Figure 1 of Ikeda (1953). Spatial and temporal variations of relative magnitudes of local uplift or subsidence are Indicated by the sizes of the gray arrows. A and A' are where aqqradatlxon is part of the surface uplift. Basin fill Piedmont foreland Mountain range Local rock uplift Late Pleistocene Present Local surface uplift Late Pleistocene Present t t t t t na cti ve Tectonic Active range- bul$e Antithetic 'Vernal bounding faul' by 1 to 4 km Alluvial-fan deposits Faulted and folded fan deposits t t t Ancient basement rocks displacement history. The location of the 1971 hy-pocenter and its 52° dip dictate a decrease in the angle of faulting toward the surface. In Ikedas model the dip abruptly decreases to about 13° at a depth of about 2.5 km (Fig. 3.15A). More than one steeply dipping fault may have been active in 1971. Heaton and Helmberger s model (1979) has two hypocen-ters: one at 13 km depth on a steeply dipping fault, and a second hypocenter on a more gently dipping range-bounding fault. Mori et al. (1995) show that the 1971 and 1994 events overlap. Figure 3.15B suggests that the internal, as well as the range-bounding front, was uplifted in 1971. In the Figure 3.15A model, a short 36° dipping fault segment was introduced to improve the fit between observed and modeled surface displacements. In the model shown, Ikeda assigns a uniform slip of 5.5 m over the two deeper fault segments. A variety of surface displacement patterns are depicted in Figure 3.15B, where 6 is the dip of the more gently dipping fault segment as a thrust fault approaches the surface. Magnitudes of modeled local subsidence increase with decreasing dip of the shallow fault segment. Ikeda concludes that an abrupt decrease in thrust-fault dip results in backtilting of the range-bounding front, and in continued uplift of the internal front where surface rupture is no longer obvious. The pronounced tectonic buckling associated with the range-bounding fault requires a large horizontal component of slip on a gently dipping fault that abruptly terminates in a set of secondary fault splays (36° in Fig. 3.15A). Once again, tectonic scrunching is an important component of rock uplift. Such spatial variations in tectonic deformation should profoundly affect consequences on the behavior of streams flowing across an active thrust fault and fold belt. For example, subsidence of the Merrick Syncline is only relative between distances of 1 and 4 km in Figure 3.15B. Streams in this reach should reflect the effects of slower uplift (Fig. 3.14) if Ikedas model is correct. Landscape assemblages test such models. Mountain Fronts 91 Lakeview (1971) \ Hypocenter Figure 3.15 Vertical displacements and fault geometry of the San Gabriel Mountains. Line of section A-A' parallels and is 2 km west of Little Tujunga Canyon. A. Cross section and calculations of theoretical vertical displacements compared with the observed 1971 displacements. Fault geometry is constrained by the 13 km depth and 52° dip of the focal mechanism of the main shock of the 1971 San Fernando earthquake. The curve is the best of three models but suggests true net subsidence between distances of 6.5 and 10 km. Note the large vertical displacement but minimal subsidence at the front that ruptured In 1971. From Ikeda (1953, Fig. 14). A tectonic geomorphologist would predict episodic attainment of the base level of erosion by looking for the presence of strath terraces. These reference surfaces would be parallel where passing through a reach with an inactive thrust fault. Strath terraces would diverge as a stream approaches an active thrust fault and then converge downstream from the tectonic perturbation, ending abruptly at the range-bounding fault zone. A broad alluvial valley floor where a stream is at the base level of erosion (Figure 2.26) might look much the same as an aggrading valley floor to most casual observers. The tectonic implications of these equilibrium and aggrading modes of stream operation are quite different. Seismic refraction surveys were used to assess stream-channel alluvial thicknesses to determine which thrust faults are active, and if the Merrick Syncline reach is truly subsiding relative to the base level of erosion reference surface. Streams that cross a belt of active thrust faults flow through alternating embayment reaches and gorge reaches. These diagnostic landform assemblages are the result of spatial variations in the rate of rock uplift. Rapid local tectonically induced downcut-ting creates a rugged gorge that ends abruptly after a stream crosses splays of an active zone of thrust faulting. Stream terraces may not form in gorge reaches f-,0.6 e = 45° 4 6 Free surface 8 km Fault Geometry Figure 3.15 Vertical displacements and fault geometry of the San Gabriel Mountains. F3. The effects of bending of a thrust fault plane on rock uplift, Au. Uplift of the interna I front, formation of a narrow tectonic bulge at the range-bounding fault, and backtilting and relative local subsidence of the piedmont foreland occur simultaneously where the dip of the shallowfault plane is 20° or less. From Ikeda (1953, Fig. 17). 92 Chapter 3 if uplift rates are sufficiently rapid, relative to stream-channel downcutting rates, to preclude attainment of type 1 dynamic equilibrium (Section 2.4.3). Abrupt widening of the active channel and the valley floor characterizes embayment reaches, which begin immediately downstream from the most downstream splay of a thrust-fault zone. Embayment reaches gradually narrow downstream and assume the characteristics of gorge reaches. The wide valley floors of embayment reaches may appear to be sites of thick alluvium. The presence of late Pleistocene and Holocene gravel strata would confirm a model of backtilting, synclinal folding, and local downwarping associated with antithetic faults that characterize part of the piedmont foreland tectonic domain. Conversely, the presence of strath terraces would attest to long-term net uplift (Section 2.6.1) of embayment reaches (Fig. 3.16). Amounts of net uplift relative to the reference datum (base level of erosion) are reduced by local tectonic downwarping in embayment reaches (Fig. 3.14). Thicknesses of stream-channel alluvium are functions of rates of stream-channel downcutting and magnitudes of local base-level change created by active thrust faulting. Local faulting of stream-beds crossing fold-and-thrust belts results in variable thicknesses of valley-floor alluvium. Streams don't remain at their base level of erosion after a surface rupture event on an active thrust fault. Each event induces aggradation downstream and degradation upstream from the fault (Fig. 3.17). A pocket of streambed gravel downstream from a thrust fault is anomalous and indicates a recent surface-rupture event. Tectonically induced downcutting will eventually eliminate anomalous thicknesses of gravel. Duration of such local deposition in a reach characterized by long-term degradation is a function of the magnitude and frequency of tectonic displacements, resistance of streambed bedrock to abrasion and plucking, and the annual tractive force of the stream. Variations in thicknesses of streambed gravel were assessed by shallow seismic refraction surveys for two streams. One survey was along Little Tujunga Canyon (Fig. 3.18). Gneissic and plutonic rocks underlie the headwater's half of the 49 km2 watershed and soft marine sediments are present in the downstream half. The headwater's reaches have numerous exposures of bedrock with intervening patches of thin gravel. Such characteristics are typical of disequilib- Figure 3.16 Late Pleistocene strath terrace in the embayment reach of Little Tujunga Canyon. The Internal mountain front rises in the background as a series of structural and fluvial benches. rium reaches that prevail in mountainous streams and provide little information about recent surface ruptures. Our basic reference surface - the base level of erosion is missing in such reaches. Non — geomor-phic studies indicate that the oblique-dextral San Gabriel fault zone may be still active (Lee and Schwarcz, 1996). Streambed gravels are only 1 m thick in the channel just upstream from the internal Sunland fault and increase almost threefold to about 2.7 m immediately downstream from the fault trace. Gravel thicknesses remain the same through the Merrick syncline reach but thin to 0.8 to 1.5 m as the stream approaches the gorge reach upstream from the range-bounding Lakeview fault. Gravel thickness then increases abruptly to 19 m only 0.1 km downstream from the fault zone (Fig. 3.18). Thus the most obvious increase in thickness of streambed gravels along Little Tujunga Canyon is associated with the most active fault zone. Sinestral-reverse oblique slip of the land surface of as much as 2 m occurred during the Mw magnitude 6.4 San Fernando earthquake of 1971 (Barrows et al., 1973), only 6 years before our survey of streambed gravel thicknesses. This historic Mountain Fronts 93 Figure 3.17 Hypothetical distribution of deposits beneath a stream flowing across an active thrust fault. A. Longitudinal profile of stream that approximates a condition of type 2 dynamic equilibrium. The thin blanket of gravel is easily entrained by large flow events, which have beveled the soft bedrock substrate and created a strath since the most recent surface rupture. F3. Thrust faulting raises the reach upstream from the fault and tilts it upstream. Scour removes the gravel. Collapse of the scarp creates colluvium downstream from the fault and the tectonic base-level fall creates space for deposition of streambed gravels. C. Valley-floor degradation continues during an interval of no additional faulting. Part of the gravel adjacent to the fault is removed and deposited further downstream. More bedrock is exposed upstream from the fault as the stream incises to deepen a bedrock narrows reach, u is uplift, cd is stream-channel downcutting, and t is time. The uplift rate-channel downcuttlng rate relation is the reverse of A. event may be partly responsible for the small stream-bed gravel anomaly downstream from the internal Sunland fault. These several variations in streambed gravel thickness might be only random variations about mean gravel thickness of covering a strath in a tec-tonically inactive fluvial system. They can also be interpreted as varying systematically with each geologic structure. If so, they indicate continuing uplift along the Sunland fault but at an order of magnitude slower rate than for the range-bounding Lakeview fault. Little deformation appears to have occurred recently in the Merrick syncline reach. Sustained synclinal folding would have depressed the streambed longitudinal profile below the base level of erosion and thick gravels would then characterize this reach. Instead, rates of long-term stream-channel downcutting exceed rates of tectonic downwarping. Total watershed relief upstream from the Lakeview fault zone has increased about 102 m. The stream of Little Tujunga Canyon upstream from the mountain front is large enough to be able to repeatedly downcut to its base level of erosion. Rapid vertical tectonic displacement has created an anomalously steep reach, especially considering the soft nature of the underlying Pliocene marine mudstones. Strath heights above the active channel record tectonic uplift. As much as 30 m separates some straths. This suggests that: 1) surface ruptures were so frequent that the base level of erosion was rarely attained, 2) the strath record is incomplete because erosion has destroyed some straths or has made them difficult to locate, or 3) modulation of times of stream-channel downcutting by climatic conditions (Section 2.6.2) may have permitted only occasional return to the base level of erosion. 94 600 - Strands of the San Gabriel Fault 500 - < 400 Kange-bounding Lakeview fault 971 surface ruptures Distance, km Figure 3.15 Variations in thickness of streambed gravel along Little Tujunga Canyon where it crosses active and inactive thrust faults. Subsurface faults and geology are diagrammatic, Chapter 3 location of maximum tectonic deformation gradually, but in an accelerating manner, from the internal fault zone to the range-bounding fault over a time span of 0.3 My. The exceptionally rapid Holocene uplift rates may not represent the true long-term increase of uplift rates. First, recent stream-channel downcutting might be anomalous if only a response to a recent temporal cluster of surface ruptures. It helps to know how the most recent event fits in with the characteristic earthquake recurrence interval for a particular fault. Second, Little Tujunga Canyon stream-channel downcutting since the mid Holocene may exceed the long-term rate as the stream catches up after an episode of climatically induced aggradation induced by the Pleistocene—Holocene climatic change. The youngest strath terrace might be an internal adjustment terrace instead of representing a return to the base level of erosion datum. Two examples of such anomalously rapid stream-channel downcutting rates are shown in Figure 2.16. The average tectonically The alfisols soils chronosequence for the Transverse Ranges (see Tables 4.2 and 4.3 in Bull, 1991), together with a flight of terraces permits evaluation of changes in late Quaternary uplift rates on the range-bounding Lakeview fault zone. The present interglacial climate is associated with the process of major strath formation rather than aggradation or degradation. A similar climate control of processes was assumed in estimating the ages of older straths. Soils chronosequence age estimates have poor precision and accuracy when compared to other dating methods (see Figure 6.1 and the comparison with eight other methods discussed in Section 6.1.1). Ages based on estimates of the time needed to form soils on terrace treads have larger uncertainties with increase of terrace age. Even so, Figure 3.19 can be used to illustrate how to use stream terraces, and it provides an obvious tectonic conclusion. Mean rates of valley-floor deepening seem to have increased from -0.15 m/ky between 330 and 125 ka to -3.17 m/ky since 6 ka. This apparently rapid rate of Holocene faulting is similar to that documented by Morton and Matti (1987) for the Cu-camonga fault zone at the east end of the San Gabriel Mountains. The most straightforward interpretation of the large increase in tectonically induced downcutting rates is that thrust-fault propagation shifted the 120 Terrace- Stream Age, ka incision, m Stream 0 T9--0.3 4 T7--6 19 T4--30 35 T3B--55 49 T3A-125 72 T2--330 102 200 Age, ka 400 Figure 3.19 Uplift-rate trend for the range-bounding fault where it truncates the terraces of Little Tujunga Creek, based on heights of stream-terrace straths above active-channel strath. Progressive increase in tectonically induced downcutting accompanied the transfer of active thrust faulting from the Internal Sunland to the range-bounding Lakeview fault. Mountain Fronts 95 induced downcutting rate of ^ 1.2 m/ky since 30 ka is a more conservative estimate than the short-term rates for estimating recent increases in uplift rate along the Lakeview fault zone. Tectonic offsets of the stream channel result in deposition downstream from the fault zone as well as incision in the upstream reach, so an apparent slip rate of 1.2 m/ky is a minimum. Total tectonic displacement needs to be increased by at least 19 m. But unlike the tectonic landform of strath terraces, we don't know when episodes of tectonically induced deposition of basin fill occurred unless many layers are dated. The late Quaternary shift in tectonic displacement from the internal to the range-bounding fault is confirmed by general trends of stream-terrace longitudinal profiles. All terraces in the reach upstream from the range-bounding Lakeview fault have straths that parallel each other and parallel the strath beneath the active channel. Unit stream power is ample to keep pace with uplift because of the quite low rock mass strength and the abrasive tools provided by cobbles and boulders derived from resistant rock types upstream from the internal fault. The Holocene strath terrace and active-channel strath are parallel in the reach that passes through the internal Sunland fault zone, implying lack of significant tectonic deformation. In contrast the 30 ka T4 terrace diverges upstream and converges with the active channel downstream from the internal fault zone, which suggests distributed tectonic deformation between 30 ka and 7ka. The 125 and 30 ka stream terraces are separated by more than 20 m in the downstream reaches of Little Tujunga Canyon. In the North Fork of the San Gabriel River (Fig. 3.3), the 55 ka aggradation event was strong enough to bury the strongly developed soil profile on the 125 ka terrace tread. Such composite alluvial fills did not form in Little Tujunga Canyon because rapid tectonically induced downcutting preserved each aggradation event as a separate stream terrace. Stream gradient (SL) indices (Section 2.6.1) have been used to evaluate tectonic and lithologic controls on streams of the San Gabriel Mountains (Keller and Rockwell, 1984). Maps of regional trends are labor intensive. Hack calculated values of the SL index, both narrow and inclusive, for 400 reaches in a single topographic quadrangle, and Keller undertook the ambitious task of mapping the SL indices for the entire San Gabriel Mountains (Figure 4.9 of Keller 0 2 4 6 Distance, km Figure 3.20 Longitudinal profiles of late Quaternary strath terraces of Little Tujunga Canyon. and Pinter, 2002). Their focus is on longer reaches (inclusive stream-gradient index), and they assume that the larger streams of the San Gabriel Mountains adjust quickly to lithologic and tectonic controlling variables. Anomalously high values of the inclusive gradient-index should be indicative of resistant rocks or long reaches that have been affected by recent tectonic deformation. The magnitude and extent of an inclusive gradient-index anomaly in the drainage basin of Little Tujunga Canyon (Fig. 3.21) records rapid uplift on the range-bounding fault. The inclusive stream-gradient index is in the highest 5% category (for the entire San Gabriel Mountains) near the mouth of Little Tujunga Canyon, despite the presence of soft Pliocene mudstones. This indicates rapid uplift rates as suggested by Figure 3.19. This inclusive stream-gradient index anomaly has been present since about 125 ka as shown by the flight of parallel strath terraces. The maximum SL anomaly extends upstream from the internal Sunland fault. Presumably this extent represents the effects of rapid uplift on the range-bounding fault with only a minor base-level fall contribution from the internal fault. The second seismic refraction survey was in rugged Cucamonga Canyon (Fig. 3.22), whose 34 km drainage basin is underlain by gneissic and metasedimentary rocks. Morton and Matti's (1987) studies of thrust-fault scarps on late Quaternary pied- 96 Chap SL>3000 SL2000 SL WOO SL<1000 to 3000 to ZOOO Figure 3.21 Relation of stream-gradient (SL) indices to fault zones in the watershed of Little Tujunga Canyon. Holocene displacements have been large and frequent on the range-bounding fault, and minimal to rare on the internal and San Gabriel fault zones. SL mapping is from Figure 4.9 of Keller and Pinter, 2002. mont alluvial surfaces show that this part of the San Gabriel Mountains also is rising rapidly, as does the SZ-index map. They estimate an earthquake recurrence interval of roughly 0.6 ky, fault displacements of 2 m, and a slip rate of 5 m/ky for the past 13 ky. Three escarpments are present in a 4 km long study reach. The Demens and Cucamonga faults cross the Cucamonga Canyon at fronts 23A and 23C, respectively, and a dissected topographic escarpment suggests the presence of a third unmapped thrust fault at front 23B. Bedrock is not exposed in the stream channel upstream from front 23A, and thickness of streambed gravel above a planar strath does not change across the fault zone. Either the Demens fault is no longer active, or stream-channel downcut-ting rates exceed fault slip rates. A thin blanket of gravel would represent the cutting tools left after the recent large streamflow events scoured the strath. Front 23B is a nice example of how geomor-phic studies can identify a previously unmapped ac- r3 tive thrust fault. A narrow gorge, only 8 to 20 m wide and flanked by vertical cliffs, is present in the reach upstream from where an escarpment, that can be seen on nearby hillslopes, crosses the valley floor of Cucamonga Canyon. This tendency for stream-channel degradation reverses abruptly at the conjectural fault zone. About 10 m of gravel are present below the stream as a wedge of alluvium that thins to 4 m at 200 m downstream from the active thrust fault. At least 10 m of throw has occurred recently on this thrust fault. Total tectonic displacement would be the sum of stream-channel changes in the two adjacent reaches. The thickness of the streambed gravel anomaly should be added to the amount of stream-channel incision into the bedrock. Streamflow diverges and becomes braided upon entering the broad fanhead trench of Cucamonga Canyon, where about 1.8 to 2.6 m of gravel overlie bedrock. Channel gravels thicken abruptly to 10 m just downstream from a streambank exposure of the range-bounding Cucamonga fault zone (same size of streambed gravel anomaly as for the postulated Front 23B active fault zone), and increase further in thickness to 19 m just downstream from a nearby piedmont fault scarp (Front 23D). I conclude that the reach at front 23A may approximate equilibrium conditions. Pulses of uplift appear to exceed channel downcutting rates at fronts 23B and 23C because tectonically induced downcutting has yet to eliminate the most recent surface rupture. Tectonic base-level falls downstream from Fronts 23C and 23D have created space for accumulation of basin fill. Distance, km Figure 3.22 Variations in thickness of streambed gravel along Cucamonga Canyon where it crosses active and Inactive thrust faults. Subsurface faults and geology are diagrammatic. Mountain Fronts 97 In summary, tectonic geomorphology studies of thrust-faulted terrains focus on the distinctive landscape assemblage of piedmont forelands. Recognition of piedmont forelands helps define the tectonic framework of a study area. Studies of the responses of streams to tectonic deformation can also help locate those active thrust faults whose surface ruptures do not extend to the surface - the "blind" thrusts. Useful fluvial landforms include stream terraces, alternating gorge and embayment reaches, streambed longitudinal profiles and stream-gradient indices, and fault offset of bedrock floored stream channels that preserve anomalous pockets of gravel. Studies using these geomorphic tools can identify which thrust faults are most likely to generate future earthquakes. 3.3 Fault Segmentation of Mountain Fronts 3.3.1 Different Ways to Study Active Faults Spectacular mountain ranges commonly are bounded by active faults that pose a hazard to cities in the adjacent basins (Fig. 3.1). Paleoseismologists use diverse information in their quest to learn more about the potentially hazardous fault zones. We date prehis-torical earthquakes and estimate their magnitudes, ascertain surface-rupture heights and lengths, make maps depicting patterns of seismic shaking, and determine if these characteristics are similar or different for consecutive earthquakes. This information in needed not only for range-bounding faults, but also for faults that terminate below the ground surface, and for submarine fault zones. Our quest has taken us to diverse tectonic settings and has resulted in new ways to study earthquakes. Paleoseismologists subdivide long fault zones into sections. Many studies suggest that surface-rupturing earthquakes occur on sections of fault zones with distinctly different structural, lithologic, and topographic characteristics (Allen, 1968; Wallace, 1970; Bull, 1978; Matsuda, 1978; Sieh and Jahns, 1984; Schwartz and Coppersmith, 1986; Schwartz and Crone, 1985; Bull and Pearthree, 1988; and Crone and Haller, 1989; Wells and Coppersmith, 1994). For some faults, recent vertical or horizontal displacements appear to be similar for consecutive earthquakes, which promoted the concept of a characteristic (typical) earthquake as part of the fault segmentation model. We realize the importance of determining if earthquake behavior is persistent for a particular fault. Attempts to simplify repetitive earthquake behavior take rupture length into account. Rupture length is crucial because moment magnitudes of earthquakes, Mw, are a function of the area of a fault plane that is ruptured. A fault zone continues to propagate and evolve with each event, so past behavior may provide only a general guide to future style, timing, and magnitude of earthquakes. Even the best models may not be applicable in all tectonic settings. The Mw 7.3 Landers earthquake of 1992 does not fit simple earthquake models. Dextral slip side-stepped through consecutive "segments" on five "separate" fault zones in the Mojave Desert of southern California (Hart et al., 1993; Sieh et al., 1993). Diverse approaches to paleoseismology are needed when hypotheses of persistent earthquake behavior don't match subsequent events. Fault segmentation studies during the past 20 years have emphasized dating of colluvial, alluvial, and swamp deposits with their intercalated pa-leosols, that are exposed by excavations across fault zones (Swan et al., 1981; Schwartz and Coppersmith, 1984; Sieh 1984). Trench studies have utilized new dating techniques such as optical and thermolumines-cence (McCalpin and Forman, 1991; Harrison et al., 1997), electron spin resonance (Lee and Schwarcz, 1996), and have re-defined the state of knowledge for specific active faults (Hancock et al., 1991). Mathematical refinements of radiocarbon dates and their stratigraphic layers (Biasi and Weldon, 1994; Biasi et al., 2002) are now used by many workers (Grant and Lettis, 2002). These worthwhile efforts have opened interesting new pages about recent earth history. The applied spin-off is much improved knowledge about earthquake hazards posed by active faults such as the San Andreas transform and the Wasatch fault zones in California and Utah. Let us consider the merits and shortcomings of the trench-and-date stratigraphic approach to paleoseismology. This sets the stage for geomorphic studies of active range-bounding faults. Major advantages of trenching fault scarps include: 1) Identification of the fault responsible for a specific surface-rupture event. 2) Presence of distorted strata and liquefaction features that unequivocally demonstrate that slip on the fault plane was seismic in nature. 3) Detailed local information about style and magnitude of surficial faulting. 98 Chap 4) Potential to estimate the times of earthquakes of the past 40 ka, using the widely preferred radiocarbon dating method. Organic materials obtained from faulted and unfaulted strata, deposited before and after the event, constrain the interval during which a prehistorical earthquake occurred. 5) Opportunities to measure offsets of bedding and buried stream channels in order to separate the events responsible for a multiple-rupture event fault scarp. Displacement amounts can be used to estimate magnitude ranges for prehistorical earthquakes. 6) Identification of which scarps contain faults and thus are potentially dangerous building sites, and which were the result of nontectonic processes. Stratigraphic studies made in trenches excavated across fault scarps continue to be important for paleoseismology (Machette, 1978; Weldon and Sieh, 1985; Rockwell and Pinnault, 1986; Sieh et al., 1989; Sieh and Williams, 1990; McGill and Sieh, 1991; Machette et al., 1992; Grant and Sieh, 1993; Lindvall and Rockwell, 1995; and in the book edited by McCalpin, 1996). The November 2002 issue of the Bulletin of the Seismological Society of America is devoted to San Andreas fault earthquakes. It is a treatise about trench-and-date paleoseismology. Age estimates of organic materials created before or after a surface-rupture event have greatly improved our perception of earthquake recurrence intervals and surface rupture characteristics (Sieh et al., 1989: Fumal et al., 1993; McCalpin, 1996; Sietz et al., 1997; Biasi et al., 2002). Fewer hazards assessments have been made for thrust faults hidden in folds (Cifuentes, 1989; Scientists of the U.S. Geological Survey and the Southern California Earthquake Center, 1994; Keller et al., 1998), or for submarine subduction zone thrust faults (Atwater et al., 1991, Atwater and Yamaguchi, 1991; Atwater et al., 1995). Three examples of the marvelous insights that trench-and-date stratigraphic studies may provide are included in this book. Critical details of stratigraphy ruptured by a thrust fault at the Loma Alta site in the San Gabriel Mountains are presented next. Using soil profiles as stratigraphic time lines is illustrated in Chapter 4. Figure 6.50 is an analysis of the Honey Lake fault zone, a major right-lateral strike-slip fault in the Walker Lane tectonic belt. Site selection, careful stratigraphic description and sampling, and making the sensible analyses indeed required the combined talents of Charlie Rubin, Scott Lindvall, and Tom Rockwell (1998) at the Loma Alta site. I summarize their project here be- ter 3 cause of their diverse trenching experience and flexibility in making interpretations. Their goal was to answer an important hazards question. Were recent earthquakes indicative of the hazard level for the Los Angeles metropolitan area? These include the Mw magnitude 6.7 San Fernando earthquake of 1971, Mw 5.9 Whittier Narrows event of 1987, and the most damaging earthquake in the history of the United States, the 1994 Northridge Mw 6.7 event. Even moderate-size surface ruptures of the Sierra Madre range-bounding thrust fault would direct enormous seismic energy southward into the adjacent densely populated metropolis. Crook et al. (1987) were unable to identify fault scarps or ruptured strata younger than late Pleistocene, and concluded that this fault segment was not as tectonically active as segments to the east and west. It may have not produced a Holocene earthquake. A thorough geomorphic reconnaissance was made before spending >$ 100,000 on a trench. Which topographic scarps might record a prehistorical surface rupture? The ancient plane-table surveys used to make the first topographic maps of the American southwest provide details not available in most modern maps. Rubin's assessment of tectonic and nontectonic scarps included re-mapping the entire mountain front using pre-urbanization aerial photographs taken between 1928 and 1935 and early versions of topographic maps, which have 1.5 m (5 foot contours). The emphasis was on locating and then field checking low steps in alluvial surfaces that might be late Quaternary fault scarps. This geomorphic approach was an exercise in understanding the behavior of fluvial systems. They needed to predict the likelihood of having discrete alluvial strata needed to define the locations and magnitudes of fault ruptures, and which sites might allow identification of multiple colluvial wedges in a thrust-fault setting. The deposits had to be young and fine grained enough to contain detrital charcoal. Small alluvial fans fed by minor canyons fit these guidelines better than the bouldery floodplains of large rivers. Loma Alta made the list of finalists. The 5 m deep and 25 m long Loma Alta trench was excavated through a 2-m high fault scarp in a Pasadena city park. It is on a fill terrace that crosses the Sierra Madre fault zone 1.2 km east of Caltech's Jet Propulsion Laboratory (which is next to Arroyo Seco on Figure 3.3). Rubin and Lindvall pegged a 50 cm grid on the trench wall and spent 2 to 3 weeks mapping in fine detail. The location of each cobble Mountain Fronts 99 was noted (Figure 3.23 shows diagrammatic cobble and boulders) and the rationale for collecting each radiocarbon sample was evaluated. Tom Rockwell did the soil profile sampling and description. Soil profiles are essential to frame this small stratigraphic section and correlate it to the San Gabriel Mountains soils chronosequence. Soils of the type sections are dated with more than detrital charcoal and provide a crosscheck of the conclusions reached by the Loma Alta batch of radiocarbon age estimates. The Loma Alta stratigraphic section contains obvious climatic and tectonic signatures. Brief pulses of regional late Quaternary climate change induced aggradation that temporarily reversed the overall trend for tectonically induced downcutting of valley floors in the San Gabriel Mountains. Climate change caused partial stripping of hillslope sediment reservoirs. The resulting fill-terrace treads became sites of soil-profiles that became more strongly developed with the passage of time. Terrace-tread soils may be buried at mountain fronts with active thrust faults. Burial stops surficial weathering and pedogenic processes, but key soil-profile characteristics are preserved in the stratigraphy. The alfisols soils chronosequence for the Transverse Ranges (Table 4.11, Bull, 1991) provides likely matches for the Loma Alta site soil profiles. The young soil above unit 4b is an A horizon that dates to the mid-to-late Holocene. The 46 cm thick buried argillic soil-profile horizon of unit 2 (Fig. 3.23) is overlain by a 36 cm A horizon. This suggests a latest Pleistocene to early Holocene age with the older age estimate preferred because the soil profile would be more strongly developed if it had not been buried. Detrital organic matter for radiocarbon dating is usually difficult to find in gravelly alluvium, even in humid regions. The latest Pleistocene aggradation event gravels in humid New Zealand have yielded few samples from hundreds of exposures. Wood floats downstream or rots. Bouldery braided streams destroy charcoal fragments. Streambank exposures and fault-scarp trenches in the arid Gobi Altai of Mongolia did not provide detrital organic matter. Paleoseismologists turned to other ways of dating displacements. Stratigraphers collected samples for luminescence dating (thermo- NE Recent slopewash buries a mid-to-late Holocene soil profile 10 m Distance 20 m Figure 3.23. Stratigraphic section exposed in the trench wall at Loma Alta, southern front of the 5an Gabriel Mountains of southern California showing fault traces and stratigraphic units 1 through 4b. Lettered x's show the locations of charcoal fragments collected for radiocarbon dating described in Table 3.2. Redrafted from Figure 3 of Rubin et al. (1995). 100 Chap luminescence, TL, and optically stimulated luminescence, OSL of silt fragments). See the work of Prentice et al., (2002) for applications of luminescence dating, and for splendid examples of trench logs that efficiently focus the reader on key aspects of the data and interpretations. Geomorphologists used terrestrial cosmogenic isotopes such as 10Be to date alluvial fan surfaces that had been tectonically translocated by strike-slip faulting or have been incised by tectonically induced downcutting. See the work of Vassallo et al. (2005) for clear-cut examples of results and description of the method. The San Gabriel Mountains provide abundant charcoal fragments to stream channels. Hill-slopes covered with chaparral consisting of waxy and highly flammable plants favor frequent fires and are the source of abundant charcoal that is flushed downstream and deposited with alluvium. The key to dating success involves recognizing three classes of charcoal fragments. 1) Fragments eroded from old alluvial and colluvial deposits, transported a short distance downstream and redeposited with new alluvium. These charcoal samples look nice in the field but yield dismaying results (wide range of ages, with nearly all being much too old). 2) Charcoal created just before deposition with alluvium. For example, a late summer firestorm leaves hillslopes barren and creates much new charcoal, which is transported and deposited by debris-flow slurries during the intense rainstorms of the next winter. Even these preferred materials for dating a single bed can't tell us the length of time between growth of a plant stem and the time of the fire. 3) Charcoal fragments created by brush fires on a sandy stream-terrace tread and then taken underground by subsequent bioturbation processes such as burrowing rodents. The radiocarbon age estimate tends to be much too young, sometimes even modern. One needs many radiocarbon dates to make appropriate age interpretations. A single date for a trench site, such as in the Figure 6.50 example, may be better than none. Having the range of possible ages suggested by the 13 radiocarbon samples at the Loma Alta (Table 3.2) site allowed selection of the most reasonable age estimates for each of the strati-graphic units. Interpretation of radiocarbon and soils information varies depending on the biases and background of paleoseismologists. The nice Rubin et al. scenario is presented first, then my thoughts that use ter 3 the buried soils to a greater extent. You, as readers, can be expected to further develop alternative models. Welcome to the intricate world of "trench-and-date" paleoseismology. Unit 1 is crudely stratified sandy boulder- to pebble-sized gravel. Moderate diagenetic weathering of diorite cobbles suggests that it is significantly older than unit 2. Unit 2 is a fine sandy gravelly loam. A buried soil (A, Bt, and Cox horizon), with a 50-cm thick Bt horizon extends into the top of unit 1. This soil-forming interval of landscape stability ended with the first of two thrust-fault events, which raised part of unit 2 sufficiently to strip it from the hanging wall and redeposit it as the colluvial wedge of unit 3. A second rupture event resulted in deposition of unit 4a, which is a massive colluvial wedge of boulders and gravel in an organic-rich silt and sand matrix. The bouldery unit 4a grades laterally into the extensively bioturbated, organic-rich silty sand of unit 4b of Figure 3.23. Ideally, radiocarbon ages are progressively younger in overlying strata. Charcoal from each of the Loma Alta stratigraphic units provided mixed results. Charcoal fragments can also be anomalously young if introduced by bioturbation after deposition of the unit. This left the research team guessing as to whether the dated fragment was detrital or bioturbated, reworked from older deposits, or created by fire just before deposition. All detrital charcoal provides a maximum age estimate of the stratum that contains it, so the youngest age in a unit is considered as the maximum possible age for a unit. Three lines of evidence indicate the presence of older reworked charcoal in younger stratigraphic units: 1) A radiocarbon age of 29 ± 4 ka was obtained for charcoal from an older stream terrace located about 1 km uphill from the excavation site. Detrital charcoal derived from such older units can be incorporated into any younger alluvium. 2) Angular and rounded charcoal fragments were analyzed from the same sample locality. Sample j of Table 3.2 was angular fragment that provided a radiocarbon age of 14.2 ± 1 ka, and the sample i rounded fragment an older radiocarbon age of 18.2 ± 1 ka. 3) Radiocarbon ages vary greatly for the same stratigraphic horizon, which suggests recycling of older detrital charcoal. An opposite problem may affect charcoal in extensively bioturbated deposits, such as Unit 4b, where young carbon ages may reflect incorporation of charcoal that postdates the time of deposition. Mountain Fronts 101 Radiocarbon sample Radiocarbon age (years B.P. ±1 a) Calendric age range (years B.P. ±2 a) Comments and interpretations Unit 2 a 16,330 + 110 10,912-19,604 All three samples are charcoal fragments in a thick buried A soil-profile horizon b 24,360 + 300 c 16,175 + ÖO 10,004-19,349 UnitS d 20,760 + 370 Recycled charcoal fragments e 29,900 + 390 Recycled charcoal fragments f 15,600 +140 10,191-10,016 Maximum age for deposition of 3 34, 300 + SOO Recycled charcoal fragments Unit 4a h 14,360 + 320 16,445-17,924 Recycled charcoal fragments i 15,235 + 95 17,901-10,405 Recycled rounded fragment j 12.160 + 75 13,097-14,530 Recycled angular fragment k 9,495 + 75 10,349-10,902 Maximum age for deposition of 4 Unit 4b 1 1,105 + 95 937-1,277 Both may be minimum ages if the charcoal dates bioturbation m 3,335 + 35 3,471-3,360 Table 3.2. Radiocarbon analyses of charcoal samples. Samples are listed in stratigraphic order; unit 2 is the oldest, 4b the youngest. See Figure 3.23 for sample locations of a-m. Analytical uncertainties are only for the laboratory measurements. From Table 1 of Rubin et al. (199S). The buried colluvial wedges (units 3 and 4a) are well stratified and show little evidence of bioturbation; thus incorporation of young carbon is unlikely. Rubin et al. conclude that units 2 through 4a are latest Pleistocene to early Holocene in age and that unit 4b is younger. Four fragments of detrital charcoal from unit 3 yielded ages of between 18 and 34 ka. Three appear to be recycled charcoal from older alluvial deposits and even 18 ka is considered as the maximum age for deposition of unit 3 colluvial deposits. Four fragments of detrital charcoal from unit 4a yielded ages ranging from 11 to 18 ka. The 11 ka estimate is considered the maximum age. Two detrital charcoal fragments from unit 4b yielded radiocarbon ages of 1 and 3 ka, and are regarded as minimum ages because it is likely that the charcoal was emplaced by bioturbation after deposition of unit 4b. Two thrust faulting events have occurred since the formation of the soil profile on the unit 2 former land surface (Fig. 3.23). The first surface rupture event is the obvious truncation of unit 2 as slip along the fault plane shoved unit 1 over unit 2. This local landscape instability created the space needed for deposition of the unit 3 colluvial wedge. It was preceded by an interval of landscape stability long enough to form the soil profile. The second event cut through the unit 3 deposits and resulted in deposition of the unit 4a colluvial wedge. Erosion has removed the unit 3 deposits above the fault and has removed part of unit 4a. The magnitudes and importance of the two recent thrust-fault rupture events are clear-cut. Total slip for the two events was 10.5 m, assuming that unit 2 was continuous across the fault zone. Restoration of the upper plate for the most recent earthquake yields a minimum slip of 3.9 ± 0.1 m. Prehistorical earthquakes at the Loma Alta site were substantially larger than the 1971 to 1994 earthquakes along reverse faults in the region. Large displacements of 4 m or greater are inconsistent with short (15 to 20 km) ruptures of the Sierra Madre fault and imply that past earthquakes ruptured much of the fault zone. The component of oblique thrust slip at depth for the 1971 Mw 6.7 San Fernando earthquake was as much as 3 m (Heaton, 1982), but the surface rupture was generally < 1 m (Kamb et al., 1971; U.S. Geological Survey, 1971). 102 Chapter 3 5 m thrust-fault displacement creates space for Unit 3 colluvial wedge Collapse and erosion of fault scarp create Unit 3 and bury Unit 2 soil profile iiiiiimiinniiiiiNHiiHitiDuiiiN i'.^.i.r. 5 m slip bisects Unit 3 colluvial wedge and creates space for Unit 4 wedge mrrnnwinnnwTTTF i w n it run uuu u mi h u ii w it unit it iih tt Collapse and erosion of fault scarp remove part of Unit 3 wedge and create Unit 4 colluvial wedge that buries remnant of Unit 3 wedge Figure 3.24. Cartoons showing development of colluvial wedges from two successive earthquakes at the Loma Klta trench site on the range-bounding Sierra Madre thrust fault, Southern California. Redrafted from Figure 5 of Rubin etal. (1995). If we assume that 5.25 m represents the average surface displacement, a regression analysis predicts a Mw magnitude earthquake of 7.5 ± 0.5. With an average slip of 5.25 m, a strike length of 65 km, and a seismogenic depth of 18 km, the seismic moment for the most recent earthquake is 2.65 x 1027 dyne-cm. Seismic moment (Aki, 1966) is the product of fault-surface area, mean displacement on the fault, and rock rigidity. Converting seismic moment to moment magnitude yields Mw magnitudes (Hanks and Kanamori, 1979) of -7.5 for both prehistorical earthquakes recorded at the Loma Alta site. Tucker and Dolan (2001) excavated a trench on the Sierra Madre fault 34 km ESE of Loma Alta. They found evidence for 14 m of slip between 24 and 8 ka. Infrequent, but large (Mw > 7) earthquakes at their site agree with the Rubin et al. (1998) conclusion of recent large earthquake (s). Both recent earthquakes at the Loma Alta site were much larger than recent historical earthquakes. The results of the trench study support Houghs (1995) fractal model that most of the seismic moment release in the greater Los Angeles area is by infrequent but large events (Mw = 7.4 to 7.5). Damage from large magnitude earthquakes along the Sierra Madre fault would be substantially larger than that caused by the 1994 Mw magnitude 6.7 Northridge earthquake. Near-field, high-amplitude ground motions would ripple through a much larger area and for a longer duration. Unlike the Northridge earthquake that ruptured northward away from the metropolitan region, a Mw magnitude >7 earthquake on the Sierra Mountain Fronts 103 Madre fault would rupture southward, directing its energy into the adjacent densely populated basin. The Loma Alta trench project resulted in re-assessment of seismic hazard and underscores the value of the stratigraphic approach to paleoseismol-ogy. The research team was careful and conservative in evaluating the times of the surface rupture events. Of course one would prefer several independent ways to cross-check the conclusions based on radiocarbon dating of charcoal fragments. Lacking these, one can use soil-profile characteristics to see if they agree with the overall conclusions. Each soil profile records a time span postdating the alluvium it is formed in. So times of faulting postdate both the stratigraphic and soil-profile ages. Unit 2 is central to any temporal analysis because it has a well-developed soil that was overridden by the first thrust faulting event. So let us consider two alternative times of faulting. Charcoal fragments constrain when the alluvium was deposited, but not the time span required for subsequent soil-profile formation. Radiocarbon samples a and c date to 19 ka and like recycled fragment b are in the A soil-profile horizon. Assume that both are contemporaneous with the deposition of unit 2, and indeed represent the maximum possible age. Deposition was followed by pedogenesis long enough to create a strongly developed soil before scrunch tectonics offset and buried the surface of the fill terrace. The youngest soil with an argillic horizon in the soils chronosequence for the San Gabriel Mountains has an age of 4 to 7 ka (Bull, 1991, Chapter 4). It might take 5 to 10 ky to form the unit 2 soil. If so, the penultimate event may have occurred in the early Holocene. A second scenario to consider. The sandy, organic, A horizon of the unit 2 buried soil would have been favorable for bioturbation. If so, samples a and c could be younger than when unit 2 deposition ceased. This leaves sample b, with a conventional radiocarbon age of 24 ky, as being more representative of the maximum age of unit 2. The likely timing of the first surface rupture event becomes much older, but still is offset by time needed to create the now buried soil profile (-24 ka minus -8 ka =16 ± 5 ka). Both unit 4b charcoal dates were considered minima. But note that the incipient soil profile capping this unit has characteristics requiring only 1 to 3 ky to form elsewhere in the San Gabriel Mountains. Both samples may have been moved about by bioturbation, but could well have been initially deposited with the recent sandy alluvium. None of my second thoughts about the timing of depositional, pedogenic, and tectonic events change the important conclusions of Rubin et al. (1998). Two large recent prehistorical earthquakes ruptured much of the Sierra Madre fault zone with important implications for seismic hazard planning. Trench-and-date paleoseismology has several drawbacks. 1) Not all faults have accessible scarps including blind thrust faults hidden in anticlines (Unruh and Moores, 1992; Bullard and Lettis, 1993; Unruh, 2001), submarine fault scarps, and faults beneath large cities. California earthquakes between 1983 and 1994 at Coalinga (Atwater et al., 1990), Kettle-man Hills (Ekstrom et al., 1992), Whittier Narrows (Hauksson et al., 1988), Loma Prieta (McNutt and Sydnor, 1990), and Northridge (Hudnut et al., 1996) call attention to severe seismic hazards posed by faults that splay into concealed branches, or that terminate in folds, before reaching the surface. Earthquakes on these blind thrust faults caused huge financial loss and cost many lives. Surface ruptures of great sub-duction zone earthquakes are concealed beneath the sea (Heaton and Kanamori, 1984). Trench-and-date stratigraphic work is not possible in such settings. 2) Organic matter dated by radiocarbon analysis either predates or postdates times of disruption of seismic stratigraphy. Using multiple samples the time of an earthquake may be bracketed by analyses of organic matter formed before and after the event, but we cannot radiocarbon date the time of an earthquake. Atmospheric radiocarbon production rates may vary sufficiently to result in multiple possible calendric radiocarbon ages for a single sample (Stuiver et al., 1998). This problem is especially acute for the crucial post-1700 A.D. interval. When combined with the usual laboratory analytical error of ±40 years, the uncertainties for many radiocarbon estimates of times of earthquakes are larger than generally acknowledged. These complications make it difficult to use radiocarbon dating to separate events on faults where the earthquake recurrence interval is 100-300 years. Detrital organic matter, such as charcoal and wood deposited by streamflows, presents additional challenges for the paleoseismologist because such materials are older than their time of deposition by an unknown amount. The material that we date in a layer that is below a faulted former land surface at the time of a prehistorical earthquake predates both its time of deposition and the time of the earthquake. Detrital organic matter in a layer above the former 104 Chapter 3 faulted land surface might have an age that is younger, older, or the same as the time of the earthquake. Fortunately, new stratigraphic and surface-exposure dating methods, such as optical stimulation luminescence (OSL) and terrestrial cosmogenic nuclides (TCN), extend the range of dating and provide cross-checks. See the excellent book by Noller et al. (2000). Precision and accuracy of nine dating methods are compared in Figure 6.1 of this book. 3) Earthquakes occurring during times of nondepo-sition at a trench site are not likely to be recognized. Omission of such earthquakes creates a false impression of irregular (clustered earthquake) behavior (Bull, 1996a). 4) Trench studies of fault segmentation are time consuming and expensive. Decades of work in two dozen trenches were required to better understand the behavior of the Wasatch Range bounding fault zone in Utah. Little is known about overlap between adjacent segments because few excavations are made in segment boundaries. 5) Trench-and-date stratigraphic studies cannot evaluate the extent and intensity of seismic shaking and are a cumbersome way to describe extent of a surface rupture. New approaches are needed in paleoseismol-ogy. Tectonic geomorphology is an important but under utilized tool with potential to complement important trench-and-date studies of earthquakes. Mapping and soils dating of alluvial surfaces allows optimal selection of trench sites by defining the extents and approximate times of prehistorical surface ruptures. Tectonic geomorphology studies use diverse data from topographic, pedogenic, stratigraphic, hy-drologic, botanical, and structural geology sources. The geomorphic approach compliments geophysical studies and may not require expensive and time-consuming trenching projects. Multidisciplinary approaches can solve more problems. Several new approaches to paleoseismology have been selected for this book. These include the work of Kirk Vincent (1995) who made detailed studies of longitudinal profiles of faulted alluvial fans. His analysis of segment-boundary behavior of tectonically active mountain fronts provides the most rigorous field test to date of the fault segmentation model of Schwartz and Coppersmith (1984). Regional appraisal of the relative tectonic activity of mountain fronts (Chapter 4) is important in selecting safe locations for nuclear power electrical generating stations. Seismic risk is minimal for range-bounding faults that have been inactive for 1 My and are distant from active fault zones. Geomorphic processes that are sensitive to seismic shaking can be used to date and locate earthquakes, estimate their magnitudes, and describe regional patterns of seismic shaking caused by prehistorical earthquakes (Chapter 6). The best approach to paleoseismology is to make both stratigraphic and geomorphic studies. The need to make geomorphic maps before trenching is stressed by McCalpin (1996, p. 34). Traditional dating of faulted stratigraphic sections in trenches is an essential next step. Geomorphic studies that define rupture lengths and displacement amounts for each event allow evaluation of how slip changes where a fault scarp extends into a segment boundary. Surface-exposure dating of coseismic landslides can recognize earthquakes that occur during times of nondeposition at a trench site (Bull, 1996a), and can be used to make maps of seismic shaking comparable to Mercalli Intensity maps (Bull and Brandon, 1998). Then, with both stratigraphic and geomorphic data in hand, we can better appraise the persistence of fault-zone behavior and potential surface rupture and seismic-shaking hazards. 3.3.2 Segmentation Concepts and Classification Even large earthquakes do not rupture the entire length of long fault zones. The world's longest known historical rupture of a normal fault may be only 70-75 km (Section 5.4.2). The strike-slip San Andreas fault of California does not rupture its entire length; instead surface ruptures occur along one of four segments that are 150 to 500 km long (Allen, 1968). The fault-segment model is used to estimate lengths of potential surface ruptures, earthquake magnitudes, and to calculate seismic moment (Aki, 1969; Tsai and Aki, 1969; and seismic moment rate, Brune, 1968). A major endeavor in active tectonics is to decipher the spatial and temporal distribution of consecutive surface ruptures on normal, reverse, and strike-slip faults. Such work provides insight about the frequency, magnitude, and style of stress release along plate boundary and intra-plate fault zones. This research has significant implications for studies of crustal mechanics, origins of mountainous landscapes, and earthquake hazards. Defining fault segments is a paleoseismic endeavor that strives for better understanding of the behavior of active faults and of the seismic risk they pose. Fault segmentation is a conceptual model that Mountain Fronts 105 provides constraints for estimating earthquake size by recognizing adjacent sections of faults, each with their own style, magnitude, and timing of surface ruptures. The theme of this model of fault behavior is that a fault segment tends to rupture repeatedly in similar-size, or characteristic, events that approximate the maximum possible magnitude earthquake. The model implies temporally constant length and displacement amounts for each of a sequence of fault segments. A basic premise is that future large earthquakes - the maximum magnitude earthquake model of Wesnousky (1986) - are most likely to occur on those faults with the largest cumulative displacements and most rapid slip rates during the late Quaternary (Allen, 1975; Matsuda, 1978). Identification of segments of a fault zone that repeatedly rupture independently of one another is not easy. Ideally, one would use repetitions of historical surface ruptures. Such data are rarely available, so paleoseismologists are forced to use less reliable approaches. Knuepfer (1989, Table 3) ranked criteria for classification of earthquake segments in terms of the likelihood of being future locations of earthquake surface ruptures. Rupture limits for historical earthquakes are ideal (100%). Rupture limits for well-dated prehistorical earthquakes could be almost as good, but even multiple trench sites usually provide information about the most recent one or two earthquakes. Topographic, lithologic, and geophysical changes along a fault zone have a reliability of less than 50% (39%), as do geologic characteristics such as structural branching and intersections with other faults (31%). Spatial changes in style of faulting or slip rates and in earthquake recurrence intervals are of little use (26%), as are changes in fault orientation such as bends and stepovers (18%). It seems that most attempts to postulate fault-segment boundaries are merely models to be tested by innovative studies and future earthquakes. We should also always allow for the possibility that some fault zones may alternate between different styles of behavior. Topographic changes are used in this chapter because they record persistent tectonic deformation quite well. Whether or not they indicate the extent of the next surface rupture depends on how many segments are involved in the event. Landscape analyses can decipher spatial variations of vertical tectonic displacements of mountain fronts that range from high escarpments to low fault scarps, using the landforms of hillslopes, streams, and alluvial fans. Mountain-front topography may not tell us much about the magnitude and extent of future slip events. Its primary value is the way it integrates late Quaternary fault displacements and erosion to create a record of the long-term rate, magnitude, and style of tectonic deformation. Having a landscape persistence perspective for the past 100 ky provides a useful counterpoint for radiocarbon dating of earthquakes for the past 40 ky. 3.3.3 Fault-Segment Boundaries Barriers to rupture propagation define the ends of fault segments in a variety of settings ranging from subduction zones (Aki, 1984) to Basin and Range Province normal faults (Wheeler et al., 1987; Fon-seca, 1988). Boundaries between segments typically are structurally complex and range in width from less than 1 km to more than 10 km. We seek information regarding which discontinuities consistently act as barriers to propagation of strain release during earthquakes. A 370-km long zone of normal faulting bounds the Wasatch Range of Utah, which is the structural boundary between the Basin and Range Province and the Colorado Plateau. Spatial patterns of cumulative displacement were used to define a characteristic earthquake model (Schwartz and Coppersmith, 1984; Schwartz and Crone, 1985). Historical earthquakes have not occurred but three decades of trench studies reveal many Quaternary surface ruptures (Swan et al., 1981; Machette et al., 1989, 1991, 1992; Machette and Brown, 1995; Mc-Calpin, 1996). Magnitudes of vertical displacement recorded at a trench site tend to be similar, and individual surface ruptures are largest near centers of segments and become less near terminations. Of course surface ruptures cannot always be zero at segment boundaries (the transition zone between adjacent segments of a fault zone), because those parts of a mountain range also have been raised relative to the valley. So surface ruptures of adjacent segments must overlap (Fig. 3.25). Assume that: 1) the center half of each segment is not affected by ruptures in adjacent segments, 2) ruptures do not stop abruptly upon entering a boundary between fault segments, but continue on for a distance of one-fourth of the length of the next segment, and 3) repetitive displacements occur with similar frequencies and magnitudes for all segments. 106 Chapter 3 Fault segment boundaries Figure 3.25 Spatial variations of sizes of characteristic surface ruptures in adjacent fault segments assuming 1) rupture events of similar magnitude and size, 2) that the middle half of each segment is not affected by displacements in adjacent segments, and 3) that the surface rupture in a given segment dies out one-fourth of the distance into the adjacent segments. Frequency and magnitude of surface ruptures are different at locations A, S>, and C of this model. Temporal patterns of surface ruptures will vary along the hypothetical segmented fault zone (Fig. 3.25). At location A, the earthquake recurrence interval is assigned a value of 10 ky and amount of displacement is assigned a value of 2 m; each displacement is similar and large. At location B the earthquake recurrence interval would have a value of 5 ky and displacements would have values of 0.5 m; each displacement is similar and small. At location C the earthquake recurrence interval would also have a value of 5 ky, but the displacements usually would alternate between large (0.8 m) and small (0.2 m). Real-world fault zones are more complicated than this model, particularly where consecutive surface ruptures have different lengths. Although diagrammatic, Figure 3.25 illustrates the difficulty of using single trench sites to estimate the frequency and magnitude of surface ruptures on range-bounding strike-slip, normal, and thrust faults. Many variations of the characteristic earthquake model have been proposed and most can be tested in the field. Using tectonic geomorphology, Kirk Vincent devised tests of the model which are described in the next section. 3.3.4 Normal Fault Surface Ruptures The fault segmentation model needs validation, particularly in regard to the style and pattern of consecutive surface-rupture events where they enter a boundary between two fault segments. This section addresses that topic by summarizing the work of Kirk Vincent on an active left-lateral oblique normal fault in Idaho (Crone et al., 1987). Discussion begins with how to use faulted fluvial landforms to estimate apparent vertical displacements. Then we correct apparent throw to obtain true values of vertical displacements. These geomorphic procedures provide insight into four consecutive late Quaternary earthquakes on the Thousand Springs and Mackay segments of the Lost River Fault, and in the 10 km long Elkhorn Creek segment boundary between the two segments (Fig. 3.26). This essay is about how to measure true tectonic displacement and what hap- Mountain Fronts 107 pens to surface ruptures in the boundary transition between two distinctive fault segments. Estimated ages and correlation of fluvial-system aggradation events were based on radiocarbon analyses, a soils chronosequence emphasizing soil-carbonate pebble coatings and stage of carbonate pedogenesis, and the 6.8 ka Mazama volcanic ash (Pierce and Scott, 1982; Scott et al, 1985; Vincent et al., 1994; and Vincent, 1995, Chapter 2). Approximate calendric ages of the prehistorical earthquakes are 5 ± 1 ka, 10.5 ± 1 ka, and 14 ± 2 ka. A Ms magnitude 7.3 earthquake in 1983 provided an opportunity to collect data not readily available for the prehistorical earthquakes. Kirk Vincent (1995) surveyed topographic profiles of faulted alluvial surfaces. Longitudinal profiles of alluvial fans and stream terraces typically are straight, or have a systematic curvature. This allows estimation of the vertical component of fault displacement for the entire width of a fault zone instead of just the main strand of the fault. It is easy to mistake the height or throw (referred to as apparent throw in this discussion) of a fault scarp as being representative of true displacement, especially where tectonic deformation along secondary faults is not obvious. See the disparities illustrated by Figure 3.27. Net throw is best measured as the vertical separation of the topographic profile for the landform that was faulted, but that too is an "apparent" vertical tectonic displacement whose correction is discussed in this section. Surface ruptures associated with normal faults of the Basin and Range Province typically consist of a zone of deformation that is 10 to 500 m wide. Secondary features such as antithetic and synthetic scarps, tilted blocks, sagged or bulged ground, and rifts become difficult to see with the passage of time (Gilbert, 1890; Slemmons, 1957; Vita-Finzi and King, 1985; and Xiao and Suppe, 1992). Visual prominence of the primary fault scarp can be a misleading indicator of the magnitude of total vertical displacement because tectonic dislocation is distributed over the entire width of the rupture zone in a style that varies along a given fault zone. Apparent net throw can be determined by surveying the land surface over a distance that exceeds three times the width of the rupture zone. Pre-faulting shapes of undissected flood plains, stream terraces, and alluvial fans can be reconstructed with confidence by making surveys with an electronic distance meter. Vincent surveyed streamflow paths over distances of 200 m to 800 m for many alluvial land-forms in his Idaho study area and found that most longitudinal profiles are straight over distances of 100 to 800 m. Straight-line profiles downslope from the Lost River fault have a gradient that generally is slightly less than the gradient upslope from the fault. First, let us describe the reference surface of a fluvial landform using a graph of surface altitude plotted against horizontal distance along the general path that streams flowed to construct or maintain the landform. The survey transect may be broadly curved, straight, or meandering and the longitudinal profile typically appears straight or changes gradient gradually. The systematic curvature of a longitudinal profile of a stream that is entraining and depositing bedload reflects orderly adjustments between many variables. The stream may be aggrading, degrading, or at equilibrium. The resulting stream channel, floodplain, or braided depositional surface has a longitudinal profile that may be described by an arithmetic, exponential, or power-function equation. Consider a reach of a stream at flood stage (when much of the work of streams is done) where Figure 3.26 Latest Pleistocene, Holocene, and 1953 surface ruptures of the Lost River fault, Idaho. From Figure 6.3 of Vincent (1995). 108 Chapter 3 Profile vertical separation Figure 3.27 Cross section of a hypothetical normal-fault rupture zone, 'bcarp height is larger than throw because the fan surface is inclined. Most range-bounding faults are fault zones, so neither scarp height nor throw at the main scarp is equivalent to net throw over the entire fault zone. Figure 5.1 of Vincent (1995) the discharge of water and sediment, channel geometry, and hydraulic roughness are uniform. In this ideal situation, stream power and resisting power are constant but may not be equal. This condition defines a straight energy grade line. A landform that evolved during uniform flow conditions will parallel the energy grade line and thus will have a straight longitudinal profile. The longitudinal profile of an alluvial surface created during equilibrium or aggradation conditions is preserved when the mode of operation switches to degradation. Vincent surveyed such stream terraces and surfaces of incised alluvial fans. Subsequent vertical displacement of the abandoned fluvial surface by a normal fault results in two sections of the longitudinal profile above and below the rupture zone that originally would have had virtually the same gradient. By projecting both undeformed profile sections into the rupture zone, one can estimate vertical offset as the vertical separation of parallel lines. Field examples of faulted longitudinal profiles are used here to illustrate ideal, and not so ideal, situations and to clarify methods of measuring vertical displacement. The straight narrow grassy floodplain of perennial Willow Creek (Fig. 3.28) has a longitudinal profile that is almost ideal for a tectonic geomorphology analysis. The surface-rupture zone for the normal fault is about 100 m wide, and is bounded by a main scarp and an antithetic thrust fault. The floodplain slopes 0.035 m/m upstream from the fault zone, and 0.034 m/m in the downstream reach. The tectonically deformed reach within the rupture zone is steep because a block of alluvium was tilted downstream during the 1983 earthquake. The mean gradient of the tilted block, including the scarp height, is 0.052. Three interpretations are possible for the 700-m long survey transect. 1) The two segments of the pre-1983 Borah Peak earthquake stream profile are statistically the same, and the essentially constant gradient reflects uniformly interacting streamflow parameters for the 700 m reach prior to the 1983 surface rupture. 2) The apparent decrease in gradient downstream from 0.035 to 0.034 is real at the 95% confidence level and reflects the pre-earthquake downstream decrease in gradient typical of most streams. 3) Minor tectonic tilting of the blocks upstream or downstream from the fault zone occurred as a result of the 1983 earthquake. Vincent assumed that either the first or second interpretation applies, which allows us to assess the magnitude of the apparent throw in the 0 to 100 m reach. Only 1 m is expressed as a fault scarp, with the remainder masked by complex deformation in the rupture zone. The straight profile segments above and below the fault zone are projected through the fault zone in order to estimate the total vertical displacement. Vertical separation of profile projections is 2.4 m at the main scarp, 2.3 m in the middle of the fault zone, and 2.2 m at the antithetic thrust. The estimate of total vertical displacement can be defined in one of two ways: as the average of the two values on the upslope and downslope sides of the fault zone, or as measured at the center of the fault zone. The extreme values are used to assign an uncertainty for the precision of the estimate. The total 1983 vertical displacement by the Thousand Springs fault at the Willow Creek site was 2.3 m ± 0.1 m. The longitudinal profile of a latest Pleistocene incised fan surface is not as ideal as the Willow Creek floodplain, but it is good enough to estimate apparent total vertical displacement by two surface-rupture events. This late Pleistocene aggradation surface is eleven times steeper than the nearby floodplain, and changes in gradient occur even within a distance of 100 m. The rupture zone consists of a graben bounded by a main scarp and an antithetic scarp, both with normal fault displacements. The situation is favorable for three reasons. The fault zone is narrow, be- Mountain Fronts 109 35 [-'-1-'-1- + is EPM survey control point Longitudinal Profile Separations ■ 200 -100 0 100 200 300 Horizontal distance from main 19S>3 fault ecarp, m Figure 3.25 Longitudinal profile of the floodplain of Willow Creek, which was faulted during the 1953 Borah Peak earthquake. Figure 5.2 of Vincent (1995) ing only 30 m wide. Topographic profiles on both sides of the rupture zone are well-constrained straight lines. Fan gradients upstream and downstream from the fault are similar: the footwall-block gradient is 0.41 m/m (22.3° slope), and the downstream hanging-wall block gradient is 0.39 m/m (21.3° slope). The profiles of the two straight fan segments are shown projected through the rupture zone in Figure 3.29. Apparent total vertical displacement resulting from the two surface rupture events is 2.4 m ± 0.2 m at the main scarp and 1.8 m ± 0.2 m at the antithetic scarp. The ± 0.2 m uncertainty is only 12%, which is good for a fan surface that has been faulted twice and has undergone dissection by consequent streams since -15 ka. Vertical offset is about 2.1 m ± 0.5 m. Unfortunately the apparent throw is much less than true vertical displacement, because of the steep slope of this landform (discussed below). In summary, Kirk Vincent's method of estimating apparent vertical displacements seems to provide maximum values at the main scarp and minimum values at the antithetic scarp. These extremes provide relative estimates of uncertainty as well as mean values. Measuring the vertical separation of straight-profile projections at the center of a rupture zone is the best measure of apparent throw. Vincent's method can be applied to the occasional case of a curving longitudinal profile, where the only valid estimate of apparent throw is at the center of the fault zone. In order to complete our discussion of his approach we need to discuss why the estimates of throw discussed so far are only apparent. Let us examine a correction procedure and its application. The vertical separation of the faulted topographic profiles of the two alluvial geomorphic surfaces (Figs. 3. 28 and 3.29) is an apparent throw because normal faulting of these fans causes both vertical and horizontal tectonic displacements of the land surface (Wallace, 1980). Apparent throw is less than true vertical displacement where a topographic profile is inclined in the direction of dip. Corrections generally can be made to obtain true vertical components of fault motion where longitudinal profile azimuths are within 15° of being perpendicular to fault strikes. Corrections are minor to moderate, but are large where fan slope approaches the fault-plane dip. Two end members are unlikely, but approximations are sufficiently common to warrant consideration. First, imagine a 20° planar hillslope with a rock slab that is resting on a slide plane with a 20° dip (hillslope and failure plane are parallel). A vertical rift develops and the rock slab slides downslope. Projection of the longitudinal profiles of the two straight hillslope profile segments would suggest that no vertical displacement had occurred, even though the slab moved to a lower altitude. The slab would be extended horizontally and lowered in altitude by the same amount. The second case is the opposite extreme. If either the fault-plane dip is 90° or the land-surface 110 Chapter 3 Horizontal Distance, m Figure 3.29 Longitudinal profile of a steep alluvial fan that has been faulted twice: during the 1953 Borah Peak earthquake and by an ~ 11 ka earthquake. Figure 5.3 of Vincent (1995). slope is 0°, apparent throw of topographic profiles would be the same as the true vertical displacements. Both cases are rare. The general case may be described by an equation and by a field of percentage correction factors illustrated by Figure 3.30, or an equation from Vincent (1995, p. 99). If V( is true vertical displacement, V the apparent vertical displacement, a the dip of the fault plane in degrees, and (3 the land-surface slope, then Vt= VJ(sina)(sin(90°+$)] sin (oc-pj 1 j Figure 3.30 shows graphical percentage correction factors to be added to apparent throw values. Typical Basin and Range Province settings have fault dips of 30° to 60° and fan slopes of <10° that result in minimal, or at least manageable, corrections to obtain true vertical displacements. Fault dip is rarely known with certainty, so fluvial landforms with gradients of < 3° (< 0.05 m/m) offer the best chance for accurate estimates of vertical displacements. Situations where corrections exceed 100% should be avoided. Consider three faulted fans. If a 3° stream profile was apparently displaced vertically 1.00 m by a 50° fault, the Figure 3.30 correction for profile displacement would be less than 5%. The estimate for true vertical displacement would be 1.05 m. For an alluvial fan sloping less than 5°, apparent throw would be increased to 1.10 m. The steep alluvial fan of Figure 3.29 had a pre-surface rupture slope of about 21.6° and at that location the Lost River Fault dips about 50°. The apparent throw of 2.1 m should be increased by 50% to 3.05 m. The vertical displacements at the edges of the surface-rupture zone that were used to calculate the uncertainty should be increased proportionately. The true vertical displacement of the faulted alluvial fan of Figure 3.29 is 3.0 m ± 0.4 m. A much clearer picture of the segmentation characteristics of the Lost River fault zone emerges when corrections are made to obtain true vertical displacements of the faulted alluvial surfaces surveyed by Kirk Vincent (Fig. 3.31). Although the 50° dip 20 30 40 50 60 70 Dip of fault-slip surface in degrees Figure 3.30 Percentage corrections of apparent vertical tectonic displacements (throw) of topographic profiles of alluvial landforms based on equation 3.1, knowing longitudinal profile gradient and fault dip. This chart provides estimates of true vertical components of slip on normal faults. The ground surface gradient has a strong influence on the correction factor where the dip of the slip surface is 30° to 70°, but is < 50% for fluvial landforms with < 10° gradients. From Wallace (1950). Mountain Fronts 111 of the range-bounding fault is uniform, fan gradients increase progressively from Willow Creek to Elkhorn Creek. So progressively larger corrections to apparent throw are necessary. Part of the Thousand Springs segment is shown from about 14 to about 5 km on the horizontal distance scale. The Elkhorn Creek segment boundary extends from about 5 to 4 km. The Willow Creek site (Fig. 3.28) is located at about 14 km, and Elkhorn Creek is located at about 3 km. All the surveyed piedmont landforms have been faulted twice, during the 1983 Borah Peak earthquake and during a prehistorical earthquake. Apparent throw (Fig. 3.31 A) appears to decrease gradually to the southeast along the entire 16 km of the fault under consideration. In contrast, the spatial pattern of true vertical displacements of longitudinal profiles remains constant at about 4.5 m for the 7 km and then decreases abruptly to less than 1 m in the segment boundary (Fig. 3.3IB). A similar procedure for all estimates of throw on the Thousand Springs segment (Fig. 3.32) shows the magnitudes of the corrections needed and reveals highly variable amounts of true displacements. -12 -10 -b -6 -4-2 0 2 Horizontal distance along Lost River fault 6 SI 5 £ u S3 4 Q_ in ^3 3 o SI o u 2 » u '■p i_ 0 Range front fault | \ Segment boundary In summary, apparent throws of longitudinal profiles of alluvial landforms provide nicely constrained estimates of fault displacement. Corrections for the effects of land-surface slope relative to fault-plane dip on true values of vertical displacements are worthwhile if slopes are steep and/or if fault dips are shallow. Both of the postglacial earthquakes on the Thousand Springs segment were large in magnitude and had nearly identical displacement patterns. Cumulative vertical displacement is more than 4 m at the center of the segment, but is only 1 m in the Elkhorn Creek segment boundary. Ruptures entering the northern margin of the segment boundary divide and displacement magnitude decreases dramatically. Up to this point our discussion has concentrated on field methods and analytical techniques. Care in selecting appropriate and diverse field sites, precise surveying of longitudinal profiles of alluvial landforms, and correction of apparent throw values are all needed to obtain reliable values of true vertical displacement. Vincent's accurate dataset allows us to better understand fault behavior in the Elkhorn Creek -i—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—i- Figure 3.31 Cumulative displacements caused by two earthquakes, the 1953 Borah Peak earthquake and the ~ 11 ka earthquake, in the transition zone between the Thousand Springs segment and Elkhorn Creek segment boundary. A. Throw (apparent vertical displacements of longitudinal profiles along fault trace). B. True vertical displacements where the apparent throw data have been corrected for the dip-slope effects using the Figure 3.30 chart. Figure 6 of Vincent and Bull (1990). -14 -12 -10 -b -6 -4 -2 0 Horizontal distance alonq fault, km 112 4 Chapter 3 Figure 3.32 Comparison of throw and true vertical displacement for the 1953 E3o-rah Peak, Idaho, earthquake. Figure 5.9 of Vincent (1995). Distance along fault trace, km segment boundary, and to test the fault segmentation models as proposed by Fraser et al. (1964), Sieh (1981), and Schwartz and Coppersmith (1984). Deciphering sequences of small and large surface ruptures is the basis for evaluating earthquake hazards, rupture initiation and termination, and evolution of mountain ranges. Several segmentation models have been proposed. Surface ruptures of the Mw magnitude 7.5 Hebgen Lake earthquake of 1959 in Montana provided Fraser et al. (1964) data for a model that assumes 50% overlap. Ruptures generated on adjacent fault segments overlap completely. The result over the long term is uniform cumulative slip. Their model accommodates discrete surface-rupture events on range-bounding fault segments, and uniform mountain-range uplift over geologic time spans (uniform range-crest altitudes). The model of Sieh (1981) also accommodates uniform slip that results in uniform range-crest altitudes. It uses major earthquakes close to the maximal size for a given fault segment. Schwartz and Coppersmith (1984) proposed a model of segmentation where similar size maximal earthquakes may recur on a segment, and that smaller earthquakes are not the norm. Rupture overlap between adjacent segments is minor. The result over the long term is nonuniform cumulative slip. A key input to fault-segmentation models is "how much do ruptures on adjacent segments overlap?" Vincent was able to answer this question in Idaho because surface ruptures are well preserved for four earthquakes on two adjacent fault segments (Fig. 3.33). Both segments have ruptured independently twice since the latest Pleistocene aggradation event, thus the interactions of four surface-rupture events can be investigated. The Thousand Springs fault segment in the northern portion of his study area ruptured in 1983 and in the early to middle Holocene. The southern portion of his study area is the Mackay fault segment, which ruptured once just after the end of the latest glacial interval and once after 7 ka (Vincent and Bull, 1990). The central portions of both fault segments are characterized by steep, straight, mountain fronts and recent faulting has occurred near the mountain-piedmont junction. The region separating the two segments is a rupture asperity referred to as the Elkhorn Creek fault segment boundary. The strike of the range-front bends within the 10-km long segment boundary forming a spur of hills, which rise in steps up to the highest peak in Idaho, Borah Peak. Both post-glacial earthquakes on the Mackay segment were large in magnitude with nearly identical displacement patterns. Ruptures on the Mackay segment penetrated the segment boundary, but details are lacking. Surface ruptures from the Thousand Springs and Mackay segments appear to overlap as much as 6 km in the segment boundary. The conspicuous cumulative slip low in the Elkhorn Creek segment boundary, which contains Borah Peak, suggests difficulty in locating late Quaternary displacements. Careful mapping by Vincent shows that the trace of the 1983 surface rupture is 14% longer than thought by Crone et al. (1987). He found 10 km of fault scarps and ground cracks within the segment boundary that are distributed over 15 km2. They range in altitude from 2000 to almost Mountain Fronts 113 c 6 I 5 ° 4 Q_ vn , ^3 ° to n o 2 p £ 1 Thousand Springs Segment boundary Mackay fault segment Earthquake at -10 to 11 ka Earthquake at -12 to 17 ka Earthquake at ~5 ka -15 -10 -5 0 5 10 15 Distance along Lost River fault trace, km 20 Figure 3.33 Calendric ages and patterns of true vertical displacements for four late Quaternary earthquakes along the Thousand Springs and Mackay fault segments and the intervening Elkhorn Creek segment boundary. Points for the Thousand Springs and Mackay segments are surface ruptures of the range-bounding faults, and points within the segment boundary are the sum of corrected throw for both range-bounding and many internal faults. Black circles are used for faulted landforms that are late glacial in age. Open circles are used for Holocene surfaces faulted at about 5 ka. Gray circles are used for surficial displacements for the 1953 Borah Peak earthquake. Large circles are used where surveys were made of large landforms with an electronic distance meter, small circles for displacements of small stream terraces estimated by eye using Suunto and hand levels. Figure 6.5 of Vincent (1995) 3000 m, and are as much as 2.5 km from the range-bounding fault. One scarp has a vertical displacement of almost 1 m, but most are less than 60 cm; 10 cm scarps and open ground cracks are common. The 1983 ruptures penetrate at least 80% of the way through the fault segment boundary. Extensive ruptures that have faulted and shattered the bedrock of the segment boundary define a much different tectonic style than the range-bounding faulting that characterizes the Thousand Springs and Mackay segments. The apparent slip deficit of about 2 m during post-glacial times in the segment boundary may represent uplift that occurred as distributed tectonic deformation as opposed to discreet and measurable slip on faults. Which fault-segmentation model best fits the Lost River fault zone? Key aspects include: 1) Earthquake magnitudes appear to be close to the maximum possible, considering the lengths of fault segments bounding the Lost River Range. 2) Displacement patterns are nearly identical for consecutive earthquakes on both the Thousand Springs and Mackay segments. 3) Overlap of surface ruptures from the two segments that is 6 km or less. The characteristic earthquake model developed by Schwartz and Coppersmith (1984) best describes the recent behavior of the Lost River fault. 3.3.5 Strike-Slip Fault Surface Ruptures This brief section compares the Idaho study with similar studies of slip variation along two strike-slip faults in the Mojave Desert of California. By using new approaches to paleoseismol-ogy Vincent was able to assess the variability of true vertical displacement along the Thousand Springs segment of the Lost River Fault. Normal-fault slip ranged greatly from 0.5 to 3 m over a distance of 5 km (Fig. 3.31). A histogram of the 38 measurements (Fig. 3.34A) shows slip amounts clustering about peaks at 1.3 and 2.1 m. An investigator finding such results for a fault zone with no known historical earthquakes might erroneously conclude that the two peaks record two prehistorical events of roughly the same magnitude. 114 Chapter 3 About 85 km of the Mojave Desert was ruptured by a right-lateral slip event during the magnitude Mw 7.3 Landers earthquake of 1992 (Sieh et al., 1993). McGill and Rubin (1999) measured many offset stream channels and vehicle tracks along 5.6 km of the central Emerson fault (Rubin and Sieh, 1997). About 60 measurements were made across the main fault zone, where right-lateral slip ranged from 1.5 to 5.3 m. Locally, variations of slip as large as 1.5 m occurred in only 30 m (Fig. 7 of McGill and Rubin, 1999). The unusual density of offset features in their study area confirmed earlier reports of variable 20 05 Q 10 0 100 200 300 Normal slip, cm 400 _ 3 A - : J 200 400 Right-lateral slip, cm 600 400 500 Left-lateral slip, cm 1200 surface slip along strike-slip faults during earthquakes (Toksoz et al., 1977; Sharp, 1982; Thatcher and Liso-kowski, 1987). Some distributions of horizontal slip along strike-slip faults are nicely unimodal, one example being the surface ruptures resulting from the Hector Mine earthquake of 1999 (Scientists, 2000, Fig. 3). The strongly bimodal distribution of right-lateral offsets along the Emerson fault (Fig. 3.34B) may be due to: 1) Distributed shear and warping in unconsolidated basin fill that is not recognized when measuring an offset. Figure 3.34 Probability density plots constructed as summations of Gaussians representing individual measurements of offset by normal and strike-slipfaults. Thearea under the curve between any two abscissa values describes the relative abundance of geo-morphic features offset by that range of slip. A. Strongly bimodal plot of normal fault offsets along Thousand Springs segment of the Lost Piver, Idaho fault shows that slip varied greatly during the 1953 Borah Peak earthquake. The means of the two statistically significant component peaks of this distribution model best as 130 and 210 cm. Data from Figure 3.26. F3. Strongly bimodal plot of right-lateral offsets along the central Emerson fault, central Mojave Desert of California, shows that slip varied greatly along the surface rupture of the 1992 Landers earthquake. From Figure 10 of McGill and Pubin (1999). C. Strongly bimodal plot of prehistori-cal left-lateral offsets along the Gar-lock fault, northern Mojave Deeert of California, suggests that two prehis-torical earthquakes occurred, each capable of displacing stream channels by 2 to 3 m. From McGill and Sieh (1991). Mountain Fronts 115 2) Bends and offsets that change the style of faulting and cause local uplift and subsidence along strike-slip faults. 3) Part of the slip occurring outside of the main trace, or on subsidiary faults. McGill and Rubin made 200 measurements across secondary fault traces up to 1.7 km from the main fault. Sums of right-lateral slip on these secondary faults were as much as 1.1 m. 4) Replication tests indicate moderately large uncertainties for some measurements. In any case, it seems unlikely that fault slip varied this much at depths of more than 1 km. McGill and Sieh (1991) measured prehistori-cal left-lateral offsets of stream channels along the Garlock fault (Fig. 3.34C). The strongly bimodal nature of the plot suggests that the most recent earthquake had a left lateral slip of 2 to 3 m, and that some landforms were also offset approximately the same amount by the penultimate earthquake. This assumes that variation in slip amounts for each earthquake is described by a unimodal distribution in a probability density plot. Of course McGill recognized that conclusions of two characteristic prehis-torical earthquakes on the Garlock fault were not supported by her later analysis of the 1992 Landers surface rupture. Less than 10% of the Garlock fault displacement measurements were 10 to 18 m, but displacements of this size should be considered as being the result of two or three earthquakes. Surface rupture offsets of landforms are not a flawless way to estimate slip on a fault plane. Vertical displacement of a stream terrace tread or horizontal displacement of a small stream channel is obvious and is nice because we can include the uncertainties of the field measurement in the estimate. But the spatial variations of apparent slip along a normal and strike-slip faults zone (Fig. 3.34A, B) show that caution is needed. Displacements along thrust fault zones are even more variable (Fig. 1.8). Suppose you have measurements for two offset landforms for a fault zone that you are studying. The resulting quandary is that you do not know if a Figure 3.31 type of distribution is present, and where your two data points would fall. Also, surface-slip measurements may not be representative of the magnitude or range of fault slip at depth. The geomorphic procedure to decipher slip distributions for prehistorical earthquakes that are not unimodal (Fig. 3.34C) is to assess the relative ages of the faulted alluvial surfaces. By using a soils chronosequence for Holocene and late Pleistocene faulted surfaces (Bull, 1991, 1996b, 2000), one does not need expensive and time-consuming surface-exposure dating methods. It is best to have at least one alluvial geomorphic surface formed between each earthquake. Situations of insufficient geomorphic surfaces are analogous to stratigraphic sites where earthquakes occur during times of nondeposition. I conclude that both geomorphic and stratigraphic studies may be needed to assess the completeness of an earthquake record. Alluvial surfaces along a given fault zone may be roughly synchronous because adjacent drainage basins tend to have similar aggradation-event response times to climate-change perturbations, and because faulting may create a base-level fall that causes synchronous stream-channel downcutting and preservation of terrace treads. Soils are climate-controlled low temperature geochemical systems, so one can expect major differences in both strength and character of soils developed under glacial and interglacial climates (Bull, 1991). Soils on a 2 ka surface may have only incipient development but should contrast nicely with nearby weakly developed 4 ka soils (Bull, 1996b). Without soils input, the paleoseismologist assumes that approximate multiples of the most recent slip event (such as 3, 6.5, and 9 m) represent repeating characteristic earthquakes (Wallace, 1968; and many subsequent workers). The paleoseismologist with soils information groups her or his slip measurements according to their relative ages in a chronosequence of faulted alluvial surfaces. Then one can proceed to determine if the slip events are characteristic or not. 3.4 Summary Chapter 3 discussions focused on mountain fronts because that is where paleoseismologists should concentrate their assessments of earthquake hazard potential. Geomorphic evolution of mountain fronts is part of the much broader subject of hillslope evolution whose erosion is strongly affected by cumulative base-level fall. Lofty, rugged mountain fronts result from prolonged uplift along either thrust or normal faults, and the fresh or degraded triangular facets at the range fronts are surprisingly similar. Straths form reaches upstream from either style of active range-bounding fault when climatic controls favor lateral beveling after tectonically induced downcutting returns the Ion- 116 Chap gitudinal profile to the base level of erosion between mountain-front uplift events. Renewed valley-floor incision creates strath terraces, and heights of straths above the active channel indicate the amounts of cumulative tectonic base-level change between times of strath formation. Streams that cross a belt of active thrust faults flow through alternating embayment and gorge reaches. Streamflow quickly erases recent surface ruptures in valley floor alluvium, but sub-alluvial fault scarps remain until removed by the longer-term process of valley floor downcutting. Seismic refraction surveys can assess alluvial thicknesses beneath stream channels to determine which thrust faults are active. Stream gradient (SL) indices are markedly different for trunk stream channels with different rates of uplift along their range-bounding faults. Both thrust and normal styles of faulting create mountain front fault scarps. Holocene scarps may not be as obvious for thrust faults because over-thrusting of alluvial materials leads to immediate collapse and slumping. Multiple thrust fault splays create complex scarps whose apparent offset and shapes are tricky to analyze. Erosion of a new thrust fault scarp can locally bury and terminate soil-profile formation on a stream-terrace tread. The elapsed time since rupture of the buried terrace tread is the stratigraphic age minus the time needed to form the post-depositional terrace tread soil profile. Using only ages of stratigraphic layers, as determined by radiocarbon or other methods, will overestimate the age of a surface rupture by an amount equal to the time needed to form the soil profile. A multiple-rupture event fault scarp at the Loma Alta site on the southern front of the San Gabriel Mountains provided unequivocal evidence for increased levels of earthquake hazard for the adjacent Los Angeles metropolis. Scarps of normal faults undergo a more systematic retreat of the surface-rupture generated free face (Chapter 5), and small grabens characterize minor faulting of the hanging-wall block. Apparent throw amounts are best estimated by projecting planar alluvial surfaces across each fault zone. Piedmont deposition is different for these two styles of faulting. Thrust faults are prone to migrate, thus creating a sequence of active mountain fronts ter 3 during the late Quaternary. The newest structural block is the piedmont foreland, or "foreberg". Paleoseismology of piedmont forelands is challenging to the tectonic geomorphologist, but these low subtle scarps are where the action is. It may take ^ 100 ky of gradual transfer of slip from an internal front out to the new range-bounding fault, which eventually becomes a prominent tectonic landform. Thicknesses of basin-fill alluvial fans are much different for active thrust and normal faults, and are a function of net base-level fall. Thrust faulting raises the mountains above the piedmont but this base-level change is reduced by the amount of concurrent stream-channel downcutting. Little piedmont deposition will occur if rapid stream-channel downcutting occurs after uplift along a range-bounding thrust fault. Thrust-fault migration further reduces the potential for piedmont deposition, and the newest fault-bounded block generally is tilted upstream. In contrast, much of the basin subsidence associated with normal fault displacements is available to quickly accumulate impressive thicknesses of basin fill. This may exceed half of the vertical tectonic displacement at the mountain front where tectonically induced stream-channel downcutting absorbs part of the base-level change. The proportion of base-level change that induces basin-fill accumulation is greatest where high rock mass strength slows the rate of channel downcutting at the mountain front. The premise of the segmentation model of fault behavior is that each section of a range-bounding fault tends to rupture repeatedly in similar-size (characteristic) events that approximate the maximum possible magnitude earthquake. The characteristic earthquake model (Schwartz and Coppersmith, 1984) best describes the behavior of the Lost River fault in Idaho. Kirk Vincent's study is the first field verification of this conceptual model. His corrections to amounts of apparent throw generally are minor to moderate, but are large where fan slopes are steep. His accurate dataset of variation of true throw on the range-bounding faults also explains apparent anomalous behavior in the boundary between two segments of the Lost River fault. The segment boundary is characterized by broadly distributed faulting which contrasts with simple range-front faults of both adjacent fault segments. The broad boundary between the Pacific and North American plates has domains of mountain fronts with different styles and rates of tectonic deformation. Mapping the geomorphic tectonic-activity characteristics of mountain fronts in a region is valuable for understanding regional tectonics or assessing earthquake hazards. Chapter 4 describes how vertical tectonic perturbations result in distinctive mountain-front characteristics. Different rates of base-level fall influence landforms as small as valley floors. Diagnostic tectonic landforms are described and used to define five classes of relative uplift. Regional maps depicting relative tectonic activity of associated groups of mountain fronts are a useful reconnaissance tool for tectonic geomorphologists and paleoseismologists. 4.1 Tectonic Setting of the North America-Pacific Plate Boundary The onshore part of the San Andreas transform boundary between the North America and Pacific plates in the southwestern United States is a 200-800 km wide transition zone extending from the Pacific Ocean into the Basin and Range Province. Two features have controlled many secondary tectonic structures of the transition zone. The San Andreas fault presently is the primary plate-boundary fault zone. This right-lateral continental transform fault slices through batholithic complexes to create the Peninsular and Transverse Ranges of southern California, passes through the Coast Ranges, and turns out to sea at Cape Mendocino in northern California to join the Mendocino fracture zone (Fig. 4.1). The Sierra Nevada microplate is equally important. The 650 km long Sierra Nevada was created by batholithic intrusions in the Mesozoic. The mountain range is huge but the microplate is immense because it also includes the adjacent Central Valley of California (Fig. 4.1). This tectonic block has undergone minimal internal deformation but the eastern side of the Sierra Nevada was raised recently in dramatic fashion. An impressive escarpment rises 1,000 m in the north and 2,000 m in the south. The microplate was much larger just 5 My ago. Its batholithic rocks extend east into the Basin and Range Province whose extensional style of faulting continues to encroach into the eastern side of the microplate. A major tectonic event - detachment of the Sierra Nevada batholithic root - referred to as de-lamination affected other tectonic elements in much of the broad transform boundary. We should note apparent coincidences of timing of many important Neogene tectonic events in the boundary between the Pacific and North American plates. The Cretaceous batholith had a thick residual root, but an Airy-type crustal root is no longer present to support the Sierra Nevada. Xeno- Tectonically active and inactive mountain fronts of the northeastern Mojave Desert, California. From U.S. Air Force U2 photo supplied courtesy of Malcolm Clark, U.S. Geological Survey. 118 Chapter 4 Study areas 1 San Gabriel Mountains 2 Mojave block 3 Death Valley region 4 Central Nevada seismic belt 5 Sierra Nevada microplate 6 Sacramento Valley 7 San Joaquin Valley & east-central Diablo Range 85 Los Anae\e Figure 4.1 Tectonic setting of southwestern United States. The Sierra Nevada microplate was wider at 5 Ma because it extended west almost to the San Andreas fault, and extended east Into the Basin and Range Yrovlnce. Area within the solid line outlines the area of accelerated extenslonal faulting since 5 Ma (Jones et al., 2004, Figure 1). This area and the area Inside the dashed line approximate the province of the Walker Lane-Eastern California shear zone, that has accommodated ~10% of the plate-boundary dextral shear since 5 Ma. lith composition changes (Ducea and Saleeby, 1996, 1998) indicate that a former dense root beneath the Sierra Nevada crest was convectively removed before 3 Ma. Mantle lithosphere is now abnormally thin beneath the Sierra Nevada and Panamint Ranges (Jones et al., 1994; Wernicke et al., 1996; Ruppert et al., 1998; Saleeby et al., 2003; Saleeby, and Foster, 2004; Zandt, et al.2004). The batholithic remnants seem to be largely supported by a buoyant upwell-ing of hot asthenosphere. Zandt (2003) refers to this flow as "mantle wind" that shifted this part of the detached Sierra Nevada batholithic root - a mantle drip - to the southwest. The detached batholithic root sank rapidly to the base of the athenosphere and it is the cold down-welling trail of the drip that is revealed by seismic tomography (Zandt, 2003). It appears that southwest directed mantle flow shifted the post-4 Ma Sierra Nevada mantle drip so it is now beneath the eastern San Joaquin Valley and the foothills of the Sierra Nevada (Zandt, 2003). Xenolith studies suggest recent removal of garnet bearing rocks triggered a brief (3.5 ± 0.25 Ma) Tectonic Activity Classes pulse of potassium-rich basaltic vulcanism in a 200 km diameter circular area in the central Sierra Nevada (Manley et al., 2000; Farmer et al., 2002; Jones, et al.2004). This event indicates the most likely timing and location of the main delamination event. Zandt (2003) points out that the highest peaks of the Sierra Nevada occur in the area where the delamination event began; the maximum strength of the geophysical perturbation coincides with maximum uplift. I like their interpretation that recent uplift of the Sierra Nevada also could be this young, and that both volcanism and uplift have a common delamination triggering mechanism. The interval between 5 Ma and 3 Ma fits a model of regional synchronous changes in many plate-boundary tectonic processes, and of landscapes that record changes in tectonic base-level controls. Recognition of the lack of a crustal root beneath the now lofty Sierra Nevada has led to much creative thinking about tectonics of the region. The brilliant synthesis by Jones et al. (2004) recognizes the late Pliocene foundering of the Sierra Nevada crustal root as a most important perturbation that changed regional plate tectonics. They reason that this crustal delamination event increased the total gravitational potential energy of the lithosphere (Jones et al., 1996), thus increasing both extensional strain rates in the western Basin and Range Province and the altitude of the eastern Sierra Nevada. They also conclude "an increase in extensional displacement rates must be accommodated by a decrease in rates of extension or an increase in rates of shortening somewhere in the vicinity of the Sierra Nevada" (p. 1411). The timing of recent thrust faulting that created the Coast Ranges bordering the Central Valley coincides nicely. And, "Lithospheric removal may also be responsible for shifting of the distribution of transform slip from the San Andreas Fault system to the Eastern California shear zone" (p. 1408). The tectonics and landscape dynamics of the southwestern corner of the Basin and Range province were much different after introduction of large amounts of right-lateral slip (Dokka and Travis, 1900a, b; Burchfiel et al., 1995; Burchfiel and Stewart, 1996; Lee et al., 2001). Important fault zones now may run through the valleys as well as along rugged mountain fronts. The more prominent strike-slip faults are shown in Figure 4.2. Cenozoic tectonism of western North America is related to the Mendocino plate boundary triple junction, which migrated northward creating the San of Mountain Fronts \\ 9 Andreas - a continental right-lateral transform fault system (Atwater, 1970; Atwater and Stock, 1998) that presently includes the Maacama and Bartlett Springs faults. They conclude that no discernible change in rates of motion of the Pacific—North American plate boundary has occurred since 8 Ma. This is a key assumption of the Jones et al. model where horizontal velocities across the Sierra before and after the delamination event match the boundary condition of Pacific-North American plate motion. Local acceleration of extensional encroachment in a belt east of the Sierra Nevada may match increased rates of Coast Ranges shortening to the west. The Sierra Nevada microplate has become narrower as a result of both processes. Several tectonic events have a timing that is coincident with or shortly following delamination. Crustal extension accelerated into the eastern margin of the Sierra Nevada, thus making the Basin and Range Province ever broader. See Figure 4.1. Part of the San Andreas fault style of dextral shear split off to create the seemingly diffuse Walker Lane-Eastern California shear zone in eastern California and western Nevada. Compression near the San Andreas fault created a fold-and-thrust belt that is encroaching northeastward into the microplate. The result of this increased rate of crustal shortening has been a regional reversal of sediment-transport direction, with thick Diablo Range alluvial-fan deposits being laid down over basin fill derived from the Sierra Nevada. The present eastern edge of the Coast Ranges was not formed by a synchronous pulse of uplift, as was the eastern front of the Sierra Nevada. A southeast to northwest migration of thrust-faulted mountain fronts profoundly influenced landscape evolution and has continued to the present. 4.2 Appraisal of Regional Mountain Front Tectonic Activity 4.2.1 Geomorphic Tools For Describing Relative Uplift Rates The above brief summary of Neogene plate-boundary tectonics suggests an intriguing variety of mountain fronts to be studied by tectonic geomorpholo-gists. These include Basin and Range Province normal faulting, Mojave Desert transtensional faulting, 120 Chapter 4 1 1 7°36.0' Figure 4.2 Important Neogene strike-slip faulting in the southwest corner of the Basin and Range Province. Mountains are gray areas and Intervening valleys are named. The Garlock and associated sinestral faults provide necessary block rotation to Impart north-south continuity of the Walker Lane-Eastern California shear zone (Dickinson, 1996). The Death Valley fault system has a releasing right step that results in rapidly subsiding pull-apart basin (upper-right inset map). Owens Valley fault is here included with the broad shear couple proposed by Monastero et al., (2005) who consider the resulting Coso Range volcanic complex to be a nascent metamorphlc core complex. The sharp releasing bend in Saline Valley is responsible for the impressive mountain front shown in Figure 2.20A. Panamint Valley has active strike-slip faults on both sides of the valley. Tectonic Activity Classes Transverse Ranges thrust faulting, and folding in the margins of the Coast Ranges. Long escarpments consist of a sequence of mountain fronts. Adjacent fronts with contrasting topography, structure, or rock type may represent fault-zone segments with different styles, rate, and magnitudes of displacement. I use landform assemblages to define and map classes of relative uplift for the mountain fronts in the broad region east of the San Andreas fault. Mountain-front tectonic activity ranges from vigorous to dormant. Dormant landscapes occur next to rapidly rising mountain fronts. Much has changed since Bull and McFadden introduced this approach to tectonic geomorphology in 1977. Their simple introduction to landscape responses to mountain-front faulting took advantage of obvious differences between rugged mountain fronts that could be designated as belonging in two active and one inactive class of tectonic activity. We now recognize the presence and importance of the Walker Lane-Eastern California shear zone. Strike-slip faulting may create fault scarps that are only a few meters high that commonly are buried by Holocene alluvial-fan deposits. Such highly active "mountain fronts" pose high risk from a paleoseismic perspective and are central in the assessments presented here. A single working model cannot encompass all styles and sizes of tectonic deformation without becoming unwieldy. The approach here, as previously, will be to emphasize large tectonic displacements. The relevant time span is the Pleistocene because magnitudes of Holocene tectonic deformation do little to change the character of a mountain front. I emphasize landforms that vary greatly with different rates of tectonic base-level fall. This approach defines broad classes of surface-uplift rates and de-emphasizes the influences of complicating factors. Both erosional and depositional landforms of fluvial systems reflect rates of bedrock uplift as affected by geomorphic processes (Fig. 1.4). Slow erosion preserves tectonic landforms longer in arid than in humid terrains, which can make mountain ranges of arid regions appear to be rising faster than their humid region counterparts. Abrupt local tectonic base-level fall associated with range-bounding faults results in more dramatic landscape contrasts than uplift associated with broad wavelength folding. So, the faulted landscapes of the Walker Lane-Eastern California shear zone are used to introduce and illustrate how to define tectonic activity classes. These mountain fronts range from lofty to low scarps. of Mountain Fronts \2l We start with numerical descriptions of key tectonic landforms. Then interactions of local base-level processes are used to define classes of relative tectonic activity that are based on definitive landform assemblages. The output consists of maps depicting areal variations of Quaternary uplift of mountain fronts that can be used by earth scientists and engineers. Paleoseismologists use regional assessment of mountain front tectonic activity in reconnaissance investigations. The emphasis is not always focused on structurally complex and fascinating zones of rapid tectonic deformation. Nuclear-power generating plants should be located in the most tectonically inactive sites. Obvious contrasts in the morphology of landforms in different tectonic settings may be described by simple ratios of topographic lengths and heights. Ratios provide dimensionless numerical indices for describing the states of tectonically induced downcutting of valley floors, and erosional retreat of hillslopes that are useful for defining classes of tectonic activity. Landscape ratios vary in their sensitivity to describe the influences of late Quaternary tectonic deformation. Those that emphasize changes in valley-floor width are good for describing tectonic uplift of the past 10 to 100 ky in arid and semiarid regions and 1 to 30 ky in extremely humid mountains. Parameters that emphasize changes in relief are good for describing drainage-basin evolution over time spans of more than 500 ky in the semiarid American West. Pedimentation operates on an even longer time scale. A kilometer of retreat of a mountain front generally requires more than 1,000 ky. Two ratios have been consistently useful for identification of tectonically active and inactive mountain fronts. These are the sinuosity of the mountain-piedmont junction and the valley floor width-valley height ratio. Frankel and Pazzaglia (2005, 2006) prefer to examine overall landscape characteristics using digital elevation models that assess the average depths of valleys expressed as a drainage basin volume-drainage basin area ratio. Such metrics and gradients of first-order stream channels (Merritts and Vincent, 1989) do indeed describe significant differences between tectonically active and inactive landscapes. They may be especially attractive for inspection of watersheds deformed by active folding with large wavelengths. However, the use of simple ratios (Bull and McFadden, 1977) may still be superior for studies of faulted mountain fronts. Tectonic base-level fall emanates from the mountain-piedmont junction so 122 Chap a case can be made to emphasize landform metrics in the immediate vicinity of the tectonic perturbation, including triangular facets, valley floor width-valley height ratios, and mountain-piedmont junction sinuosities. Upstream propagation rates for range-front base-level falls might be best explored through studies of the longitudinal profiles of trunk stream channels that are the connecting link between the mountain-front tectonic perturbation and slowly evolving streams and hills of headwater reaches. 4.2.1.1 Mountain-Front Sinuosity The straight or gently curving nature of most faults or folds allows evaluation of the degree of erosional modification of a structural landform. Rapid uplift along a range-bounding fault maintains the linear nature of the front. Erosion dominates landscape evolution after cessation of uplift and creates a sinuous mountain-piedmont junction, especially where lithologic resistance to erosion is weak. Intermediate scenarios involve the interplay of ongoing uplift and continuing fluvial degradation, which varies greatly with climatic setting. ter 4 Streamflow becomes the dominant process shaping the mountain-front landscape in tectonically quiescent settings. Streams downcut quickly to their base level of erosion by removing small amounts of rock, and then slowly widen their valley floors by removing large amounts of detritus derived from hill-slopes. Maximal concentrations of stream power at canyon mouths result in erosional embayments that extend up the larger valleys. The result is a highly sinuous mountain-piedmont junction. Relatively slow uplift may be continuing but fluvial erosion is dominant over uplift. Sections of a mountain front between the principal watersheds are eroded at a much slower rate, which minimizes development of embayments. The sinuosity index used here (Fig. 4.3) is measured in the same way as for meandering rivers. Sinuosity of the mountain-piedmont junction, J, is the ratio of the planimetric length of the topographic junction between the mountains and the adjacent piedmont, L, to the length of the range-bounding geologic structure, Ls. j-h. (4.D J L Figure 4.3 Sinuosity of mountain-piedmont junctions. A. Tectonically active mountain front associated with an oblique right-lateral fault in the central Mojave Desert of California. The length of the thick, straight white line Ls, is the length of the range-bounding fault. The thin, sinuous white line, L , is along the mountain-piedmont junction. Its sinuosity records embayments at mouths of watersheds, and minor departures from fault-zone linearity. Single-lane dirt track for scale. Tectonically translocated fan aggradation event surfaces have been offset in a right-lateral sense from their source watershed. Ages of surfaces are summarized in Table 1.2. Tectonic Activity Classes of Mountain Fronts 123 Figure 4.3 Sinuosity of mountain-piedmont junctions. F3. Tectonically inactive mountain front. Advanced stage of erosion leaves few clues as to the actual location of the now dormant range-bounding fault, so endpoints are arbitrary. The thin, sinuous white line, L, Includes distances around detached parts of the mountain front that are now inselbergs. J commonly is less than 3, and approaches the minimum value of 1.0 where steep mountains are being raised rapidly along a range-bounding fault or fold. Sinuosity can also be expressed as departures from a value of 1.0, or as departures from mean values for a specific tectonic province. Consistent procedures should be used when dealing with three potential complications. Wide alluvial valley floors of large watersheds may extend far upstream. To trace the mountain-piedmont junction most of the distance to the watershed divide would produce an unnecessarily large value of J. When tracing such mountain-piedmont junctions, the operator should go only a constant distance up such valleys, such as 0.5 or 1.0 km, or a basin-position coordinate of 0.95, before crossing to the other side of the valley floor and returning to the outer mountain front. Doing this emphasizes the mountain front instead of the sides of anomalously large valleys. The straight-line distance between two points on the mountain-piedmont junction should be used as the reference distance where the location of the range-bounding fault is not known. Parts of former spur ridges become isolated from the range as valley floors widen and the mountain front retreats. These erosional outliers are termed inselbergs. For pedimented landscapes, the mountain-piedmont junction, L, is measured as the sum of the perimeters around inselbergs plus the length of the mountain-piedmont junction (Fig. 4.3B). Images may be better than maps for measuring range-front sinuosities. The piedmont is smooth in the Figure 4.3A image, except for two areas of dissected terrain. Measurements made on a topographic map would incorrectly suggest that these are part of the overall erosional terrain and should be included with the mountains. Such remnants of old alluvial-fan deposits may have considerable relief but are not included in the measurements because they are basin fill, not mountains. Inclusion of such basin fill would raise the sinuosity value and place this front in a class of relatively less tectonic activity. Alluvial fanheads extend far up the embay-ments shown in Figure 4.3B. Mountain-front length focuses on the bedrock-alluvium contact, including remnants that are now isolated from the mountains. Once again, smooth remnants of Q2 age dissected alluvial fans are not included as part of the mountain- 124 Chap piedmont junction. This distinction can be made on this image, but not on topographic maps. A variety of topographic information sources can be used to map mountain-piedmont junctions. Replication measurements are precise when the mountain-piedmont junction is viewed on images larger than 1:60,000. Landsat is borderline, but SRTM radar images can be ideal (Fig. 4.3C). Radar images focus on landscape roughness and piedmonts adjacent to a mountain front commonly show as smooth, nearly featureless plains. Not having the clutter of towns and fields makes radar images especially nice for studying mountain fronts. Topographic maps with a scale of ^ 1:25,000 and a 5 to 10-m contour interval yield good results, but 1:250,000 scale maps with 30 m contour intervals are inadequate for estimating the location of the mountain-piedmont junction. Meticulous plane-table surveys of old surveys may provide excellent maps for geomorphic evaluations of mountain-piedmont junctions. The Figure 4.3D example uses a 1.5 m contour interval that provides detailed information. Four domains of contour characteristics are obvious, but domain 4 is where the contour interval changes from 5 feet to 25 feet (1.52 to 7.62 m). The smooth depositional surfaces of the alluvial fans (domain 1) contrast nicely with the erosional topography of domain 2. Domain 3 is also underlain by tectonically deformed Pleistocene unconsolidated deposits but has much more local relief. The sinuous nature of ter 4 the internal mountain front suggests the presence of a second thrust fault. The range-bounding front may also record thrust faulting but is not nearly as sinuous because it is in part a function of deposition of adjacent piedmont alluvium. Different uplift rates (for a given climate and lithology) have distinctive ranges of sinuosity. Sinuosities of highly active mountain fronts generally range from 1.0 to 1.5, moderately active fronts range from 1.5 to 3, and inactive fronts from 3 to more than 10. Sinuosities greater than 3 describe highly embayed fronts. The range-bounding fault along which the mountains were raised may be more than 1 km downslope from the mountain—piedmont junction (Section 4.2.2). Edges of folded mountain ranges whose strata strike parallel to the front may retreat with minimal embayment, and have anomalously low sinuosity values of J. Changes in mountain front sinuosity typically require long time spans, because geomorphic processes change hillslopes fairly slowly. Nonetheless the index complements another index that changes at a faster pace because concentration stream power in valley floors can accelerate local landscape evolution. 4.2.1.2 Widths of Valleys Another sensitive index to recent and ongoing uplift is the valley floor width-valley height ratio, or for brevity the V^ratio (Fig. 4.4A). If Vfw is the Figure 4.3 Sinuosity of mountain-piedmont junctions. C. Radar image of the southern end of the North Island of New Zealand. Mountain front of the active Wellington fault (1) has a sinuosity of 1.0. Rimutaka Range front-ANalrarapa fault (2) has a sinuosity of 1.1 and ruptured in the magnitude Mw Ö.2 earthquake of 1Ö55. This figure is from NASA/JPL/ NIMA image PIA02742 from the Shuttle Radar Topography Mission. Tectonic Activity Classes of Mountain Fronts 125 Figure 4.3 Sinuosity of mountain-piedmont junctions. D. Example of wee of detailed topographic mapping for study of a mountain-piedmont junction. From 1926 plane-table mapping of the Levis Quadrangle on the folded eastern margin of the central California Coast Ranges. Contour Interval of 5 feet (1.5 m) changes to 25 feet in the southwest corner of the map. width of a valley floor and Ald, Ard, and Asc are the altitudes of the left and right divides (looking downstream) and the altitude of the stream channel, then f (Ald-Aj + {Ard-Aj (42) 2 Valley-floor widths increase with watershed size, erodibility of rock type, and with decrease of uplift rate. Valley heights decrease with the passage of time after cessation of uplift, but not nearly as fast as valleys widen. The valley floor width-valley height ratio is especially sensitive to late Quaternary tectonic base-level falls because narrowing of a valley floor is accomplished quickly by the downcutting action of streams. Bull and McFadden (1977) found significant differences (at the 0.99 confidence level) in the means of V. ratios of tectonically active and inactive mountain fronts. Care is needed in selecting sites to measure valley floor width-valley height ratios. Vf values are more likely to be representative of the relative degree of tectonic base-level fall if determined in similar rock types and at the same basin-position coordinate for a suite of similar size drainage basins along a given mountain front. Rock resistance to erosion may not change, but annual unit stream power to do the work of erosion increases downstream. Consequently, part of the variation in valley-floor width is a function of drainage-basin size. A limited range of drainage-basin sizes is preferred because stream discharge increases in a nonlinear manner in the downstream direction, particularly for streams of humid regions (Wolman and Gerson, 1978). Measuring valley floor widths can be tricky. Ideally they should be the mean value of Distance, m Figure 4.4 Topographic cross sections that illustrate valley-morphology definitions. Starvation Canyon, Yanamlnt Range of California. A. Valley floor width-valley height ratio, Vf. 126 Chap several measurements made in the field. High quality topographic maps succeed fairly well. Digital sources give erratic results if poor resolution integrates valley floor and footslope relief, thus failing to detect abrupt margins common to many valley floors. Determine Vf ratios at basin-position coordinates of 0.9, or at a constant distance upstream from the mountain—piedmont junction (for example 1 km). Avoid a basin-position coordinate of 1.0, especially if it coincides with intensely sheared bedrock of a range-bounding fault. This local decrease of rock mass strength is a complication that should be minimized. The mouths of canyons also are the most likely sites for climate-change induced aggradation to greatly widen a valley floor that may be subject to continuing degradation in upstream reaches. One should void mixing data from degrading reaches with reaches whose valley-floor width records episodes of aggradation. The statistically different widths of valley floors for tectonically active and inactive mountain ranges record much different stream-channel processes. The much larger 1/ ratios of tectonically inactive watersheds record widening of bedrock valley floors (type 1 dynamic equilibrium of Section 2.4.3). Such watersheds also have pervasive fill stream terraces created by climate-change induced aggradation events (Bull, 1991). The edges of these valley floors are where valley-floor alluvium meets the footslope. Valley shape landforms can be selected for a particular time span of interest. The Vf ratio can be used to detect rapid uplift during the Holocene in humid regions and uplift since the middle to late Pleistocene in arid regions. Longer response times are required to change the cross-sectional dimensions of a valley. Valley cross-section ratios (Mayer, 1986), and mountain front sinuosity, would be appropriate for semiarid regions studies of late Cenozoic tectonics and topography. Mayer (1986) uses a numerical index of valley morphology, V , that compares the area of a valley cross section, A , with the area of a semicircle whose radius is equal to the height of the lowest adjacent drainage divide, Ac (Fig. 4.4B): ter 4 width, V, to valley height - for brevity, the V ratio - may be useful for describing long-term decreases of relief for various lithologies and climates. V = A _v A, (4.3) The Vc index is intermediate in sensitivity for describing the effects of tectonically induced downcut-ting compared to the Vf and V ratios. Ratios of valley V 2V (A —A ) + {A -A ) v Id sc' v rd scJ (4.4) The V ratio can be used to discriminate between highly active and inactive mountain fronts but is not as useful for distinguishing intermediate classes of tectonic activity. V ratios of 4 to 6 are representative of rising mountains. Inactive mountain fronts generally have values of V greater than 7. Vf ratios are more sensitive to fluvial base-level controls than V ratios because changes in the widths of valley floors occur much more rapidly than hillslope relief is reduced. Valleys may be incised in less than 10 ky in response to a base-level fall. Valley-floor narrowing typically is more than the concurrent increase of valley cross-section relief. Then valley-floor width may double in 100 ky, but >1 My may be needed to reduce valley height by half at the same position in a drainage basin. Highly active mountain fronts commonly have V ratios between 0.5 and 0.05. Strath terraces are absent in such narrow canyons if downcutting does not pause (times of attainment of the base level of erosion). The miniscule widths of such valley floors, relative to the adjacent watershed ridgecrest heights, result in Vf ratios that clearly define situations where tectonically induced downcutting maintains the mode of stream operation far to the degradational side to the threshold of critical power. Moderately to slightly active valleys have flights of strath terraces that record pauses in stream-channel downcutting. Inactive valleys may lack 1000 £ 600 500 WOO Distance, m Figure 4.4 Topographic cross sections that illustrate valley-morphology definitions. Starvation Canyon, Yanamlnt Range of California. F3. V ratio. A Is the cross-sectional area of the c v valley. Ac is the area of a semicircle whose radius is equal to the height of the lowest adjacent drainage divide. Tectonic Activity Classes strath terraces, but only if they remain at the base level of erosion. The presence of a strath terrace requires not only beveling during a period of attainment of the base level of erosion but also subsequent renewed degradation resulting from climate change or response to long-term isostatic uplift. Isostatic uplift resulting from millions of years of denudation of a tectonically inactive mountain range may eventually favor an episode of renewed stream-channel down-cutting that leaves pediments and associated valley floors as strath terraces. Such infrequent episodes of strath-terrace formation probably coincide with climate-change induced changes of bedload transport rate and stream discharge that shift the mode of operation to the degradational side of the threshold of critical power. 4.2.1.3 Triangular Facets Daunting challenges face paleoseismologists when asked to assess earthquake hazards of vast regions. "Please identify every surface rupture younger than 10 ky in 50,000 km2 of the Basin and Range Province". "Tell me which mountain ranges are rising faster than 0.1 m/ky". Seemingly not possible but these requests are actually quite straightforward for tectonic geomorphologists. The key to efficiently doing such work is to use simple techniques that quickly discern obvious differences between informative landscape parameters. Holocene surface-rupture identification is easy. Pleistocene-Holocene climatic changes affected geomorphic processes profoundly in the American southwest, causing piedmont aggradation events (Bull, 1991, Chapter 2). Pedogenic processes and the resulting soils were as different as the global contrast between ice-age and present climates. So the solution is simply "Does the surface rupture disrupt an alluvial fan or stream terrace that has a Holocene soil profile?" Soils geomorphology provides the answer without expensive, lengthy laboratory analyses or modeling. In California the Alquist-Priolo Earthquake Fault Zoning Act (Hart and Bryant, 1997) defines an active fault as one that has "had surface displacement within Holocene time" and no structure for human occupancy is permitted within 50 feet of an active fault. Heights and stages of dissection of triangular facets (Section 3.2.1, Figure 3.5, Table 3.1) are indicative of relative tectonic activity (Bull and McFadden, 1977). Basal sections of triangular facets may resem- of Mountain Fronts \2J ble degraded fault planes (Hamblin, 1976; Menges, 1990b; Ellis et al., 1999). Obvious landscape contrasts in the Great Basin of west-central Nevada were used by dePolo and Anderson (2000) to estimate slip rates for hundreds of normal faults. Rapidly rising mountain fronts have 1) fault scarps on the piedmont and at the mountain-piedmont junction and 2) high triangular facets. Tectonically inactive mountain fronts have neither. Their emphasis was on normal faults because tectonic increase of relief provides the potential energy to erode fault blocks into drainage basins whose characteristics reflect the rates and magnitudes of vertical displacement. Active plate tectonics extensional processes create distinctive fluvial systems. Pure strike-slip faulting is locally important but does little to change hydraulic gradients of streams. Instead strike-slip faulting disrupts and tears apart the fluvial systems created by base-level fall. The dePolo and Anderson dataset of 45 mountain fronts provided three classes of estimates of normal-fault slip rates (dePolo, 1998). Type 3 mountain fronts have slip rates of -0.001 m/ky, which is so slow that these landscapes lack fault scarps and triangular facets. Type 1 mountain fronts have both fault scarps and triangular facets with minimal dissection that rise more than 30 m above the mountain-piedmont junction. All but one Type 1 mountain front has a Holocene surface rupture. Type 2 normal faults Type 3 Type 2 No scarps Alluvial scarps No facets No facets Typel Facets above active fault Alluvial scarps 1.001 0.005 0.01 0.05 0.1 0.5 Vertical slip rate, m/ky Figure 4.5 Comparison of vertical slip rates for normal faults associated with three types of landscape. Dark boxes are Type 1 faults, white boxes are Type 2 faults, and the two light gray boxes are Type 3 faults. Figure 3 of dePolo and Anderson (2000). 128 Chapter 4 600 ■51 § E Sf TO -a -P E o => .to TO 400 200 ~i-1-1-1-1-1-1-r R2 is 0.63 1c is 0.239 or ± a multiplicative factor of 1.7 in slip rate _l_ _l_ _l_ _l_ _l_ 0 1 2 3 4 5 Vertical slip rate, m/ky Figure 4.6 General relation between maximum height of basal triangular facets and vertical slip rate. Figure 6 of dePolo and Anderson (2000). 5^ is vertical slip rate (m/ky) and H is maximum basal facet height in meters. have alluvial fault scarps, but lack active basal sets of facets. Relict triangular facets may be present. Comparison of these simple geomorphic characteristics with known normal fault slip rates was internally consistent (Fig. 4.5). The dePolo-Ander-son method highlights Type 1 landscapes with triangular facets indicative of those range-bounding faults most likely to rupture next. They focused on the heights of active triangular facets, which have undergone minimal dissection in the arid Great Basin of west-central Nevada. Triangular facet height for the most rapidly rising Type 1 mountain fronts increases systematically with increasing rate of uplift (Fig. 4.6). dePolo and Anderson assessed the relative tectonic activity of normal faults for much of the state of Nevada. The most rapid normal fault slip rates occur in the Walker Lane-Eastern California shear zone, perhaps because of recent encroachment of Basin-and-Range extension into the Sierra Nevada microplate, and/or because of local pull-apart basins created by lateral fault displacements. 4.2.2 Diagnostic Landscape Classes of Relative Tectonic Activity Local base-level processes change streambed altitudes relative to adjacent reaches. Four base-level processes occur at mountain fronts of arid and humid regions. These are the dependent variables of channel downcutting (cd) in the mountains, aggradation (pa) or degradation (pd) of the piedmont downslope from the escarpment, and the independent variable of uplift (u) of the mountains relative to the adjacent basin. Local base-level processes affect erosion of valley floors and hillslopes in mountains, and deposition and erosion that create piedmont landscapes. Geomorphic processes on alluvial fans and pediments are discussed here in the context of base-level changes and tectonic activity classes (Bull, 1984). The affects of rapid uplift of mountains relative to an adjacent basin result in a unique landform assemblage in either arid or humid regions. Consider the accumulation of thick alluvial-fan deposits adjacent to mountains where streams are flowing on bedrock (Figs. 4.7A, 4.8A). The stream channel will tend to become entrenched into the fanhead as valley floor degradation continues to lower the stream channel in the mountains. The resulting downstream shifts of the threshold-intersection point are promoted by either channel downcutting in the mountains and/or by fan aggradation on the piedmont. Uplift of the mountains along a range-bounding fault or fold counteracts this tendency to entrench the fan-head. Tectonic elevation promotes sustained channel downcutting in the mountains and piedmont fan deposition, but only when the uplift rate equals or exceeds the sum of the two local base-level processes that tend to cause fanhead trenching. Au Acd Apd At At > At (4.5) A tectonically active landscape assemblage may be defined in terms of the three interacting base-level processes of equation 4.5. Unentrenched alluvial fans are present immediately downstream from mountain valleys that have only a veneer of alluvium on narrow bedrock floors. These are base-level interrelations between geomorphic processes, so equation 4.5 is not generally meant to be used in situations of great thicknesses of basin fill. Basin subsidence below sea level is indeed part of mountain front tectonic deformation, but rarely can be related to present streamflow dynamics. Of course one can change the uplift term from rock uplift (Au/Af) to a measure of tectonic deformation, "Ltd, in order to accommodate stratigraphic information and a greater spatial vertical scale of mountain-front deformation over longer time spans than usually intended by equation 4.5. Tectonic Activity Classes of Mountain Fronts 129 Class 1 3. Class 3 A m < A cd . At At > At Fan »«•r *.«'««*^ »,,i'«.'> »,*■•.'> f »i^t . «* »1.; • ? «; Class 5E3 Pediment Au^Acd &pd TIP- Truncated Alluvial Fan Figure 4.7 Longitudinal profiles of diagrammatic fluvial systems showing the landforms associated with different relative uplift rates. TIP is threshold-intersection point. A. Alluvial-fan deposition continues next to the mountains where uplift is the dominant base-level process. F3. The apex of fan deposition shifts down fan where stream-channel downcutting is the dominant base-level process. C. Erosion is the dominant base-level process in the mountains and on the piedmont In a tectonically inactive setting. Landscape assemblages defined by equation 4.5 are characterized by straight mountain fronts that coincide with range-bounding faults or folds, triangular facets whose younger (basal) portions have undergone minimal dissection, and V-shaped cross-valley profiles with straight or convex footslopes. This suite of characteristic tectonic landforms defines a class of relative uplift (class 1) of the mountain front regardless of the prevailing climate and rock types, and with no need to determine the rate of uplift or the time at which it began. Climate and erodibility of materials greatly affect rates of erosion (Fig. 2.19) and deposition but classes of relative tectonic activity are based on relative rates of uplift, erosion, and deposition. A low-altitude panorama of the arid western front of the Panamint Range (Fig. 4.9) shows most geomorphic features indicative of rapid base-level fall. A typical abrupt switch between the erosional and depositional subsystems is present. This coincides with the range-bounding Panamint Valley fault 130 Chapter 4 Figure 4.5 Block diagrams contrasting the landscape assemblages associated with active and inactive normal faults. Buried stream channels record episodes of climate-change induced stream-channel downcutting on an aggrading piedmont. Drawings by Bus Royce. A. Class 1 landscape where rapid tectonic displacement rates are partitioned between creating space for continuing aggradation of alluvial-fan deposits and increasing relief of mountains with narrow valley floors and triangular facets along a straight mountain-piedmont junction. B. Class 5 landscape where cessation of normal faulting allows fluvial systems to achieve a long-term base level of erosion recorded by a single surface composed of straths in mountain valley floors, beveled pediment surface with inselbergs rising above it, and truncated basin fill. Broad valley floors extend far upstream from a sinuous mountain-piedmont junction lacking triangular facets. zone, which here is a fairly narrow high-angle oblique normal fault. Rapid deposition of gravel by braided distributary streams has created unentrenched fan surfaces of Holocene age. Base-level rise resulting from this piedmont aggradation has promoted alluvial deposition over the fault zone. Subsequent surface ruptures create piedmont fault scarps. Stream-channel downcutting induced by episodes of tectonic base-level fall preserves several ages of late Quaternary fan surfaces on the edge of the footwall block. Some of these remnants may be preserved as they are raised, as if on an escalator, into the rising Panamint Range. A Saline Valley example is shown at the left side of Figure 2.20A just above the level of the waterfall. The fine textured (closely-spaced) drainage density shown in Figure 4.9 suggests the presence of soft rocks. Cichanski (2000) mapped this part of the range flank asal5°to35° low-angle normal fault that became inactive when the range-bounding transten-sional Panamint Valley fault zone became active. See the Figure 1.7 discussion about how changing from low- to high-angle faulting changed fan deposition. Tectonic Activity Classes of Mountain Fronts 131 shorelines Figure 4.9 Aerial view of a class 1A mountain front at Bighorn Canyon, west side of the Panamint Range, southeastern California. Landforms indicative of vapid uplift during the late Quaternary include the presence of highly elongate drainage basins, very low valley floor width-valley height ratios, a straight mountain front, stage 1 triangular facets, undissected alluvial fans, and multiple ages of alluvial surfaces separated by fault scarps of different heights. Four faulted fan surfaces range in age from late Pleistocene (1) to late Holocene (4). A remnant of late Pleistocene lake deposits and shoreline ridges is at lower right. Two-lane dirt road for scale at lower left. The width of this view is about 1.9 km. Parts of the low-angle normal fault are capped with a gently dipping, loose to locally cemented, fanglom-erate with a particle-size distribution similar to that of the modern alluvial fans. Highly fractured and altered mylonitic leucogranite orthogneiss below the fault plane is so soft that it is cut by closely spaced rills, much like the Inyo Range example shown in Figure 2.20A. Even these narrow spur ridges have triangular facets (see the left side of Figure 4.9). The relatively undissected triangular facets of the piedmont fault scarps merely indicate that fan gravels have a much higher infiltration rate than the bedrock, so there is less runoff to erode rills. Fluvial erosion varies along this uniformly rising escarpment. Stream power is proportional to drainage-basin area. So the drainage basin of Bighorn Canyon in the center of the Figure 4.9 panorama has a deeper valley and a larger alluvial fan than the smaller adjacent fluvial systems. Tectonic controls on canyon cutting are complicated here because Bighorn 132 Chapter 4 Canyon was created in part by base-level falls caused by movements along a sequence of now inactive low-angle normal faults (Cichanski, 2000). Valley floors upstream from the active range-bounding faults of most class 1 fronts are exceptionally narrow, and the adjacent spur ridges may be high where powerful streamflows have had sufficient time to increase valley relief. Class 1 drainage basins generally are quite narrow relative to their length. Stream-channel entrenchment into the fan-heads of class 1 fans, if present, is temporary and most likely is the result of climate-change perturbations. Temporary and permanent are easily defined by taking advantage of profound differences between Pleistocene and Holocene climates. Temporary fan-head trenches have formed since the most recent climate-change induced aggradation event. These fan surfaces have Holocene soil profiles. Conversely, the presence of a soil profile on alluvium deposited during the Pleistocene is indicative of permanent stream-channel entrenchment. Post-glacial aggradation surfaces are younger than 13 ka in the deserts of the American Southwest. These soil profiles are less than 0.5 m thick and commonly lack argillic horizons. Argillic horizons of Late Pleistocene soils are red and the depths and amounts of pedogenic calcium carbonate illuviation are substantially greater than in Holocene soils. Arroyo cutting and backfilling of basin fill occurs on a still younger time scale of 0.1 to 0.3 ka. Climatic change impacts, locally enhanced by humans, can result in temporary entrenchment (Bull, 1997) of fans composed of cohesive sandy deposits. Equation 4.5 best describes the appropriate set of interre- lations where such temporary entrenchment has occurred downstream from a mountain landscape with class 1 erosional characteristics. Impacts of long-term and short-term climate changes are present, but the intent here is to classify landforms on the basis of tectonic controls instead of adjustments to changing climates. Class 1 fronts in weak rocks or unconsolidated alluvium have a different landscape assemblage. Such mountain fronts lack the rock mass strength required for development of narrow, steep canyons in rugged mountains. Triangular facets are common along the front shown in Figure 4.1 OA, which has a sinuosity that would approach 1.0 (a straight line) were it not for the presence of prominent wide valleys. The larger streams have valley-floor widths that are much wider than their flood-discharge widths because they have been at the base level of erosion for a long time. Small streams, with less unit stream power, still have valley-floor widths similar to their flood-discharge widths. They continue to degrade at the times of infrequent large rainfall events in the Mojave Desert. Deposition of alluvial fans continues immediately downslope from the range-bounding fault. The broad mountain valleys suggest a slightly active (class 3) landscape but the straight mountain-piedmont junction and a piedmont of class 1 alluvial fans implies a much higher level of tectonic activity. The importance of lithologic controls on landscape evolution is acknowledged for the active fault zones shown in Figure 4.10 and such landscapes are assigned to class IB. Use of class IB recognizes the importance of rock mass strength on response times of geomor-phic processes to tectonic uplift in this region. Class 1B is a convenient way to rank strike-slip faults slicing through the basin fill as belonging in the highest slip-rate class. Eight different landscape assemblages are used to define the five tectonic activity classes of Table 4.1. Class 1 has been subdivided into A and B parts in order to accommodate the broad valley floors that are typical of noncohesive materials. No difference in uplift rates is implied by this subdivision of class Figure 4.10 Aerial views of class 1F3 mountain fronts. A. Cenozolc alluvium and fractured, weathered quartz monzonite have been ruptured by the right lateral strike-slip Calico fault northwest of Hildago Mountain, central Mojave Desert, California. Dirt road In upper-left for scale. Tectonic Activity Classes 1. Classes 2, 3, and 4 have permanently entrenched alluvial fans; classification is based on erosional stages of the mountain-piedmont junction that require progressively more time to form. Examples include embayment of valleys to create a more sinuous mountain-piedmont junction, and erosional deterioration of triangular facets. Class 5 consists of landscape assemblages that describe three types of pediments, the piedmont landform indicative of prolonged tectonic inactivity. The attainment of equilibrium stream channels, as indicated by the presence of straths and strath terraces, varies with tectonic activity class. Class 1A terrains of semiarid regions typically have valley-floor widths that are the same as the widths of peak stream discharges. Strath terraces are not present in such narrow canyons because downcutting is not interrupted by prolonged attainment of the base level of erosion. The miniscule widths of such valley floors, relative to the valley widths, result in ^ratios that clearly define situations where tectonically induced downcutting maintains a mode of stream operation far to the degradational side to the threshold of criti- of Mountain Fronts 133 Figure 4.10 Aerial views of class 1F3 mountain fronts. F3. Faulted basin fill along the west side of the Panamint Pange, southeastern, California. Large valleys have broad floors indicative of attainment of the base level of erosion, but small streams at the mountain front continue to actively downcut. Note the triangular facets in uplifted alluvium and low fault scarps of late Holocene age. cal power. Of course large streams of humid regions have sufficient annual unit stream power to remain at their base levels of erosion nearly all the time, so typically have strath terraces. Class 3 valleys of semiarid regions have strath terraces. Class 5 valleys may lack strath terraces because of insufficient uplift (very slow isostatic) to allow creation of more than one obvious base level of erosion - the beveled bedrock surface of pediments. The "typical landforms" of Table 4.1 are but a sample of the many landscape responses to uplift. Examples are listed in Table 4.2. Class 1 landscapes are much different than class 5 landscapes with similar total relief, climate, rock type, and drainage-basin area. Class 1 landscapes have more convex ridgecrests, steeper footslopes that in extreme cases are a continuation of the convex ridgecrests, narrower and steeper valleys, less sinuous mountain fronts, predominately young soils on rapidly aggrading piedmonts, and thick accumulations of Quaternary basin fill. Shapes of valley cross sections are useful for identification of tectonic activity classes. Each valley has a shape that is a function of rates of tectonically induced down-cutting, with short response times for changes in valley-floor width after a base-level fall, and long response times for ridgecrest relief and convexity. The independent variables of climate and rock type profoundly affect response times to uplift for each particular study area. The five-part classification of Table 4.1 may be too detailed when there is insufficient time or funds for more than reconnaissance field work in large study regions. A three-part classification may be preferable: highly active class 1, grouping of classes 2, 3, and 4 into a moderately to slightly active second class, and grouping of classes 5A, 5B, and 5C into an inactive third class. Bull and McFadden (1977b) and Bull (1977) used this three-part classification to identify tectonically active mountain fronts and to determine spatial patterns of late Quaternary tectonic uplift in southeastern California. 134 Chapter 4 C\aee of Relative Tectonic Activity Relative Uplift Rate Typical Landforms Piedmont* Mountain** Active Class 1A - maximal Au/At>Acd/At+Apa/At Unentrenched alluvial fan [0.6-0.9] V-shaped valley profile in hard rock [1.1-1.4] Class 113 - maximal Au/At>Acd/At+Apa/At Unentrenched alluvial fan [0.6-0.9] U-shaped profile in soft rock [1.0-1.2] Class 2 - rapid Au/AtApd/At Entrenched alluvial fan [1.0-1.1] V-shaped valley [1.1-1.3] Class 3 - slow Au/AtApd/At Entrenched alluvial fan [1.1] U-shaped valley [1.0-1.1] Class 4 - minimal Au/AtApd/At Entrenched alluvial fan [1.1] Embayed front [1.0-1.1] Inactive Class 5A Au/At«Acd/At>Apd/At Dissected pediment [1.1] Dissected pediment embayment [1.0-1.1] Class 513 Au/At«Acd/At=Apd/At Undissected pediment [1.0] Dissected pediment embayment [1.0] Class 5C Au/At«Acd/At Af (4.6) Equation 4.6 pertains to classes 2, 3, and 4 of relative tectonic activity (Table 4.1). The erosional land-forms of mountain valleys and mountain-piedmont junction are used to differentiate between these three classes whose piedmont landforms consist of entrenched alluvial fans. Uplift is relatively less rapid than for class 1 alluvial fans of the same study area. Uplift rates of landscapes described by equation 4.6 may be slow (<0.1 m/ky), where ephemeral streams flow over resistant rocks. Resistant fluvial systems respond slowly to tectonic perturbations, so tend to retain landscape signatures indicative of uplift. Conversely, similar landscape assemblages may be associated with rapid uplift (>2 m/ky) for humid-region streams flowing over soft rocks. These rivers have sufficient annual unit stream power to quickly adapt to tectonic base-level fall. Stream-channel downcutting shifts the apex of active fan deposition downslope from the mountain-piedmont junction. Deposition can no longer Tectonic Activity Classes of Mountain Fronts 135 Class 1A Badwater fan, Black Mountains, Peath Valley National Monument, California; Peer Canyon, Lytle Creek, San Gabriel Mountains, Los Angeles County, California; Panoche Creek, Laguna Seca Creek, Piablo Range, Western Fresno County, California; Wassuk and Stillwater Ranges, of Western Nevada. Classic Faulted alluvium northwest of Hidalgo Mountain, central Mojave Pesert, California; San Andreas fault, Carrizo Plains, California, Peath Valley fault zone. Class 2 Westfront of Grapevine Mountains and Pesolation Canyon area, Peath Valley National Monument, California; Eaton Canyon, San Gabriel Mountains, Los Angeles County, California; East front of Toiyabe Range, central Nevada. Class 3 Hanuapah Canyon, Panamint Range, Peath Valley National Monument, California; Little Panoche Creek, Piablo Range, western Fresno County, California. Class 4 Wilson Canyon, Argus Range, Inyo County, California; 49 Palms Canyon, Joshua Tree National Monument, California. Class 5A South front of Whipple Mountains, eastern Mojave Pesert, California; West front of Santa Catalina Mountains, Pima County, Arizona. Class 5E3 South front of the Granite Mountains, north front of the Coxcomb Mountains, eastern Mojave Pesert, California. Class 5C East front Riverside Mountains, eastern Mojave Pesert, California; Northeast front of Sheep Mountain, Gila Mountains, western Arizona. Table 4.2 Mountain fronts whose landscape assemblages illustrate five classes (Table 4.1) of relative tectonic activity in the southwestern United States. occur on the fanhead, and the strength of soil-profile development indicates how much time has elapsed since the fanhead last received deposits from the source watershed. About 50 to 500 ky may have passed since class 1 conditions prevailed. The strongly developed soil profiles on permanently entrenched fanheads also attest to minimal rates of piedmont degradation. Ape/ generally is much smaller than Au. Streams of class 2 landscapes downcut sufficiently rapidly to maintain V-shaped cross-valley profiles upstream from the range-bounding fault, whose location generally coincides with the mountain-piedmont junction. Longitudinal stream profiles of such mountain reaches may indicate attainment of type 2 dynamic equilibrium (Section 2.4.3). Deep fanhead trenches, and the dark surficial pavements indicative of Pleistocene age alluvial surfaces (Fig. 4.11) indicate that fanhead trenching is permanent. Minimal backwearing of the mountain-piedmont junction has occurred, but triangular facets generally are more dissected than class 1 escarpments. Greater tectonically induced downcutting near the mouths of rising watersheds may result in a drainage basin that is progressively narrower toward the basin mouth. These are referred to as "hourglass" or "wineglass" valleys (Beaty, 1961), and may develop in either class 1 or class 2 fluvial systems. Pleasant Canyon in the Panamint Range is an example. Class 3 landscapes are indicative of slow or intermittent uplift during the Quaternary. Holocene age fault scarps are rare. Some, like the Hanaupah fault cutting the ^55 ka alluvial fan surface near the toe of the Panamint Range piedmont, are related to adjacent pull-apart basin tectonics, not the now-dormant range-bounding fault. Valley floors are much wider than the floodplain widths, and the ubiquitous presence of U-shaped valleys (Fig. 4.4A) implies pro- Chapter 4 longed attainment of type 1 dynamic equilibrium. Alternating periods of lateral erosion and renewed downcutting result in strath terraces being a typical landform of mountain reaches of class 3 terrains. The eastern flank of the Panamint Range is an interesting class 3 mountain front. A sinuous moun- Figure 4.11 Aerial view of the class 2 mountain front along the west side of the Funeral Range, Death Valley National Park, southeastern California. Drainage basins are progressively more narrow downstream. The mountain-piedmont junction coincides with the trace of a normal fault except in fanhead embayments. The fanhead surface may have been created at the time of the Q2b, 125 ka aggradation event (Table 1.2). Two-lane paved highway near the base of the piedmont for scale. tain-piedmont junction (Fig. 4.2) is a short distance upslope from the range-bounding fault zone. Fault scarps cut the oldest Pleistocene alluvial fan. Stream-channel entrenchment has preserved the alluvial surfaces of three climate-change induced Pleistocene aggradation events (Table 1.2). The combined affects Figure 4.12 Hanaupah Canyon alluvial fan and class 3 mountain front, east side of Panamint Range, Death Valley, southeastern California. Poorly preserved triangular facets mark approximate location of range-bounding normal fault. The fanhead trench is 100 m deep at the fan apex. Approximate ages of fan aggradation event surfaces are: 04, the active channels; Q3, 7-9ka; Q2c, 55 ka; Q2b, 125 ka; and Q2a, >500 ka. Tectonic Activity Classes of Panamint Range tilting continue to lower the base level of erosion for the trunk stream channel. It has been at least 500 to 700 ky since class 1 conditions prevailed, which is a much longer time span than the 125 ky estimate for the class 2 front shown in Figure 4.12. Straths and typical class 3 cross-valley topographic profiles are shown in Figure 4.4. Class 4 mountain fronts border on being tec-tonically inactive. Indeed, some may have been inactive since the early Quaternary and their landforms merely reflect long response times to old tectonic perturbations. The numerous embayments of class 4 fronts create a sinuous mountain-piedmont junction that no longer coincides with the range-bounding fault. Wilson Canyon in the granitic Argus Range of southeastern California is an example (Fig. 4.13). Although bordered by steep hillslopes, the broad valley floor of this deep erosional embayment has a U-shaped cross-valley profile. Triangular facets are not obvious and have been notched by deep valleys that extend more than two-thirds of the distance to the apex of the facet (stage 5 of Table 3.1). The wide valley floor has fill terraces, which are clear evidence of temporary reversals of long-term valley-floor degradation caused by climate-change induced aggradation events. Note the presence of an unentrenched alluvial fan downstream from entrenched valley fill (Fig. 4.13). It should not be considered as evidence for tectonically active (class 1) conditions, particularly in view of the stage of landscape erosion represented by the erosional landforms. Instead this fan is the result of major alterations in geomorphic processes during the change from Pleistocene to Holocene climate. Watersheds underlain by granitic rocks typically yield huge amounts of sediment in response to increases in runoff and decreases in protective plant cover (Bull, 1991, Section 3.3). Response times to this recent climatic perturbation in the Argus Range are 5 to 10 ky. So this fluvial system has yet to switch to a stream-channel entrenchment mode of operation after an aggradation event caused by the Pleistocene-Holocene climatic change. Class 4 terrains may seem inactive until they are contrasted with class 5 terrains. Compare the granitic landscapes of Figures 4.13 and 4.15A. Class 4 fluvial systems still have well-integrated drainage networks and inselbergs are rare. Diagnostic landforms of class 5 terrains consist of three types of pediments, each representing a different relative rate of stream-channel and pied- of Mountain Fronts 137 Figure 4.13 Aerial view of the class 4 mountain front along the east side of the central Argus Range at Wilson Canyon, southeastern California . A highly embayed mountain-piedmont junction, stage 5 triangular facets, and U-shaped mountain valley cross sections all suggest virtual cessation of uplift along the range-bounding fault. mont degradation. Degradation of the pediment ipd) may be less than, equal to, or exceed the rate of downcutting by stream channels (cd) flowing across the piedmont. Permanent (Pleistocene) and temporary (Holocene) channel entrenchment of pediments is defined in the same way as for alluvial fans: At£«4^>4P^ (47) Af Af Af Au< 100 km of lateral displacement. Slip rates may decrease as second-generation primary faults become more active but their total slip is constrained by widths of fault blocks. Complexity increases still more when the diverse secondary faults and folds are added to the structural mix. Antithetic faulting occurs as cross faults between the primary faults in a parallel set. They may be more obvious than synthetic faults because of their steep angle to the primary faults (Fig. 4.19). Tectonic geomorphologists need to be alert for landscape characteristics where tectonic deformation is presently occurring in the diverse structural settings created by evolving sets of parallel strike-slip faults. Fault scarps and associated fissures are useful for locating the most recent tectonic deformation. The degree to which primary faults have become locked needs more study. The way in which the Mw magnitude 7.3 Landers earthquake of 1991 side-stepped along five faults instead of rupturing a single fault zone suggests that rotation of the set of Figure 4.19 Diagrammatic model of set of fault blocks that are rotated counter-clockwise by dextral faulting. From Figure 4 of Nur et al., 1956. A. Original vertical 30° orientation related to initial angles of shear for a set of blocks created by regional strike--slip faulting. F3 About 10° more block rotation brings the dextral fault system to the rotation limit. At the threshold time depicted here the primary faults become locked, and a set of secondary faults starts to rupture the primary blocks. Secondary faulting will continue until they too approach the fault-mechanics threshold, whereupon a third-order set of fault blocks will develop. C. Second-order faulting creates a complex pattern of strike-slip faults and associated mountain fronts. Slgmoldal characteristics have been added to the cross-fault terminations. Mojave Desert faults is important, or that the orientation of the stress field has changed. Elsewhere, such as the Marlborough section of the Alpine fault of New Zealand, rupture continues to occur mainly on the primary faults. Although conceptually appeal- Tectonic Activity Classes ing, the model of Nur et al. (1986) may be difficult to apply to specific field situations. Using tectonic activity classes to assess the relative displacement rates on strike-slip faults may seem illogical. Many strike-slip faults generate minimal topographic relief and tend to be concealed beneath basin fill. Tectonic geomorphologists use the secondary compressional and tensional structures generated by wrench-fault tectonics, or rely on partitioning of crustal deformation into lateral and vertical domains. The Blue Cut and Pinto Mountain sinestral fault zones in the eastern Transverse Ranges are active. Both are important elements of the transrota-tional model of plate-boundary shearing of the Walker Lane-Eastern California shear zone as a single tectonic entity. The Blue Cut fault zone is virtually hidden along most of its length but its tectonic activity during the late Quaternary is revealed by impressive normal-faulted mountain fronts (Fig. 4.20). As elsewhere in the shear zone, faulting of pedimented terrain is creating new mountain ranges. The uplifted pediment shown in Figure 4.20 is 20 to 150 m above actively aggrading alluvial fans. The quartz mon-zonite of these watersheds is especially sensitive to Pleistocene-Holocene climatic change perturbations, resulting in vigorous aggradation events (Bull, 1991). The -55 ka Q2c fan surface has a prominent fault scarp where it was not buried by the unfaulted -7 ka Q3b aggradation event. This front was classified as Class 1 on the basis of a mountain-piedmont junction sinuosity of 1.1 and a mean valley floor width-valley height ratio of only 0.21. I presume that the hidden sinestral-fault component of the partition is also worthy of a Class 1 tectonic activity class. The hidden sinestral-fault component of the Pinto Mountain fault is much more active than the adjacent mountain-front fault. Trenching revealed five surface ruptures since 14 ka (Cadena et al, 2004). Thrust and normal faults may also occur as sets where uplift is distributed amongst several faults. Tectonic landforms usually reflect cumulative uplift across a zone of faulting. Tectonically active thrust-faulted mountain ranges commonly have several active fault zones (Section 3.2.3) that define reaches of streams with different uplift rates. This favors analysis of landscapes associated with external (piedmont), range-bounding, and internal faults. Erosional land-forms of internal fronts reflect the cumulative influences of downstream base-level falls that are transmit- of Mountain Fronts 145 ted upstream, as well as the effects of local faulting. Although tectonic perturbations decrease exponentially in the upstream direction (Section 2.2.5), they may keep a stream on the degradational side of the threshold of critical power even adjacent to an active internal fault. Small-scale pockets of deposition (Figs. 3.18, 3.22) may be temporary. Folding results in even broader tectonic perturbations, so presentation of tectonic activity class maps starts out with the simpler situation of base-level changes caused by faulting. 4.2.3.2 Maps of Relative Uplift Geomorphic analyses of tectonic activity of mountain fronts complement stratigraphic paleoseismic input about timing and magnitudes of surface ruptures provided by trenching of fault zones. Data about the locations, depths, magnitudes, styles, and frequencies Figure 4.20 Normal faulting of class 1 Eagle Mountains of the eastern Transverse Ranges has raised a pediment. Sinestral Blue Cut fault is parallel partitioned and largely hidden beneath valley fill but is an inferred class 1 strike-slip fault. Ages of alluvial aggradation events are: Q3b ~7 ka, Q3a -12 ka, and Q2c -55 ka. 146 Chap of historical earthquakes provided by seismologists tell us much about historically active fault zones. But we also need a quick, inexpensive way to assess potential earthquake hazards of large numbers of range-bounding faults that are presently locked and may be presumed to be seismically inactive. Scientists, engineers, and planners all need regional assessments of late Quaternary tectonic activity. This section examines the areal distributions of mountain fronts that range from tectonically active to inactive. Diverse settings include the normal-faulted Basin and Range Province, folded Coast Ranges, thrust-faulted Transverse Ranges, and lateral-faulted Mojave Desert. These maps can be compared with other geomorphic assessments of uplift rates. Published evaluations include the Death Valley region (Smith et al., 1968; Hooke, 1972; Smith, 1975, 1976, 1979; Schweig, 1989; Anderson and Densmore, 1997; Slate, 1999; Machette et al., 2001; Walker et al., 2005), the Great Basin of Nevada (Wallace, 1978, 1987a,b; Fonseca, 1988; Pearthree, 1990; dePolo et al., 1991; dePolo and Anderson, 2000), the Transverse Ranges of California (Rockwell et al., 1988; Keller et al., 1998; Spotila and Anderson, 2004), and the central Coast Ranges of California (Lettis, 1982, 1985). The state of this art has changed. Studies of several responses of mountain-front landscapes to base-level fall began as a way to class relative tectonic activity of Mojave Desert mountain fronts near proposed nuclear generating power stations. The next step was to study the obviously active Death Valley region, where Bull and McFadden (1977) assessed potentially useful landforms. Their work focused on impressive fronts of lofty mountain ranges, but now we more fully realize that active tectonics is influenced by the pervasive strike-slip faults of the Walker Lane-Eastern California shear zone. Tectonic deformation is partitioned. Strike-slip faults traverse valley floors. Alluvium, instead of rock, is the main material in low piedmont fault scarps and offset stream channels so these strike-slip faults are an important element of tectonic activity class IB. As for low normal- and thrust-fault scarps, low scarps in Holocene alluvium created by offsets along strike-slip faults catch the attention of ever-vigilant paleoseismologists. The map of tectonic activity classes north of the Garlock fault (Figs. 4.2, 4.21) has consistent trends. Strike-slip faults are rated in the highest tectonic activity class. About 110 km of the Owens Valley fault zone was the source of a Mw magnitude ter 4 7.6 earthquake in 1872. This fault zone may extend all the way to the Garlock fault (Unruh et al. 2002; Monastero, et al. 2005). The 310 km long Death Valley fault zone has a dextral rate of slip of ^ 4 to 9 m/ky (Reheis et al., 1995; Klinger, 1999, 2001; Machette et al., 2001; Frankel et al., in press for 2007). Examples of pull-apart basins include Death Valley by the Black Mountains, Saline Valley, and Deep Springs Valley. Mountain front characteristics indicate progressively less relative uplift rates from north to south towards the Garlock fault. Bull and McFadden suggested that this was a regional pattern of decreasing tectonic activity. Using the modern transrotational model (Dickinson, 1996) we now realize that the left-lateral Garlock fault and the domains of left-lateral faulted mountain fronts on both sides of the Garlock fault in the northeastern Mojave Desert and the eastern Transverse Ranges accommodate the overall continuity of the Walker Lane-Eastern California shear zone. The Argus, Slate, Panamint, Black, and Grapevine-Funeral ranges have less Quaternary tectonic activity on their eastern mountain fronts than on their west sides. Part of the Owlshead Mountains is shown here as being class 4, which fits the Quaternary landscape assemblage. But this is another example of rejuvenated mountain fronts: piedmont scarps created by thrust faulting of the piedmont downslope from a pedimented mountain front (Christopher Menges, personal communication). I suspect that most of the Owlshead small mountain fronts should be given a tectonic activity class of 1 - the highest rating from a hazards standpoint. The Sierra Nevada front is only class 3. The Owens Valley strike-slip fault is class 1 and carries on through the Coso complex all the way to the Garlock fault. Pull-apart basin subsidence that created Owens Lake basin has bounding faults that have encroached into the Sierra Nevada. The result is a local base-level fall and a class 2 mountain front. Then the Sierra Nevada mountain front tectonic activity decreases progressively to the south: 2 to 3, 3 to 4, and 4 to 5. Each of these changes in range-front landscapes is accompanied by a jog in the mountain-piedmont junction (Fig. 4.21). Piedmont alluvial-fan deposition opposite Mt. Whitney consists of climate-change induced pulses of aggradation whose ages appear to match the times of glacial-moraine formation (Le et al., 2007). Instead of being thick tectonic alluvial fans, these are thin climatic fans. Key evidence for Tectonic Activity Classes of Mountain Fronts 147 Figure 4.21 Tectonic activity classes of mountain fronts in the Death Valley - Sierra Nevada area. climatic instead of tectonic fans is the way in which shallow fanhead trenches have cut through thin sur-ficial alluvium to expose buried surfaces of old fans. 10Be dating of faulted fans by Le et al. indicates a slow rate of uplift of about 0.2 to 0.3 m/ky since 125 ka. Like the eastern Panamint Range, the rugged landscape is mainly inherited from earlier times of rapid uplift. Big mountains take forever to wear down in this climatic setting. Relative uplift rates of mountain fronts in the extensional terrain of the Basin and Range Province in west-central Nevada were appraised using five tectonic activity classes. Normal faulting has created mountain ranges of volcanic, sedimentary, and plu-tonic rocks. The climate is moderately seasonal, arid to semiarid, thermic to mesic. So these mountains preserve diagnostic tectonic landforms for a long time, even where rates of tectonic displacement are modest. This study area is part of the central Nevada seismic belt, a 300 km long zone of surface ruptures that occurred in 1903, 1915, 1932, 1934, and 1954. These mountain ranges are fairly small (Fig. 4.22), so all drainage basins larger than 3 km were used in the Vf analysis. The Table 4.3 numerical values do not overlap where the three moderately active classes (2, 3, and 4) were lumped together, much like the Bull and McFadden (1977) three-class approach. Overlapping of data is typical when five classes are used, but is minimal here. Even so, it makes sense to use several tectonic landforms to assess relative tectonic activity. Anomalous situations may not be recognized if you use only one landscape statistic. Only some of the range-bounding faults have been active during the Quaternary. One characteristic of the Basin and Range Province is diverse normal-fault slip rates. Some fronts have been raised during Chapter 4 Figure 4.22 Digital image of the Pleasant Valley (PV)-Dixie Valley (DV) region in west-central Nevada. historical earthquakes, for example the 1915 magnitude 7.1 earthquake that ruptured several fronts on the west side of the Tobin Range over a time span of several hours (Wallace, 1977). Other mountain fronts, such as those bounding the East Range, have been inactive for more than 1 My. Uplift rates are moderate: class 1 fronts are being raised only about 0.3-0.5 m/ky (Wallace, 1978; Pearthree, 1990). Uplift should be considered relative. Range crests rise "0.3 m and the intervening basins subside "0.7 m during magnitude Mw 7 earthquakes (Stein and Bar-rientos 1985; Stein and Bucknam, 1986). The locations of Holocene faulting in Dixie-Pleasant Valley shift from the east side of the Stillwater Range to the west side of the Tobin Range. The cross-over is in the Sou Hills: a barrier that arrests surface ruptures propagating from either north or south (Fonseca, 1988). Multiple zones of base-level fall complicate assignment of tectonic activity classes. Streams along the west flank of the Tobin Range cross the active range-bounding fault zone and then flow south, crossing a zone of dispersed faulting in the Sou Hills before descending into Dixie Valley. Minor relative uplift of the Sou Hills, and subsidence of the north end of Dixie Valley, occur with each faulting episode. The net effect is a base-level fall that induces stream-channel entrenchment, which migrates to distant upstream reaches. For this reason, the piedmont west of § o - Sou Hills Tobin /n. Range - r£?en ultimate 'event ^ J 1915 A., D. event i i i i i 10 20 30 40 Km Figure 4.23 Fault scarp vertical displacements that decrease south towards the Sou Hills from the west flank of the Tobin Pange for the 1915 A.D. (Wallace, 19S4) and early Holocene surface ruptures. From Figure 11 of Fonseca (19SS). the Tobin Range consists mainly of class 2 fans (entrenched), rather than the class 1 fans (unentrenched) that one would expect from the presence of other obvious class 1 tectonic landforms. Fonseca (1988) describes how the Sou Hills act as a transverse barrier that arrests propagation of fault-ruptures. Total vertical tectonic displacement for the two most recent Tobin Range surface ruptures dies out towards the south where faulting was halted by the Sou Hills transverse barrier (Figs. 4.23, 4.24). The 1915 surface rupture had a normal-fault displacement that decreased from a maximum of 6 m along the Tobin Range to 0.3 to 1.0 m where it terminated as many small scarps in the Sou Hills. The penultimate surface-rupture event had the same magnitude and trend. It is older than the 6.8 ka Mazama volcanic ash (Wallace, 1984). Was this a characteristic behavior? Fonseca used mountain front tectonic activity classes to see if landforms with distinctive tectonic signatures define a persistent tectonic style during the Quaternary. Total mountain-range relief of both the Tobin and Stillwater Ranges decreases towards the Sou Hills. Geo-morphic indicators of Quaternary uplift rates, such as valley floor width-valley height ratio and mountain-piedmont junction sinuosity, decrease towards the Sou Hills from both the north and the south. Climatic and lithologic controls on landscape evolution vary little in this area, so the changes of tectonic activity classes towards the Sou Hills (Fig. 4.24) demonstrate systematic decreases of magnitudes and rates of tectonic base-level fall towards the Sou Hills. Both geomorphic analyses, using short- and long-term landscape characteristics, define a consistent tectonic style. The National Earthquake Hazards Reduction Program of the U.S. Geological Survey is a ma- Tectonic Activity Classes of Mountain Fronts 149 Mountain front Sinuosity of Mean valley floor Range of Kf ratios Triangular facet of Figure 4.25 mountain-pied- width-valley height dissection stag- mont junction, J ratio, Vf es of Table 2.1 Tectonically Active Mountain Fronts (Class 1) 16 1.0Ö [-0.92] 0.15 [-1.Ö6] 0.0Ö to 0.35 1 through 4 10 1.07 [-0.93] 0.06 [-1.95] 0.05 to 0.09 1 through 5 Tectonically Active Mountain Fronts (Class 2) 1 1.3 [-0.7] 0.16 [-1.Ö5] 0.0Ö to 0.51 1 through 5 6 1.2 [-03] 0.13 [-1.ÖÖ] 0.11 to 0.16 1 through 5 7 1.2 [-03] 0.13 [-1.ÖÖ] 0.06 to 0.53 2 through 5 & 1.2 [-03] 0.17 [-1.Ö4] 0.07 to 0.25 2 through 5 15 1.09 [-0.9] 0.31 [-1.7] 0.23 to 0.42 2 through 6 17 1.19 [-0.1] 0.51 [-1.5] 0.15 to 0.Ö7 2 through 5 19 1.19 [-0.1] 0.23 [-1.7Ö] 1 through 5 Moderately Active Mountain Fronts (Class 3) 2 1.9 [-0.1] 1.7 [-0.31] 0.43 to 3.33 1 through 5 5 1.6 [-0.4] 1.2 [-0.Ö1] 0.23 to 2.59 2 through 4 Slightly Active Mountain Fronts (Class 4) 4 2.2 [+0.2] 1.3 [-0.71] 0.36 to 3.53 2 through 5 11 3.2 [+1.2] 7.0 [+4.99] 0.91 to 39.4 4 through 5 12 2.3 [+0.3] 1.2 [-0.Ö1] 0.3 to 3.75 3 and 5 13 2.2 [+0.2] 1.0 [-1.01] 0.15 to 2.25 3 and 4 14 2.0 [0] 1.9 [-0.11] 0.71 to 2.Ö6 3 through 6 20 1.Ö [-0.2] 2.94 [0.93] 0.71 to 11.4 4 through 6 Inactive Mountain Fronts (Class 5) 3 3.5 [+1.5] 2.5 [+0.49] 1.3Ö to 3.Ö3 7 9 2.1 [+0.1] 2.0 [-0.01] 1.25 to 2.50 6 10 2.6 [+0.6] 73 [+5.79] 1.33 to 16.3 4 through 6 Table 4.3 Data used to assess five tectonic activity classes of mountain fronts in west-central Nevada (Fig. 4.24). Other ways to view the data are shown inside brackets [departure from mean value for entire study area]. Data from Julia Fonseca and John Partridge. jor source of funds for paleoseismology projects. One was an evaluation of mountain front tectonic activity in the central Mojave Desert (Bull, 1977b), where right-lateral faulting predominates. Two earthquakes occurred after this study, the Mw magnitude 7.3 Landers event of 1992 and the Mw 7.1 Hector Mine event of 1999. Unmodified 1977 results are shown here to better evaluate this geomorphic approach for estimating earthquake hazards. The results of these studies are combined in the regional assessment of Quaternary tectonic activity shown in Figures 4.25A, B. Tectonically active Chapter 4 crustal processes in the Walker Lane-Eastern California shear zone. Some parallel right-lateral faults of the central Mojave Desert are highly active and others inactive (Fig. 4.25B). Uplift rates decrease towards the north: first to class 2 fronts and then to inactive fronts as the fault zones approach the Garlock fault. Jennings (1994) has classed some fronts as active that a tectonic geomorphologist would consider virtually inactive because no surface ruptures appear to have occurred during the past 1 My. The active Garlock fault runs through a region of inactive mountain fronts, with the relative activity increasing to both the north and south (Figs. 4.21, 4.25B). A few short class 1 left-lateral fronts in the northeastern Mojave Desert support the presence of a local tectonic domain proposed by Garfunkel (1974). Ongoing field studies show that this area of active transrotation (Dickinson and Wernicke, 1997) turns out to be much larger and includes both the Avawatz Mountains south of the Garlock fault and the Owlshead Mountains north of the fault (Chris Menges, 2006, personal communication). Active fault zones bound the impressively high Transverse Ranges. These ranges become progressively lower and less active towards the east. They Figure 4.24 Tectonic activity classes (upper right corner) of normal faulted mountain fronts, bedrock is mainly volcanic and sedimentary rocks with some intrusives. See Table 4.3 for numerical data for the numbered fronts. Data from analyses by John Partridge and Julia Fonseca. and inactive regions are bounded by the Garlock and San Andreas faults. The coastal region to the southwest has many active strike-slip faults and the thrust faults associated with the "big bend" of the main trace of the San Andreas fault transform boundary. The large tectonically inactive region of the eastern Mojave Desert includes southwestern Arizona. It extends right up to the edge of the Salton Trough, a rhomochasm connected to the Gulf of California spreading center to the south. The western Mojave Desert is a strange tectonically inactive landscape. The way in which Quaternary surficial faulting appears to die out to the North in the central Mojave Desert describes how unlikely surface ruptures may be, but may not be representative of complex deeper Tectonically active 'ac " o Ocea i Figure 4.25 Tectonic activity classes of mountain fronts In the arid Mojave Desert and semiarid eastern Transverse Ranges of southeastern California. A. Pegions of tectonically active and inactive mountain fronts in the Mojave Desert of southeastern California. Tectonic Activity Classes of Mountain Fronts f1 a u I t 116° .i-Jiefort 35° X fa, Manix- arstow P c . o Quaternary tectonic activity class Slightly active Inactive V 34° 151 Figure 4.25 Tectonic activity classes of mountain fronts in the arid Mojave Desert and semiarid eastern Transverse Ranges of southeastern California. B.Central and western Mojave Desert. Displacements along oblique strike-slip and normal faults have created mountain ranges of resistant crystalline basement rocks. Barstow is a large town. still have locally active mountain fronts (Figs. 4.20, 4.26) and are essential for transrotational tectonic connections within the Walker Lane-Eastern California shear zone. Rather surprisingly, mountain fronts immediately adjacent to the plate boundary Salton Sea Figure 4.26 Tectonic activity classes suggest a left-stepping pattern of active mountain fronts Indicative of right-lateral displacement along several faults of the postulated Sheephole shear zone in an otherwise inactive terrain. trough are as inactive as the mountain fronts of the eastern Mojave Desert and Arizona. Neither area has experienced active range-front faulting during the past million years. The Landers and Hector Mine surface ruptures sidestepped across several fault zones (Treiman et al., 2002). Spacing of tectonic activity classes of other mountain fronts suggests a similar behavior. An example is shown in Figure 4.26. These mountains are at the eastern edge of the tectonically active central Mojave Desert. The inselbergs and sinuous front of the Calumet Mountains signal long-term tectonic inactivity. However, a series of class 2 fronts describes a left-stepping pattern to the northwest that is suggestive of occasional movements dispersed along a right-lateral shear zone. How well do the tectonic activity classes of Bull (1977b) compare with the subsequent surface ruptures of 1992 and 1999? The 1977 analysis had most of the Landers surface rupture as discrete class 1 and class 2 mountain fronts, which included obvious fault scarps on desert plains. Each of the five fault zones of the 1992 event had parts that were classed as "most active" in 1977, with intervening sections of lesser earthquake potential. The Landers event involved only mountain fronts expected to be involved in the next surface rupture. 152 Chapter 4 Figure 4.27 Comparisons of 1977 estimates of the most active fault zones of the central Mojave Desert based on mountain-front tectonic activity classes of Figure 4.27F3. The complex surface ruptures of the Mw magnitude 7.3 Landers earthquake of 25 June 1992 and the Mw magnitude 7.1 Hector Mine earthquake of 16 October 1999 are shown by gray lines. The southern part of the surface rupture crosses young alluvial plains (Fig. 4.27) and the 1977 method, like the area north of the Garlock Fault, did not address strike-slip faulting. The same deficiency applies to the analysis of the northern part of the Hector Mine surface rupture, which crosses the plains of the Lavic Lake basin. The 1999 Mw magnitude 7.1 Hector Mine earthquake produced a maximum slip of 5.2 m on a -48-km long surface rupture involving the Lavic Lake fault and a portion of the southern Bullion fault. Bull (1977a) did not regard the Hector Mine area as being particularly active, being a mix of active and inactive fronts. The 1999 surface rupture coincided with some mountain fronts, such as along the Lavic Lake fault, that have obvious characteristics of minimal amounts of late Quaternary uplift. The Hector Mine event may be more unusual than the Landers event. This event might have a longer earthquake recurrence interval or the locus of strike-slip faulting might have shifted eastward. Trench stratigraphy analyses support the geo-morphic conclusion that the Lavic Lake fault passes through mountains with a low level of tectonic activ- ity. The work of Lindvall et al. (2000) is presented here because it underscores the advantages of using soil stratigraphy in paleoseismology (Rockwell, 2000). The "Bullion Fan" trench (Fig. 4.28) revealed a late Holocene surface rupture. But the "Drainage Divide" trench on the Lavic Lake Fault provided a much different and quite valuable fault history. The 1999 rupture cuts steeply sloping alluvial aprons with entrenched stream channels. Massive gravel deposits are weakly stratified, so soil-profile horizons were mapped. Figure 4.25 Map of the surface ruptures caused by the 1999 Mw magnitude 7.1 Hector Mine earthquake (solid heavy black lines) and the locations of nearby active fault zones (dashed lines) of the central Mojave Desert of southern California. Base map and pattern of faulting furnished courtesy of Tom Rockwell. Tectonic Activity Classes of Mountain Fronts 153 o s_ Av, vesicular A 3w, cambic calcic F3 3tk, argillic calcic F3 3k, calcic F3 K, calcic Only pre-1999 fault Figure 4.29 Soil stratigraphy for two late Pleistocene aggradation events exposed in the Drainage Divide trench, Lavic lake fault zone. A. Sequence of faulted soil-profile horizons on both sides of a fault zone fissure filling. All the black lines are fractures and faults created by the Hector Mine earthquake of 1999, except for a single heavy black line for the penultimate event. Figure 2 of Lindvall et al. (2000). Only soil-profile horizons, not sedimentary strata, were mapped in the wall of the 2-m deep trench. Surficial gravel with pronounced black varnish reveals a lack of Holocene deposition. As throughout the Mojave Desert, pulses of climate-change induced aggradation buried former surfaces with new increments of detritus stripped from adjacent hillslopes. The pulsatory to continuous nature of deposition depends partly on location within a fluvial system (Fig. 2.14). The trench-site location favors incremental accumulation of alluvium during aggradation events. Alluvial thicknesses and the strength and style of soil-profile development at the "Drainage Divide" trench site are nicely detailed in Figure 4.29A. These appear to be typical 125 and 55 ka aggradation events of the Mojave Desert (Bull, 1991; Slate, 1999; Machette et al., 2001). Evidence for a pulse of Holocene aggradation is not present in the trench, but this is not unusual. Alluvial geomorphic surfaces with black rock varnish are common throughout the Mojave Desert. Each represents a situation where the magnitude of stream-channel entrenchment was deep enough to contain the thickness of aggrading alluvium, thus preventing streams from spreading out and depositing another sheet of detritus stripped from the hillslope sediment reservoir of the watershed. The two soil profiles of Figure 4.29A do not represent time lines even though pedogenesis began after cessation of the -125 and ^ 55 ka aggradation events. Each of the two soil profiles of Figure 4.29A records an interval of nondeposition. The Q2b interval is 125 to 55 ka, and would be 125 to 0 ka were it not for the deposition of Q2c, which has been offset by two surface rupture events. Displacements are several meters right-lateral with a small vertical component. The right side of the faulted section looks like it has been lowered, but this is only apparent, being mainly the result of lateral displacements of stratigraphic units of nonuniform thickness. The 1999 event produced abundant fractures and faults, many of which extend to the surface. These are fresh and obvious now, but will be much less apparent after 20 to 40 ka of overprinting by pedogenic processes. The only pre-1999 fault cuts and offsets the argillic soil-profile horizon of the older (Q2b) soil (Fig. 4.29B), which started forming after the 125 ka aggradation event. The penultimate event also offsets the Bk horizon of the younger (Q2c) soil and terminates in the Q2c argillic horizon (Fig. 4.29A). It appears that no fault displacements occurred during the 125-55 ka soil-forming interval. The penultimate 154 Chapter 4 West Figure 4.29 Soil stratigraphy for two late Pleistocene aggradation events exposed In the Drainage Divide trench, Lavic lake fault zone. F3. Alluvium of the 125 ka (Q2b) and 55 ka (Q2c) aggradation events based on Figure 4.29A. A single pre-1999 fault displaces the fissure filling and adjacent deposits but does not extend to the surface. surface rupture event may have occurred shortly after deposition of Q2c alluvium, but before much soil-profile development had occurred. Holocene surface ruptures are not present. Recognizing what might seem as innocuous "mountain fronts" is essential for testing several models proposed for new orientations of plate-boundary faulting. Erosion obliterates small scarps in hilly ter- Figure 4.30 Relation of the surface rupture of the 1975 Galway Lake earthquake on a secondary fault to the orientations of the primary Emerson and Calico fault zones of the central Mojave Desert. From Figure 1 of Hill and Beeby, 1977. rain much faster than on gently sloping piedmonts. Finding and evaluating Holocene surface ruptures that cross mountain ranges requires field studies above a reconnaissance level. More use should be made of LiDAR (Section 5.6.1) in order to discern diminutive piedmont fault scarps. Rotation of the tectonic block between the Emerson and Calico faults has been sufficiently large that ruptures now occur on shorter secondary faults (Fig. 4.30). A Mw magnitude 5.1 earthquake occurred on the secondary Galway Lake fault in 1975 A.D. The angle between the primary and secondary faults is - 37° - quite similar to the 39° and 41° angles noted for other secondary faults in Figure 4.30. This kinematic style is significant. If the primary direction of faulting is locked because limits of rotation have been reached (Nur et al., 1986), then long surface ruptures such as that of the Mw magnitude 7.3 Landers 1991 earthquake can be expected to sidestep between many faults instead of being along a single fault zone. Secondary faults might have become more important. Long-term maximum shearing occurs in the center of a belt of parallel dextral faults, such as the Walker Lane-Eastern California shear zone. The Hector Mine event occurred on the eastern edge of the zone of active strike-slip faulting in the central Mojave Desert and only two fault zones were involved. Perhaps the primary faults on the fringes of the central Mojave Desert set of fault blocks have yet to become fully locked. Changes in kinematic styles of faulting may complicate the use of mountain front topographic characteristics to classify relative tectonic activities. Transfer of slip from the Emerson to the Galway Lake and other secondary faults diminishes the rate of uplift along the principal Emerson fault, but not enough to lower its designation from tectonic activity class 1 to class 2. This is still an impressive, rugged mountain front. Once again, the response times of sinuosity of the mountain-piedmont junction, and the valley floor width-valley height ratio, are sufficiently long that >100 ky is needed to discern decreased uplift rates in landscape characteristics in this arid realm. What are the mountain-front characteristics for the Galway Lake mountain front? Almost none! There is no mountain front for this surface rupture zone. It is so young that its location cannot be discerned on topographic maps. The 7-km long rupture barely broke the surface. Hill and Beeby (1977) Tectonic Activity Classes note at least one previous Holocene surface rupture and a very short remnant of an older 1-m high fault scarp. Perhaps the primary faults have only recently become sufficiently locked to transfer slip to secondary zones of surface rupture such as the Galway Lake fault zone. Patterns of faulting may change with time. Only 35 m/ky of the present 49 m/ky of relative motion between the Pacific and North American plates (DeMets et al., 1990, 1994) presently occurs along the San Andreas fault (Sieh and Jahns, 1984; Wel-don and Sieh, 1985). Roughly half of the remaining 14 m/ky of right-lateral shift between the Pacific and North American plates occurs along faults in eastern California and Nevada (Savage et al., 1990). Geodetic measurements suggest that cumulative right-lateral strain for the faults of the central Mojave Desert is occurring at about 7 m/ky (Sauber et al., 1986). Tectonic rotation of dextral structural blocks with time places them in less favorable positions, relative to the regional principal compressive stress. Paleoseismolo-gists should note the implications of regional structural re-alignment with the passage of time. Nothing remains static in the world of predicting earthquake hazards. Several structural models can be evaluated by using the tectonic geomorphology data of Bull (1977), Morton et al. (1980); Matti et al., (1985), and Hart et al., (1989). The northwest-trending faults should lose their ability to accommodate crust-al deformation as they become progressively more unfavorably oriented according to the model of Nur et al. (1993a, b). They use six historical earthquakes to define the initial stages of development of a new 120-km long fault, and conclude that the Landers event slip is partitioned between the old trend that has been rotated to N40°W and the new N15°W trend. Sowers et al., (1994, Fig. 5) present evidence that further supports the North-South directed shear model. The Figure 4.25B geomorphic evidence does not recognize this speculation. Perhaps more time is needed for new faults to create hills and scarps with a different orientation in a new version of a tectonically active landscape. The structural realignment, if real, has just started to form. A second model concerns a larger version of the North-South directed shear concept - the Eastern California Shear Zone, extending from the central Mojave Desert into the west-central Nevada area of Figure 4.24. Miller et al. (1993, p. 871) postulate that the major "seismic gap" in the historical of Mountain Fronts 155 earthquake record north of the town of Barstow is a likely candidate for a future major earthquake. Figure 4.25B does not support this hypothesis. Slightly to moderately active fronts near the town of Barstow grade into an area of mountain fronts that lack evidence of persistent faulting. Most of the terrain closer to the Garlock fault has been virtually inactive during the Quaternary, when compared to either the central Mojave Desert or the Death Valley region. Garfunkel (1974) recognized that the northwest striking dextral faults of the Mojave tectonic block do not extend beyond the Garlock fault. So his model emphasizes conjugate domains of right-lateral and left-lateral faults. Right-lateral is dominant but he recognized the importance of left-lateral domains in the long-term process of rotation and deformation of the Mojave block. These include the eastern extension of the Transverse Ranges, a left-lateral block in the Northeast part the Mojave Block, and the Garlock fault. Both left-lateral domains have active mountain fronts (Fig. 4.25B). The lack of persistent faulting north of Barstow remains a problem for this model too. A more complete kinematic model is needed, one that accounts for the large expanses of seemingly inactive terrain in the western and northern portions of the Mojave block. The dextral faults of the central Mojave Desert intersect the thrust faults of the eastern Transverse Ranges. Spotila and Anderson (2004) have studied the kinematics of these complex fault systems. Their model has dextral slip on the Helendale fault accompanied by movements on the range-bounding thrust fault along the western part of the San Bernardino Mountains. This implies a concurrent decrease in faulting east of the junction with the Helendale fault. The result is a quasi-stable triple junction that separates a western domain dominated by thrusting and an eastern domain characterized by a larger component of strike-slip displacement. They speculate that this domain junction has migrated to the west as the Mojave block has been translated southeastward along the San Andreas fault. The tectonic activity classes of Figure 4.25B fit their model, perhaps with the Lenwood fault being the junction. Alternatively the triple junction could be in the process of migrating from the Lenwood to the Helendale fault, or perhaps the triple junction involves an area instead of a single fault intersection. The presence or absence of coseismic rock-falls is especially obvious in the Mojave Desert because flat ground allows blocks from several events Chapter 4 Figure 4.31 Comparison of rockfalls from seismically active and inactive hillslopes of plutonic rocks in the central Mojave Desert, California. A. Abundant large boulders of biotite diorite remain on the hillsides of a class 5 mountain front. Only a few cobbles have been washed onto the adjacent piedmont. F3. Earthquakes have dislodged many large boulders of porphyry breccia from the hillside. Boulder at right is 12 m high and has a volume of -140 m3. This is a class 1 mountain front bounded by the right-lateral Lenwood fault. to accumulate and there are no trees to hide them. The contrast in rockfall block abundance between tectonically active and inactive mountain fronts can be striking (Fig. 4.31). Rockfalls are such a great way for tectonic geomorphologists to study earthquakes that Chapter 6 is devoted to this subject. The western Mojave Desert is a bit of an enigma. A triangular chunk of Mesozoic plutonic rocks is caught between the very active San Andreas and Garlock faults (Fig. 3.2). The area lacks the rugged mountain fronts of the central Mojave Desert. Instead a pedimented plain with shallow playa basins are the norm. It seems as inactive as the landscape between the Colorado River and Salton Trough of southern California. Closer inspection (Figs. 4.32, 4.33) suggests that the pediment domes may not be entirely erosional in origin. Subtle, long wavelength buckling folding may play a part in exposing much of the inherited pediments of this area. This block may be quite different from both the eastern and central Mojave Desert, but still seems to be an area of minimal surface-rupture risk. The thrust faulted south side of the San Gabriel Mountains of southern California (Figs. 3.1-3.3) is part of a large transpressional bend in the San Andreas fault system. The setting is different than either the Mojave Desert or Nevada, where most mountain ranges are small and the climate is arid to semiarid. Thrust faulting has created a rugged mountain range of sheared and altered plutonic and metamorphic rocks. The long, lofty San Gabriel Mountains promote orographic lifting of storms from the nearby Pacific Ocean. Precipitation increases rapidly with increase in altitude. The climate ranges from strongly seasonal semiarid and moderately seasonal thermic (Table 2.1) on the southern piedmont to strongly seasonal subhumid and moderately seasonal mesic in the mountains below about 1400 m. The higher parts of the range are humid and mesic to frigid. A pronounced rainshadow is present on the northern Tectonic Activity Classes of Mountain Fronts 157 11 s 117' Figure 4.32 Pediments of the tectonically inactive western Mojave Desert. A. Percentage of exposed pediment in this arid landscape. Dark gray shows area of most common exposure. lee side of the mountains, which declines to the arid western Mojave Desert. Long dry summers are followed by mild wet winters. Less than 200 mm of precipitation may fall at a mountain weather station one winter and 1000 mm may fall the next. Years with frequent winter rainstorms are remembered for their devastating floods and debris flows. The resulting landscape of this rapidly rising mountain range has steep, bouldery braided stream channels resembling those of the Southern Alps of New Zealand. The San Gabriel Mountains fronts were analyzed using five tectonic activity classes. Data for mountain-front sinuosity and triangular-facet stages were adequate, but values for the valley floor width-valley height ratio (Fig. 4.33) had considerable overlap between tectonic activity classes (Karen Demsey, 1986 ,written communication). This scatter is the result of: spatial variations in rock resistance that affects rates of valley-floor widening, and a large range of drainage-basin areas upstream from measurement sites. Small watersheds with low annual unit stream power have downcut less in response to late Quaternary uplift. Consequently they have less relief between valley floors and adjacent ridgecrests than do adjacent large watersheds. The tendency for narrow valley floors of small watersheds is partially offset by lower values of valley floor-ridgecrest relief. Figure 4.32 Pediments of the tectonically inactive western Mojave Desert. F3. Low-sun-angle aerial view of bedrock knobs rising slightly above the surface of a pediment dome of a former mountain range. Two-lane dirt road for scale. 100 100 1000 2 Drainage basin area, Ad, in km Figure 4.33 Relation of drainage-basin area to drainage-basin length for watersheds draining the south side of the San Gabriel Mountains, California, east of Big Tujunga Canyon. Filled points Indicate the narrow range of drainage-basin sizes used to assess valley floor width-valley height ratios. 158 Chap Variation in ratios caused by both factors was reduced in two ways. First, valley-floor widths were measured at basin-position coordinates of 0.9. This reduced the number of sites where valley-floor width was anomalously wide because of soft, crushed rocks in the range-bounding thrust-fault zone. Secondly, we used only the narrow range of watershed size noted in Figure 4.33. The constants in the power function, Af, drainage-basin length (straight-line length, not the length along the sinuous valley floor), and Ad, drainage-basin area A =0.25 A °A52 I a (4.10) reflect the tectonic environment. Long, narrow drainage basins are indicative of the first stages of watershed development on an actively rising escarp- Figure 4.34 Contrasts in mountain-front sinuosity for active and inactive tectonic settings. The bedrock in both cases is soft Pliocene marine sedimentary rocks. Bedrock-alluvium contacts are from the geologic map by Barrows et al. (1973). See Figure 4.35 for locations of front numbers. A. Inactive front 2B. B. Active front 7C. ter 4 ment. Initial stages of watershed development have lower coefficients and exponents than more circular watersheds indicative of an older or of a less tectonically inactive mountain range, such as John Hacks study areas in the Appalachian Mountains. The exponent of 0.45 is less than 0.5, which would describe geometric similarity (Vacher, 1999) where all length dimensions increase in the same way as they change size in the downstream direction. Sinuosity varies greatly with age and rate of uplift of mountain fronts. Thrust faults have migrated basinward during the Quaternary, and the outermost fault is the most active at fronts 5, 7, 9, 16, 18, 19, 20, 23, and 28 (See Figure 4.35). Range-bounding fault zone 7C ruptured in the 1971 Mw magnitude 6.8 San Fernando earthquake (Grantz et al. 1971). The youngest fronts are the external (piedmont) faults of the San Gabriel Mountains. They are straight with sinuosities close to 1.0. Range-bounding faults have sinuosities ranging from 1.2 to >10. Sinuosities could be evaluated for only a few internal faults because these former mountain-piedmont junction transition zones have been converted into mountainous terrain. The contrast in sinuosity between class 1 and class 4 fronts is dramatically illustrated in Figure 4.34 where climate and rock type are the same, but uplift rates are much different. Comparison of relative uplift rates should be considered using a late Quaternary temporal framework because the tectonic signatures in this landscape result from more than 50 ky of uplift. Uplift is rapid for most of the 60 fronts of Figure 4.35. Estimated uplift rates for class 1 fronts range from about 0.5 to >2.0 m/ky and Holocene fault scarps are common. Combined uplift along several overlapping thrust faults suggests that parts of the range are rising faster than 2 m/ky (McFadden et al., 1982; Morton and Matti, 1987). The most active faults are north of the San Fernando Valley and west of where the San Jacinto and San Andreas faults pass through the Transverse Ranges. The fronts between BigTujunga and San Antonio Canyons (west-central part of range) appear to be less active. These are class 2 and class 3 fronts and only one Holocene surface rupture appears to have occurred (Fig. 3.23). Front 8 not only has degraded triangular facets but also is the only front with anomalously low stream-gradient indices (Keller and Rockwell, 1984). The range-bounding fault is not always the most active (Fig. 4.36). Front 2B is highly sinuous Tectonic Activity Classes of Mountain Fronts Figure 4.35 Tectonic activity classes of mountain fronts along the south side of the San Gabriel Mountains of southern California. The fronts are numbered consecutively from west to east. (Fig. 4.34A) and the valley floor width-valley height ratios for Limekiln Wash are large. Note the abrupt increase of V^ratios just downstream from the active Santa Susana fault zone. The footwall of the Santa Susana fault was raised during the 1994 Northridge earthquake (Yeats, 2001). Dip-slip rates are about 6 ± 4 mm/yr on the Santa Susana fault (Huftile and Yeats, 1996). Small values of ratios in the watershed headwater may be in part a function of stream size as noted in the discussion of Figure 4.33. The piedmont foreland alluvium is old. These gravels were deposited as thin alluvial fans before uplift of front 2B caused dissection of the block between fronts 2A and 2B. Uplift rates from ongoing geodetic surveys may differ from the rates suggested by tectonic geo-morphology studies, which focus on identification of persistent long-term uplift. The mere presence of Figure 4.36 Active internal and inactive range-bounding faults of the Santa Susana Mountains. Mountain-front locations are on Figure 4.35. Valley floor width-valley height ratios are in bold numbers along Limekiln Wash. Flat-topped ridges of the piedmont foreland are capped by remnants of alluvial-fan deposits. i IE t _i T ft ! i V I Kin 1 M , \ 4.4 „ge bounding front 20 _ _ t -kJ J_ 160 Chapter 4 low fault scarps in Holocene piedmont alluvium records recent shifts in the locations of active faulting. We should also remember that active thrust faulting is largely hidden beneath the deposits of the Los Angeles basin (Hauksson et al., 1988; Davis et al., 1989; Scientists, 1994; Mori et al., 1995; Hudnut et al., 1996). Many active faults are relatively narrow linear zones of base-level fall. Complications resulting from secondary antithetic and synthetic faulting may create broad fault zones. Appraisal of landscape response is even more difficult in broad areas of minor uplift. Such upwarps are more akin to areas of regional uplift, and their diffuse border locations of base-level fall may be difficult to discern. Landscape responses to folding are more difficult to evaluate than responses involving single faults. ^ Stream poweri components QReeieting power components 0 20 AO 60 &0 IOC Jasin-position coordinate 0 20 AO 60 bO 100 Jasin-position coordinate Threshold of critical power 1.1 1.2 1.1 1.0 0.9 03 Anticline Syncline Distance Strength of tectonically induced base-level fall Distance Increase of aggradation induced base-level rise Folds, from a tectonic geomorphology perspective, are broad bands of tectonic base-level change between the fold axes (Fig. 4.37A). Magnitudes of cumulative tectonic base-level fall increase with distance away from the axis of an anticline (Figs. 4.37A, D). Rate of increase of base-level fall increases downstream, attains a maximum at the inflection point of the fold curve (location TIP of Figure 4.37A), and then decreases to the axis of the adjacent syncline. Rock type and climate determine where erosion and deposition occur in landscapes being deformed by folding. Drainage basins underlain by resistant rocks with low sediment yields in a humid Figure 4.37 Tectonic perturbations in the folded Kettleman Hills landscape. A. Diagrammatic section of a folded fluvial system. Deposition prevails downstream and erosion predominates upstream from the threshold-Intersection point at TIP, which in this case coincides with the inflection point of the fold curve. F3. Downstream trends in factors affecting stream power between 5 and TIP on Figure 4.37A. Flow gradient becomes more gentle with increase of discharge and steeper as anticlinal folding Increases valley-floor slope. Actual stream gradient is a function of these and other Interacting variables. C. Downstream trends In factors affecting resisting power [see Section 2.3, Figure 2.13) between 5 and TIP on Figure 4.37A. Hydraulic roughness presumably decreases as flow depth Increases and size of streambed alluvium decreases. Channel and hillslope incision Increases bedload transport rate sufficiently to change the mode of fluvial system operation from degradation to aggradation at the point TIP the threshold-intersection point. D. Steepening of the valley floors of consequent streams on the flanks of the anticline Increases slowly near the flat anticline axis and is a maximum at the fold-inflection point at TIP E. Base-level rise resulting from piedmont aggradation is largest at point V. Pate of thickness increase becomes less nearer the axis of the syncline. Tectonic Activity Classes Figure 4.35 Topographic map showing the abrupt, straight mountain-piedmont junction of the North Dome of the Kettleman Hills anticline. Location is on the eastern flank of the Diablo Range, about 2 km north of the town of Avenal and just southeast of the Kings-Fresno county line (upper left). 25 foot (7.6 m) contour- Interval. Altitudes are in feet on this 1930 survey by the U.S. Geological Survey. setting have streams that flush sediment into the intervening synclinal valleys and then downstream. Piedmont deposition may be minimal. Arid settings and soft rocks favor deposition, especially where a piedmont of coalescing alluvial fans creates a base-level rise adjacent to rising folds (Figs. 4.37, 4.38). The proportion of an actively folding landscape exposed to erosion is less than in a humid setting. Interactions of folding, erosion, and deposition along the western margin of the San Joaquin Valley have allowed deposition by consequent streams to occur well up on the flanks of folds, perhaps to the fold-inflection point (TIP on Fig. 4.37A). However, the transition from mountains to piedmont occurs at the same place in adjacent large and small watersheds. The mountain-piedmont junction (and the reference length, L., of equation 4.1) is straight or gently curving in these class 1 landscapes. Base-level rise exerted by piedmont deposition is significant, even where adjacent watersheds have greatly different unit stream power. of Mountain Fronts \ g \ The behavior of fluvial systems, such as that diagrammed in Figure 4.37, is best understood using the threshold of critical power (see Section 2.3 and Figure 2.13). Rates of stream-channel downcutting into a rising fold are largely a function of flow discharge and hydraulic slope, both of which increase between locations S and TIP of Figure 4.37A. Offsetting this tendency for streams to downcut are resisting power factors such as hydraulic roughness and bedload transport rate. Bedload input from the hill-slopes increases greatly downstream from the headwater divide. Erosion is so minimal in the headwater of the Kettleman Hills that the shape of the growing anticline is visible when looking across the accordant ridgecrests. This is why the assigned stream power/ resisting power ratio is only 1.1 in Figure 4.37A. Introduction of gravelly bedload increases quickly as flow discharge increases downstream to erode deep gorges - the departure from threshold of critical power indicated by a relative value of 1.2 on Figure 4.37A This is an exponential increase of sediment yield from the mid- to mid-upper part of a non-steady-state landscape. This trend means that eventually the unit stream power of these occasional ephemeral flows is no longer capable of transporting the entire sediment load. The coarsest fraction is deposited downstream from the threshold-intersection point, indicated by the short reach of no departure from a value of 1.0. The mode of operation switches ever more strongly to deposition downstream from the reach that is neither aggrading nor degrading. Deposition of permeable basin fill is a self-enhancing feedback mechanism. Increased stream-bed infiltration of water from ephemeral stream-flows, compared to bedrock channel floors in the mountains, further decreases stream power. The assigned stream power/resisting power ratio decreases to 0.9, then 0.8. Basin-fill aggradation is a base-level rise that tends to offset tectonic base-level fall or rise caused by subsurface folding. Piedmont deposition maintains a straight mountain front regardless of variable drainage-basin area. Folding can encroach from steep, rising mountain fronts into the fanhead areas of adjacent piedmonts (Keller et al., 2000). This can change the threshold of critical power values near crests of anticlines, thereby decreasing rates of deposition or initiating degradation. Dispersion of uplift by fold migration and propagation may result in thinner alluvial fans next to the mountain front. This does not indicate a lesser degree of relative tectonic activity. Field studies 162 Chap may be needed to show that such cases deserve a class 1 instead of a class 2 rating because the mountain front is in the process of migrating into the adjacent basin. The rapidly rising eastern margin of the Coast Ranges of central California is a good place to apply the tectonic-activity-classes approach to folded mountain fronts (Fig. 4.39). This fold-and-thrust belt is near the plate bounding San Andreas fault, about 40-60 km to the west. These monoclines and anticlines of the late Quaternary foothill belt of the Diablo Range are encroaching on the San Joaquin Valley as the result of northeast-southwest compression. We presume that mountain fronts along the western and southern margins of the San Joaquin Valley have blind thrust faults at depth (Namson, and ter 4 Davis, 1988, Wentworth and Zoback, 1989; Namson et al., 1990; Wakabayashi and Smith, 1994; Keller et al., 1998). Earthquakes occur here. The first of a northwest-to-southeast sequence of earthquakes on the same 100-km long hidden thrust fault (Stein and Yeats, 1989; Stein and Ekstrom, 1992; Lin and Stein, 2006) was the Mw magnitude 5.4 New Idria earthquake of 1982. The Mw magnitude 6.5 Coalinga earthquake in 1983 (Hill, 1984; Rymer and Ellsworth, 1990) emanated from beneath Anticline Ridge, and the Mw magnitude 6.1 earthquake of 1985 from beneath the north end of the Kettleman Hills (Wentworth et al., 1983). The crest of Anticline Ridge rose abruptly about 500 mm during the 1983 earthquake Mountain front tectonic activity classes 1A 2 3 4 Relative uplift rate 2D km Figure 4.39 Tectonic activity classes of folded mountain fronts along the eastern margin of the Coast Ranges in central California. AR, Anticline Ridge; CH, Ciervo Hills; CH, Coalinga Hose; K.H, kettleman Hills; K.Y, Kreyenhagen Hills; LGC, Los Gatos Creek; L5H, Laguna Seca Hills; Nl, New Idria; PC, Panoche Creek, PH, Panoche Hills; RR, Reef Ridge. Tectonic Activity Classes (King and Stein, 1983). In hindsight, I now realize that comparison of first-order level-line surveys across Anticline Ridge in March, 1962 and March, 1963 recorded a precursor event to the destructive Coalinga earthquake. Benchmarks rose as much as 24 mm in a pattern that mimicked the topographic profile of Anticline Ridge (Bull, 1975b, Fig. 4). These "hidden earthquakes" (Stein and Yeats, 1989) underscore the continuing active nature of thrust faults in the cores of these folds. This landscape reflects rapid, continuing uplift. The predominance of Class 1 mountain fronts (Fig. 4.39) describes a highly active tectonic environment, despite the deceptive appearance of smooth, rolling hills underlain by soft marine sedimentary rocks. Large antecedent streams emerge from structural lows between rising folds (Panoche Creek), or are superimposed on rising anticlines (Los Gatos Creek). Even though the larger streams have intermittent flow characteristics, they easily bevel broad strath surfaces in the soft Tertiary marine mudstone and sandstone upstream from the mountain fronts. Alluvial fans downstream from these fronts have increased in thickness by as much as 700 m thick since deposition of the Corcoran Lake Clay member of the Tulare Formation. Sanidine crystals from a tuff in the upper part of the Corcoran have a K7A age of -625 ka (Janda, 1965). This suggests that mountain-front uplift rates exceed 1 m/ky. Pliocene—Pleistocene stratigraphic formations have been folded on all of the Class 1 range-bounding faults. Stein and Yeats of Mountain Fronts \ 63 note "The youngest folded sediments are less than two million years old, which indicates that the fold began to form since that time. If a 75-centimeter growth is typical for events at the Coalinga site, then Anticline Ridge could have been built by roughly 1,000 ancestral earthquakes of magnitudes from M = 6 to M = 7 recurring every 1,000 to 2,000 years." Class 2 mountain fronts have a less obvious juncture with the alluvial-fan piedmont, and mountain-piedmont junction sinuosity increases greatly. Much broader valleys extend upstream from the mountain front. Class 3 and 4 mountain fronts have another notable characteristic. Differential erosion of the mix of sandstone and mudstone beds results in a mountainous landscape where hard strata form prominent ridgecrests, and soft rocks underlie the broad valley floors of even small tributary streams. Reef Ridge in the southern Kreyenhagen Hills is an example. Class 1 mountain fronts are being raised so fast that soft mudstone is at the same position in the landscape as the somewhat harder sandstone. Note that the tectonic activity in the Kings County part of Figure 4.39 appears to have shifted eastward with the thrust-fault propagation that resulted in the formation of the Kettleman Hills. The influence of uplift on landslide distribution is illustrated by the semiarid Arroyo Ciervo fluvial system in the Ciervo Hills (Figs. 4.40, 4.41). This drainage basin is underlain by 60% diatomaceous shale, 8% mudstone and clay, and 32% soft sand- Landslide Monocline Figure 4.40 Distribution of landslides upstream from tectonically active Monocline Ridge in the Arroyo Ciervo drainage basin, located at CH in Figure 4.39. Oldest landslides are at F3 and youngest are at C. Area A has yet to experience the pulse of renewed landslide activity. Chapter 4 Figure 4.41 South fork of Arroyo Clervo where landslides are encroaching Into the stream channel from both sides of the valley. stone and sand. Soft clay-rich rocks occur throughout the basin but landslides are not common near the mouth or near the headwater. Mean annual (wintertime) precipitation increases with altitude from about 110 to 360 mm. An active monocline at the basin mouth is so young that uplift has yet to create high local relief (see Figure 4.3D for 1.5 m contour-interval map of this front), but steep slopes and 300 m high hills are common in the shale and sandstone upstream from tectonically active Monocline Ridge, which is the class 1 internal front on Figure 4.40. Recent accelerated valley-floor downcutting has caused many landslides that occupy 8% of the watershed. Arroyo Ciervo provides a nice example of how uplift interacts with fluvial processes to influence locations of landslides. Tectonically induced downcutting caused by presumed recent uplift of Monocline Ridge has extended to within 2 km of the headwater divide. Area A (Fig. 4.40) lacks landslides even though the rock type is the same diatomaceous shale as in downstream reaches. Mass movements consist mainly of young to active footslope slumps (Fig. 4.41) that encroach into the two trunk stream channels. In contrast, landslides in that part of the fluvial system immediately upstream from Monocline Ridge (location B) occur higher on midslopes or on footslopes adjacent to tributary streams. The location B landslides are a mixture of ages; some are only a few decades old, but others are so old that the typical hummocky topography has become quite subdued. Tributary landsliding has just begun as a few small slumps in area C. This landslide succession illustrates the substantial reaction time for increased incidence of landslides as the effects of a tectonic perturbation are transmitted by the stream subsystem to progressively more distant hillslopes. 4.3 Summary Future work should focus on upgrading the tectonic-activity-class model, which compares relative mountain-front uplift rates. Ideally, we should be making quantitative estimates of total Quaternary tectonic displacement for a mountain front, and of how rates of faulting and folding have changed with time. Such advances require improved understanding of how lithology and structure, and temporal and spatial variations in climate, affect rates of landscape denudation. The roles of these independent variables are fundamental to all geomorphology. Defining mountain front uplift rates requires descriptions of stream-channel downcutting rates in the mountains and piedmont aggradation rates in order to understand base level controls in those reaches of fluvial systems that cross range-bounding faults. Better dating is essential. Advances in optical and thermoluminescence now are being applied to diverse stratigraphic settings. Terrestrial cosmogenic nuclides now rival standard radiocarbon dating because of their ability to estimate ages of both surfaces and deposits. Mountain-front denudation is more varied and complex than for low piedmont fault scarps because larger scale makes for more diverse lithologic and climatic controls, and because of uplift may emanate from multiple zones of tectonic base-level fall. The need for better understanding of lithologic and climatic controls of mountain front tectonic activity classes has a counterpart in diffusion-equation modeling of fault scarps. We now turn our attention to such low, comparatively simple mountain fronts. Fault scarps intrigue earth scientists who use both landscapes and stratigraphy to better understand earthquake hazards. In contrast to the multitude of surface rupture events needed to create an impressive mountain front, fault-scarp studies emphasize the size of the most recent surface rupture and when it occurred. Tectonic geomorphologists study fault scarps located at mountain-piedmont junctions and on the adjacent piedmonts. Humans use both sites extensively. These simple landforms have much to offer modelers of landscape evolution as they seek to better understand hillslope erosion and the genesis of mountain fronts. A fault scarp may record the latest of many surface ruptures that have created an impressive mountain front, or it may represent its mere beginnings where thrust faulting has propagated out onto a piedmont. Fault scarps are a mainstay of paleoseis-mology investigations. All the previous chapters have incorporated selected aspects of fault scarps (Figs. 1.9, 2.12, 3.17, 3.23, 3.27-3.29, 4.9-4.11, 4.23, 4.29). 5.1 General Features I introduce useful terms and concepts and then consider independent variables affecting diffusion-equation modeling of scarps. These include climatic and lithologic controls. An example of fault-scarp processes with an earthquake recurrence interval of >200 ky is presented to balance the emphasis on Holocene fault scarps. The advent of cosmogenic dating creates intriguing possibilities for breakthough studies of fault scarps in alluvium. Cosmogenic-isotope studies of smooth limestone fault planes hold much promise for future studies of bedrock fault scarps. LiDAR (Light Distance and Ranging) radar provides exquisite details of surfaces ruptured by recent faulting. Uplift along range-bounding faults adds fresh increments to scarps that rupture bedrock but these can be difficult to separate from former increments and analyze in a quantitative manner because of the 1) great diversity of lithologies and structures, and 2) the steep and irregular slopes of mountain-front escarpments are the product of many surface ruptures. Single-event surface ruptures in alluvium that create simple convex-concave curves are much easier to model. Piedmont fault scarps are the main topic of this chapter because: 1) The surfaces of alluvial fans and treads of stream terraces provide geometrically simple shapes, such as segments of cones and planar surfaces (Section 3.3.4) that are ideal for analyzing tectonic deformation and consequent erosional surface-rupture landforms. 2) Stream terraces and fanhead embayments cross range-bounding faults. Faulting of these planar surfaces records surface-rupture characteristics of one or several prehistorical earthquakes. The ages of adjacent faulted and unfaulted surfaces in a flight of stream terraces bracket times of most recent faulting. 3) The ubiquitous sandy gravels of the piedmonts of the western North America provide an approximately uniform material for studies of fault-scarp denudation and for defining chronosequences of soil profiles of flights of faulted and unfaulted stream terraces. An exceedingly long time span elapsed between the penultimate and 1887 surface ruptures of the Pitaycachi fault, Mexico. This gently sloping hill has no obvious previous surface rupture, even though it too created a 3 to 4 m high fault scarp. 166 Chapter 5 4) Many piedmonts have a great variety of alluvial-surface ages. Offsets of young and old stream-terrace treads can be compared. Cumulative displacement of older surfaces produces multiple-rupture event fault scarps. Paleoseismologists seek to estimate characteristic earthquake and surface-rupture size and displacement (Section 3.3). 5) Alluvial fans and terrace deposits may contain organic materials for radiocarbon dating, volcanic ash that can be correlated with previously described and dated tephra, fluvial and eolian silt for luminescence dating, and depth-dependent systematic trends of terrestrial cosmogenic nuclides. Surficial boulders can be surface-exposure dated with terrestrial cosmogenic nuclides and weathering rinds. 6) Diffusion-equation modeling of changes in piedmont fault-scarp morphologies is a standard paleo-seismology technique. Rates of degradation of unce-mented gravelly deposits are sufficiently rapid to allow differentiation of scarps formed by single-rupture events during the late, middle, or early Holocene. 7) Stream terraces and alluvial fans are readily accessible sites for excavation of trenches to study faulted and unfaulted strata that record earthquake recurrence intervals, liquefaction features, colluvial wedges, paleosols, and fault-pond deposits (Sieh, 1978b; Weber and Cotton, 1981, McGill and Sieh, 1993; Pantosti et al., 1993a, b; McCalpin, 1996; Machette et al., 1992, 2005; Machette and Brown, 1995). Many fault scarps have undergone minimal dissection, which suggests a late Quaternary surface rupture. More than 10,000 km of active faults occur just in the Great Basin part of the Basin and Range Province (Wallace, 1977). Most have low scarps in unconsolidated piedmont alluvium, or in colluvium at the mountain-piedmont junction. Fault scarps provide a convenient way of describing late Quaternary segmentation of faults (Section 3.3.4). Features such as scarp height, degradation slope morphology, and multiple-rupture-event crests of late Pleistocene and Holocene scarps identify differences between adjacent fault segments (Machette, 1986; Crone and Haller, 1989; Turko and Knuepfer, 1991). These segments can be classed as persistent or nonpersistent where compared to the tectonic activity classes of the adjacent mountain front (Section 4.2.3.2). By analyzing piedmont fault scarps in 17,000 km2 of central Nevada, Wallace (1977) estimated that about seven earthquakes of Mw magnitude 7 to 8 have occurred during the past 12 ka. The 1915 earthquake in Dixie Valley created much-studied fault scarps. Historical fault scarps in arid regions tend to have prominent free faces (steep cliffy part) and narrow crests and bases (Fig. 5.1A). The skylined Upslope original ace^ Penultimate scarp 1915 Scarp ^^^tak Grabe Downslope original surface Figure 5.1 Surface ruptures along the range-bounding fault of the Tobin Range, Pleasant Valley, west-central Nevada. A. This 3-m high scarp was formed by the 1915 earthquake. Left side of view is gullied and scarp-crest sinuosity has increased because of more runoff from hillslope than on the right side of view. F3. Faulted fanhead embayment at Figure 5.1A site. The convex ridgecrest gradually decreases in slope upslope from the graben in the faulted alluvium. This preceding (penultimate) fault scarp now appears as a bevel adjacent to the free face of the 1915 surface rupture. Truck for scale. Fault Scarps 167 scarp (Fig. 5. IB) is the result of two surface ruptures. Wallace found that topographic profiles normal to the fault trace that described scarp steepness and crest sharpness were particularly useful guides to the approximate times of late Quaternary faulting (Figs. 5.2, 5.3). Substantial progress has been made in understanding the evolution of fault scarps. Diffusion-equation modeling estimates times of formation of scarps with easily described topographic shapes. Pioneering studies by David Nash emphasized simple situations. Single-rupture-event scarps created by normal faulting of nearly planar alluvial surfaces are ideal for modeling. Section 5.2 explores the subject of how scarps become progressively less steep and the crest more rounded with the passage of time. Complex scarps, such as those created by multiple thrust (Fig. 1.9) or normal faults, are not amenable to modeling. Multiple fault scarps are a composite of overlapping degradation and aggradation. The resulting gently sloping scarp is a composite topographic profile that appears anomalously old. Fault-scarp terminology emphasizes processes that create distinctive landforms (Figs. 5.2, 5.3). Rupture of a smooth alluvial surface by a normal fault is a tectonic perturbation that abruptly increases local relief and creates a low linear ridge. This new hillslope changes shape with time. The fault scarp initially consists of a prominent free face (erosional part of scarp steeper than the angle of repose for loose sandy gravel) and an adjacent debris slope (area of colluvial deposition). The debris slope is initially formed mainly by slumps and rockfalls from the free face (Fig. 5.1 A). The angle of repose for uncohesive materials ranges from 28° to 37°. The free face retreats upslope and quickly disappears. It may be regarded as a weathering-limited slope whose rate of roughly parallel erosional retreat is largely controlled by the rates at which materials are made available for transport by geomorphic processes of rainsplash, slopewash, and creep (Gilbert, 1877). A sharp crestal break in slope (Fig. 5.2) persists as long as the topographic discontinuity of a retreating free face remains. The crest and base define a new hillslope that ranges from 40° to 70° for 10-year old scarps and 3°-10° for 200 ky old scarps. A scarp that no longer has a free face may be regarded as a transport-limited slopewhose rate of overall slope decline is largely controlled by the ability of erosional processes to remove weakly cohesive soil and alluvium. Fluvial processes broaden the crest slope or scarp crest (convex slope element between Upslope original surface Crest Free face~ Debris slope Wash slope Base —13 m Fault zone Downslope original surface Figure 5.2 Topographic elements of a diagrammatic single-rupture event fault ecarp. From Wallace (1977). Each element has a characteristic morphology formed by different processes. the upslope original surface and free face) and wash slope (concave area of fluvial deposition that buries the downslope original surface to form a new alluvial slope). Readers will note a variety of terms used for magnitude of fault-scarp tectonic deformation. Most terms are only approximations of the actual throw as used in a structural geology sense. Scarp height is visibly obvious, but is larger than throw for sloping alluvial surfaces. Projected net separation of the topographic profile across a fault zone (Fig. 3.27) may be close to representing the true displacement but should be regarded as an apparent throw unless corrected for land surface and fault plane dips. Common synonyms for apparent throw include "scarp offset" and "2a", where a is half the apparent throw above and below the scarp-profile midpoint. Piedmont fault scarps are like range-bounding faults in that they separate reaches of tectonically induced aggradation and degradation. After each surface rupture, the consequences of the tectonic perturbation spread to the adjacent reaches of the fluvial system. Tectonically induced downcutting occurs rapidly in larger antecedent streams (streams that were present before faulting), but smaller consequent streams (whose channels form as result of the uplift event) also begin to erode headward into newly created scarps (Fig. 5.1 A). Sediment derived from 168 Chap Age ranges for scarps in sandy gravel Fault scarps Fault controlled 5° "Oky~~ 4>/f * t t i * J— * f 1 Gravity and i ( debris ^Free face Pebris slope 5°-10° ^ 0.1 - 0.5 ky WasrTslope Pebris-controlled 0.3 -1 ky ■ Pebris and wash-controlled 10 - 50 ky s^p Wash-controlled 50-100ky Figure 5.3 Sketches showing stages in the sequence of degradation of a typical normal fault scarp in alluvium. Each dashed gray line represents the solid line of the previous stage. From Wallace (1977). stream-channel incision is deposited on alluvial fans that radiate downslope from threshold-intersection points on the new alluvial slope. From a geomorphic sense, a new fault scarp is indeed the creation of a new mountain front. Scarp geomorphic elements may expand in area and change altitude, which changes fault-scarp height (Figs. 5.3, 5.9). Scarp height x& generally measured as the vertical separation between obvious topographic breaks in slope termed the base and crest. Geomorphic responses to faulting extend much farther, which is obvious when the pre-surface rupture ter 5 soil profile is examined. Truncation of a formerly present A soil-profile horizon upslope from the crest reveals where rates of erosion now exceed rates of soil-profile formation. Extent of recent burial of the soil profile downslope from the debris and wash slope defines the limits of the new alluvial slopewhere technically induced aggradation now is occurring. Fault scarp terminology emphasizes processes that create distinctive landforms when piedmont alluvium is ruptured. The terms have a strong genetic connotation (Wood, 1942; Young, 1972; Cooke and Warren, 1973). The same types of processes that shape larger hillslopes control the slope elements and degradational stages of a single-rupture event fault scarp (Figs. 5.2, 5.3). The term "crest slope" implies dominance of rainsplash and sheetflow processes, the "free face" gravity-controlled erosional processes, the "debris slope" a combination of gravity and fluvial depositional processes, and the "wash slope" fluvial processes. These genetic terms are quite appropriate for new hillslopes abruptly created by faulting. However, additional changes in the new landforms of a ruptured original surface also need to be considered. Crestal convexity and wash slope concavity decrease with time at similar or different rates. The original surface commonly is degraded near the crest, buried near the base, and absent between the fault trace and the crest (Fig. 5.4). The upslope original surface may be subdivided into eroded and uneroded elements. Significant amounts of crestal rounding may occur even when a retreating free face is still present. Although the magnitude of surficial lowering is minor compared to the magnitude of free-face retreat, one or more soil-profile horizons may be removed. A fairly typical example is the Drum Mountain fault scarps in western Utah (Bucknam and Anderson, 1979); their age estimate is 7-10 ka and the original surface seems to have been erosionally steepened by about 1° in the upslope 4 m of the scarp crest. The upslope eroded original surface has been modified by minor erosion. The presence of a buried or unburied soil profile is helpful in defining the position of the downslope original surface so that estimates of surface displacement can be made. Thin A soil-profile horizons or the boundary between the B and C horizons are useful approximations of planes that parallel the original alluvial geomorphic surface prior to offset by faulting. The presence of an A horizon usually is a good indication of minimal modification of the original surface by fluvial processes. This use of soil-pro- Fault file horizonation features avoids the pitfalls of 1) estimating displacement of the land surface where there has been erosion of and/or deposition on the original surface and 2) variations in thickness and dip of faulted sedimentary strata, both of which may introduce unknown amounts of error into estimates of vertical fault slip. Describing the soil-profile characteristics in the context of a soils chronosequence constrains the time of the surface-rupture event. Examples of using soils in studies of faulted surfaces and stratigraphy of colluvial wedges include studies by Chadwick et al., 1984; Hecker, 1985, 1993; Demsey, 1987; Forman et al., 1991; Machette et al., 1992, Amit et al., 1995, 1996, 2002; Enzel et al., 1996; McCalpin, 1996, Rubin et al., 1998, Birkeland, 1999; Lindvall et al., 2000; and Phillips et al., 2001. Erosional terminology should not be used for slope elements downslope from the fault trace because the lower segments of fault scarps typically are depositional areas. The debris slope that is upslope from the fault trace is mainly a slope of transportation or is being denuded, although it may have a temporary veneer of detritus on it. Downslope from the fault trace, the debris slope is aggradational and merges with the depositional piedmont of the new alluvial slope. The new alluvial slope consists of areas of sheetflow deposition and may include small, actively aggrading alluvial fans. The threshold-intersection points represented by the fan apexes tend to Upslope original surface Eroded \ Uneroded Greet Downslope Newalluvial o original slope Debris slope <ü surface Scarp height - « 05 CD M J-'„Ö ^Tjß o ^■ ' 0 '. o' Q (?' a 1 a '-a-^ •' ' a '-a-^ •' ' a ■ . ■ . . 9 ■ ^ o •=> . . ■ . {?- <=,'■'■=>•.° o- <=> ■ ^ a. c " o. ^ C -6'. ö c ■'■ a ■■a-"'-& ■ ■ Ö 0 Ci. ■. t= 3B.'?->o :■:■.■ • -9:'=>.■ ■ 9-^'o c Pleistocene alluvial-fan deposits Figure 5.4 Fault scarp terminology and landforms as related to the pre-rupture event soil profile. Scarps 169 be downslope from the base of the debris slope for antecedent streams, and coincide with the base of the debris slope for consequent streams. Surface rupture, even in the form of 1 m fault scarps, affects local base-level processes as described in Figure 4.7. Part of the downslope original surface is shown in Figure 5.4. This part generally is buried except where the scarp is so young that depositional processes have yet to form an extensive new alluvial slope. The length and height of historical fault scarps provide general information about earthquake size, but scarp height rarely is the same as throw (vertical fault displacement). Scarp height equals displacement for vertical faults that displace horizontal surfaces. Otherwise scarp height exceeds throw. Scarp height may also be less than vertical displacement where a faulted horizontal surface is raised by aggradation on the downthrown surface. Several factors cause anomalous scarp heights. The difference between scarp height and displacement increases with decrease in dip of normal faults, increase in slope of an original surface that slopes in the same direction as the fault, downslope migration of the scarp base, and upslope migration of the scarp crest. A common procedure that is illustrated in Figure 3.27 is to use projections of the upslope and downslope parts of the displaced original surface to the estimated position of the fault trace to make an uncorrected estimate of throw. Wallace (1980) devised nomograms for determining the components of slip when scarp height and slopes of the fault plane and original surface are known (Fig. 3.30). Additional complications to extrapolating surface slopes in order to estimate fault displacements (Section 3.3.4, Fig. 1.9) include 1) re-faulting depositional surfaces that change slope gradually or abruptly, 2) rotation of one or both of the adjacent fault blocks so that pre-faulting surface slopes are changed, and 3) movements along multiple faults. Earth deformation may be concentrated within a few meters of the fault, but commonly it is dispersed within a 10 to 100 m band of shearing and is different in the hanging wall and footwall blocks of normal faults. Fault-zone stratigraphy may define multiple ruptures (Figs. 3.23 and 4.29), give clues regarding frequency of rupture and cumulative displacement, and show how tectonic deformation influences fault-scarp morphology. Several significant geomorphic, structural, and stratigraphic features of normal faulting are illustrated by the unusually complete exposure shown 170 Chapter 5 1<3<37 scarp Holocene Terrace I 4 I 1V| Co I lu vi um Stream Channel Oriented clasts in fault zone Gray Water-Laid Alluvium Holocene qq1^ Coarse gravel Pleistocene Medium gravel Red Clayey Alluvium Pleistocene Medium grained water laid gravel Debris-flow deposit Scale in meters 1 2 _l_ J Fine grave Pebbly sand Fault Fracture, open or partly filled with orqanic matter Boulder Figure 5.5 Cross section of Pitaycachi normal fault and alluvial deposits exposed in south bank of Arroyo Hondo, northern 5onora, Mexico. Fault planes of the 1557 and mid-Pleistocene surface ruptures are shown. Figure 6 of Bull and Pearthree, 1955. in Figure 5.5. A 60° to 70° west-dipping fault zone is exposed in a 6 m high streambank. Two markedly different alluvial sequences with different deforma-tional styles have been juxtaposed by faulting and are capped by bouldery Holocene gravel. Gray-brown (7.5YR 7/2; Munsell, 1992), indurated, water-laid alluvium of Plio-Pleistocene age comprises the up-thrown block where tectonic deformation is minimal. Drag-folding has not occurred and only a few closed fractures parallel the 1887 rupture plane. The other two faults and greatly differing alluvial sequences on opposite sides of the fault zone are evidence for multiple fault ruptures and large cumulative displacement. Red-brown (5YR 5/6) alluvium of mid-Pleistocene age exposed in the downthrown block consists of water-laid gravels and debris-flow deposits. The debris-flow matrix contains 20% silt and 22% clay. Two normal faults of the penultimate surface rupture do not penetrate into the capping Holocene alluvium, and drag-folded beds abut the faults. Fault Scarps 171 The downthrown block also has numerous vertical fissures, largely filled with loose sand and fresh-appearing organic debris, that do not extend from the cohesive reddish clayey Pleistocene alluvium into the overlying noncohesive sandy Holocene gravel. The structures associated with the faulting shown in Figure 5.5 represent three different styles of deformation near a major fault. Minor fractures in the upthrown block that parallel the main 1887 rupture plane were probably generated by tensional stresses during the 1887 surface-rupture event, as the basinward block moved away and down from the relatively stable "upthrown block". Drag-folding associated with Pleistocene faulting in the downthrown block may represent soft-sediment deformation of saturated deposits. The open fissures (shown in black) on the downthrown block represent a more brittle style of deformation during the 1887 event that affected cohesive dry deposits well above the water table. The orientations of these fissures are appropriate for extensional fractures (Davis, 1984, p. 272) if the dip of the fault becomes slightly less at depth. Pervasive Assuring during the 1887 event, and soft-sediment deformation during the prior event, reflect maximal extension of materials close to the fault zone on the downthrown side. They may be major causes of surface subsidence (commonly referred to as back-tilting; Fig. 1.6) near normal faults, which generally are not exposed as well as at Arroyo Hondo. Many normal faults have a single trace, but the position of the fault trace in the topographic cross section of the scarp is quite variable. The initial position of a fault trace is at the base of the newly created escarpment. Scarp-crest degradation and scarp-base aggradation move the fault trace to a more central location, especially for multiple-rupture event scarps. Multiple-rupture event fault scarps are a composite of two or more times of surface rupture. Most multiple-rupture event fault scarps have several features indicative of rejuvenated surface rupture. Renewed fault displacement may abruptly steepen the preexisting scarp slope producing a segmented topographic profile. For example, renewed normal faulting might create a scarp with 25° to 35° crest and footslopes above and below a new 70° free face (Fig. 5.IB). Multiple-rupture event fault scarps also may have multiple colluvial wedges and associated paleosols in the debris-slope stratigraphy (Figs. 3.23 and 5.38 discussed later). Scarps formed by frequent surface ruptures may not have sufficient time to create the segmented appearance of multiple-rupture event fault. Occasionally, scarp segmentation is the combination of tectonic and fluvial processes, for example faulting followed by lateral erosion induced base-level fall that steepens the basal portion of the scarp. One of the best ways to identify a multiple-rupture event fault scarp is to study stair-step flights of stream terraces that cross the fault zone (Mason et al., 2006, Fig. 2). Terraces may range in age from less than 2 ka to more than 200 ka. The youngest terrace tread may not have been faulted and the fault trace is buried by terrace alluvium. Older terraces commonly have progressively higher scarps with more complex topographic profiles. Uplift history and characteristic earthquake estimates for such a paleoseismology site can be made by estimating the age and displacement and age of each terrace in the flight. A soils chrono-sequence is an essential ingredient for success, as is having more terraces than surface-rupture events. The Buttersworth graben has multiple surface ruptures that are revealed by stream terraces, but not by the fault scarp. The graben formed on internal normal faults in a valley fill 275 m above the class 1 mountain-piedmont junction of the Stillwater Range of west-central Nevada (Fig. 4.24 and Table 4.3). Topographic profiles of the high scarp along the west side of the graben (Figs. 5.6, 5.7) suggest only one 8 to 10 m surface rupture. Profile A has a Figure 5.6 The 7 m high multiple-rupture event fault scarp at the Buttersworth graben site, Stillwater Range, west-central Nevada. The view is from the east side of the graben. Topographic profiles of the fault scavp at A and E3 are shown in Figure 5.7. The fan-head trench in which the stream terraces of Figure 5.5 are located is in the background. 172 £ 15 1.0 Figure 5.9 Horizontal and vertical changes in the boundaries between slope elements of a fault scarp, and Increases (+) and decreases (-) in scarp height, after faulting at time t . The morphology of the scarp is shown at time t, and the position of the crest upon elimination of the free face is shown at time t„. 174 Chapter 5 The hypothetical example of Figure 5.9 illustrates progressive increases in altitudes of both scarp crest and scarp base between times t and t. The net effect of these nontectonic base-level processes operating upslope and downslope from the fault trace is to increase scarp height to a value that exceeds fault throw. This trend generally is reversed after elimination of the free face. Geometry of piedmont faulting (Section 3.3.4) and nontectonic changes in scarp height (Fig. 5.9) provide many pitfalls for those tempted to use scarp height as an indicator of earthquake size. In summary, the base-level processes of vertical fault slip, scarp crest and free-face degradation, and aggradation on the new alluvial slope determine the heights of fault scarps. The downslope change from net erosion to net deposition coincides roughly with the fault trace, which is the position of the tectonic base-level fall. Thus, the location of an abrupt change in mode of fluvial-system operation is tec-tonically controlled, just as it is for class 1 mountain fronts (Figs. 4.9-4.11). Scarp height approximates uplift only at the time of displacement, and only for vertical faults that rupture horizontal surfaces. Scarp height tends to increase until the free face has been removed by erosional retreat. Then both crest degradation and aggradation of the new alluvial slope may progressively reduce scarp height. 5.2.2 Decreases in Maximum Scarp Slope Bucknam and Anderson (1979) recognized the importance of scarp height on the steepness of fault scarps of various ages. They compared maximum slope angle and height for scarps formed by piedmont faulting and lacustrine shoreline processes. Slope angles were measured for a range of scarp heights created by the same surface rupture event. Scarp steepness increases with increasing scarp height (Fig. 5.10) but at a rate dependent on scarp age. Free faces were no longer present at their three sites. The climate is moderately seasonal, semiarid, and the lithology is sandy gravel. The times of initial fault-scarp formation are approximately 1, 10, and 100 ka. The three regressions of Figure 5.10 show that maximum scarp slope becomes less with increasing age and decreasing scarp height. For a scarp 3 m high, the maximum slope angles are 27°, 18°, and 10° for the 1, 10, and 100 ka scarps. Free faces persist longer in higher scarps. Scarps that are 2 to 5 m high expose materials that are more resistant to erosion than do scarps that are less than 1 m high. Even in sandy gravel, surficial soil horizons are less resistant to erosion than the more cohesive or cemented deeper horizons. These factors together with the need to remove larger masses of materials to denude high, as compared to low, scarps may explain much of the variation in maximum scarp in -10° ' / I '\ I I I I I ......I I 0.5 1-0 5 Scarp height, m 10 Figure 5.10 Relation of maximum slope to height for Pleistocene and Holocene fault scarps in western Utah. From Bucknam and Anderson (1979). Dashed lines show the regression significance at the 95% level. Materials eroded Materials deposited ^2FT Figure 5.11 Cross sections showing nonlinear fourfold Increase of volumes of earth materials eroded from and deposited between times t and t for scarp heights of H and 2H as a linear scarp slope changes from 60° to 30°. Fault Scarps 175 Figure 5.12 Modeled stages in the evolution of a 5 m high scarp In sandy gravel In the 15 ky after the free face has raveled to the startling slope of 33.5°. The scarp profiles that would be predicted by diffusion-equation modeling at 5,10, and 15 ky use a degradation rate coefficient of 12 x 10"1 mz/ky. Figure 1 of Pierce and Colman (1956). slopes. A doubling of scarp height greatly increases the volume of materials that need to be eroded for a given change in scarp morphology (Fig. 5.11). 5.2.3 Diffusion-Equation Modeling Progressive changes in the shape of two fault scarps on different piedmonts would be virtually indistinguishable if the variables of lithology scarp height and piedmont slope, climate, and vegetation were the same. Only the factor of time would be responsible for differences in steepness and curvature of the topographic profiles of the two scarps. Such reasoning encourages modeling of fault scarp topographic profiles in order to estimate their times of surface rupture. Hold the other variables constant and model the landform evolution. Diffusion-equation modeling involves matching of surveyed topographic profiles of fault scarps with topographic profiles of different ages as modeled by an equation that treats transport-limited hillslope processes as diffusion. The example shown in Figure 5.12 illustrates the distinctly different predicted topographic profiles, and a decreasing rate of change of scarp morphology, during a 15 ky time span after elimination of the weathering-limited free face. Accuracy and precision of calibrated modeled fault-scarp ages depend on the degree of variation of the controlling variables within a study region. In a general sense, piedmont fault scarps in the western United States are 1 to 5 m high, and occur on 2° to 20° piedmont slopes of sandy gravel in a semiarid climate. This rough similarity in controlling factors is largely responsible for much of the apparent success of diffusion-equation modeling. However, we should not expect too much from diffusion-equation modeling (Mayer, 1984). It generally cannot distinguish ages of fault scarps that are separated by only 2 ky, but it commonly discriminates between surface ruptures classed as late Pleistocene, and early, middle, and late Holocene. The lack of better precision is attributable mainly to seemingly minor variations of geomorphic variables between sites. For example, sand with 25% gravel has a narrow range of alluvium mass strength, but variation in argillic and calcic soil-profile horizons results in a tenfold variability of alluvium mass strength along with major decreases in diffusivity coefficient. Such variations are sufficiently important that Sections 5.3 and 5.4 are devoted to climatic and lithologic controls of fault-scarp morphology. Diffusion-equation modeling generally cannot be used where displacement along several faults has created a multiple fault scarp, nor is it usually appropriate for study of multiple-rupture event fault scarps. Many workers have modeled scarps created by tectonic, lacustrine, and fluvial processes in the northern Basin and Range Province. Scarps formed by lateral erosion of stream channels incised into alluvium provide a great variety of formation times 176 Chap and orientations and thus are well suited to evaluate microclimatic controls (Pierce and Colman, 1986). Scarps formed by highstands or recessional stillstands of lake levels can be assumed to have formed at the same time within a given lake basin (Hanks and Wallace, 1985; Hanks and Andrews, 1989). Scarps that have resulted from nontectonic processes, such as lateral erosion induced base-level fall, do not have the displaced original surfaces that characterize tectonic fault scarps on piedmonts. Only the upslope original surface constrains the time of the lateral erosion base-level fall to a maximum age. The younger downslope surface is fluvial or lacustrine in origin. Modeling the ages of such nontectonic scarps involves an apparent displacement of the original surface. This term also is appropriate for fault scarps where the downslope original surface has been buried to create a new alluvial slope. Characteristics of scarps created during the highstands of late Pleistocene Lake Bonneville in Utah and Lake Lahontan in Nevada are used to test diffusion-equation modeling, and evaluate the importance of lithologic and scarp-height controls. Highstand scarps are notched into a variety of gravelly piedmont deposits. The scarp ages are about 14.5 ka for Lake Bonneville (Scott et al., 1983) and 13 ka for Lake Lahontan (Benson and Thompson, 1987a,b; Adams et al., 1999; Adams and Wesnousky 1999). A graph of maximum scarp slope and apparent dis- ter 5 placement (Fig. 5.13) suggests that the two datasets may be considered as a single population. Variations in lithology and microclimate result in some scatter of points. Figure 5.13 also conveys a strong dependence of maximum scarp slope on the magnitude of apparent displacement. Such tests encourage the use of diffusion-equation modeling to estimate ages of Basin and Range Province fault scarps. Diffusion-equation modeling can be used where transport of surficial materials is directly proportional to the energy gradient. Transport of sediment by flowing water can be described as a diffusion process, but gravity-controlled rockfalls from sandy gravel free faces of fault scarps cannot be considered a diffusion process because they are abrupt mass movements. So, diffusion-equation modeling can be used to describe systematic changes in fault-scarp morphology only after the free face has been eliminated,. This generally requires at least two years (Pierce and Colman, 1986) to several centuries (Wallace, 1977). Fluvial degradation of scarp crests by overland flow, raindrop splash, and wet or dry soil creep - and the concurrent aggradation of debris and wash slopes - is directly proportional to slope steepness. With the passage of time the convexity of the scarp crest, the concavity of the scarp base, and the steepness of all segments of the topographic profiles decrease (Fig. 5.12). The rate of change in altitude of any point, y, on the topographic profile of a scarp with time, f, can be described by the diffusion equation: 5 10 Apparent displacement, m Figure 5.13 Relation of maximum ecavp slope and apparent displacement for Lake Bonneville and Lake Lahontan highstand shorelines. The diffusion-equation model curve with a tc value (a constant relative ecarp age) of 16 mz implies a single regional dataset. From Hecker (1955) as modified from Hanks and Wallace (1955). by/St = c Ô2y/Sx2 (5.1) where 82y/§X2 is the slope curvature, and X and y are the horizontal and vertical coordinates of a point, and C (the diffusivity coefficient) represents the rates of fluvial degradation and aggradation, which are assumed to not vary with time or space. The rate at which altitude of a point on the fault scarp changes, 8y/5f, is equal to the hillslope curvature at that point, 82y/§X2 times the diffusivity coefficient, C. The diffusion equation has been applied to a variety of scarps and larger hillslopes (Culling, I960, 1963, 1965; Hirano, 1968, 1975; Kirkby, 1971; Nash, 1980, 1984, 1986; Colman and Watson, 1983; Hanks et al., 1984; Mayer, 1984; Andrews and Hanks, 1985; Pierce and Colman, 1986; and Nash and Beaujon, 2006). Subsequent applications that use the diffusion equation for modeling slope evolution in larger-scale landscapes and more complex scarps include publications by Begin, 1988, 1992; Fault Hanks and Andrews, Koons, 1989; 1989; Kirkby, 1992; Martin and Church, 1997; Arrowsmith et al., 1998; Pearthree et al., 1998; Hanks, 2000; Niviere and Marquis, 2000; Camelbeeck et al., 2001, Hil-ley et al., 2001; Lund et al., 2001; Mattson and Bruhn, 2001: Carretier et al., 2002; and Amoroso et al., 2004. A variety of solutions to the diffusion equation all provide products of age and the diffusiv-ity coefficient — tC. Values of tC products may be regarded as relative scarp ages where the value of Cis unknown, but is assumed to be constant. C can be estimated where f is known, and f can be made estimated where C is known. The -15 and -13 ka ages of the Pleistocene Lake Bonneville and Lake Lahontan highstand shorelines provided an opportunity to estimate values of C for sandy gravel. The combined dataset provides an appropriate value of the tC product of 16 m2 (Fig. 5.13). Substituting the known ages of the lakes in equation 5.1 provides an estimate of -1.1 m2/ky for the diffusivity coefficient, C. Noting a similarity with New Mexico work done by Mike Machette (1982), Hanks et al., 1984 concluded that the -1.1 m2/ky value might be widely applicable throughout the Basin and Range Province and Rio Grande rift valley. Values of the diffusivity coefficient may not be as uniform as suggested by lake shoreline studies. Nash's calculations of C range from 1.2 m2/ky for sands of wave-cut bluffs of Lake Michigan to 0.44 m2/ky for gravelly alluvium of a dated Utah Surveyed profile t 2a = 3.3 m Modeled profile J^^^ tc ( • ) = 3 m2 tc (A ) = 4 m2 1954 ^^^^ 2a = 1.9 m Offset^^k^r--tr^ tc ( • ) = 3 m2 tc (A ) = 4 m2 Figure 5.14 Scarp-profile modeling with different diffusivity coefficients. A. Comparison of surveyed (dashed line) and modeled (points) topographic profiles of fault scarps at the mountain-piedmont junction, Stillwater Range-Dixie Valley, Nevada. Single-rupture event fault scarps using a tc value of 3.0 mz and 4.0 mz. From Hecker (1955). Scarps 177 fault scarp. It is likely that lithologic controls cause an order of magnitude variation in the diffusivity coefficient of scarp degradation, even when restricting scarp lithology to just alluvium (Section 5.4). One should be cautious when using an age estimate from diffusion-equation modeling, especially when using an assumed value of diffusivity coefficient, C. The instructive Figure 5.14B examples 0 Initial ^T"""Ba===----__ offset C2 1 „ „ 1, 1; ,j. „ clack line: t=10 ka, Black line: «X c=0A m2/ka ■ t=50 ka, ^ c=0A m2/ka Gray line: t=4.5 ka, Gray line: t=21 ka, c=1.0 m2/ka 1— c-\.0 m2/ka 0 5 10 15 20 25 Horizontal distance, m Figure 5.14 Scarp-profile modeling using different diffusivity coefficients. F3. Four modeled scarp profiles for a hypothetical 2 m offset for young and older stages of degradation. Similarity of the black and gray line model results underscores sensitivity of scarp degradation to the assigned value of diffusivity coefficient, C. Diffusion-equation modeling does not provide unique solutions for scarp age. Figure 1 from Phillips etal. (2003). clearly show that tweaking C gets much different values for time, f. Getting a nice match to surveyed data is not enough. Independent, local determination of the C value is advisable; see Fig. 5.18 (Demsey, 1987). Scarp height also influences age estimates for scarps analyzed by diffusion-equation modeling (Pierce and Colman, 1986). Their diffusivity coefficients, C*, increase with scarp height and change with the microclimates of different slope orientations. C* = [1.35 Da + 3.03] x [0.1 m2/ky] (5.2) where Dq is apparent displacement. The next step is to use appropriate values of C, or C*, for fault scarps in piedmont sandy gravels throughout the same region. The examples used here 178 Chapter 5 Wassuk Range bedrock Latest Pleistocene Lake shore platform Early-mid Holocene alluvium Mid-Late Holocene alluvium Present depositional area Recent Older scarp scarp Normal faults Stream channel Figure 5.15 Map of piedmont fault scarps 1.5 km north of Copper Canyon that rupture alluvial surfaces postdating the 13 ka Lake Lahontan shoreline, Wassuk Range, west-central Nevada. From Demsey (1957). are from west-central Nevada. They are based on studies in the central Nevada seismic belt made by Hecker (1985) and Demsey (1987). Hecker studied piedmont fault scarps along the east side of the Stillwater Range (Fig. 4.26), which is the area of the 1954 Dixie Valley Mw magnitude 6.9 earthquake (Slemmons, 1967). Five of the prehistorical fault scarps can be nicely modeled with a tc product of 3.0 m2 or 4.0 m2 (Fig. 5.14A). Differences between modeled and surveyed apparent displacements of less than 5% are encouraging. Diffusion-equation modeling of fault scarps may be used to study fault segmentation. Scarps in the zone of overlap between fault segments may be poorly modeled with a single tc value if they are multiple-rupture event fault scarps. One of Hecker s topographic profiles was modeled crudely with a tc value of 3.0 m2. It is nicely modeled with tc values of 5.0 m2 and 8.0 m2 for the basal and crestal portions 40 50 O 20 «5 10 - — _ ^—' Frlor i.i.i, -0.2 0 0.2 OA 0.6 Log of scarp offset (H, in meters) 0.& Figure 5.16 Lines for regressions of maximum slope (S in degrees) and scarp offset (H, in meters) for three classes of scarps along the northern Wassuk Range. A is for segment A recent fault scarps. 5 = 17.5 (log H) + 19.1; n Is 17, r2 is 0.75. C is for segment C recent fault scarps. 5 = 17.0 (log H) + 20.4; n Is 13, r2 Is 0.77. Frlor Is for segments A, F3, C for the penultimate surface rupture event. 5 = 16.5 (log H) + 14.5; n Is 12, r2 Is 0.56. From Figure 7 of Demsey (1957). of the profile, respectively, which suggests a second surface rupture in this zone of overlap between two fault segments. The faulted and unfaulted alluvial surfaces of the piedmont adjacent to the 3,400 m high Wassuk Range in western Nevada have nicely constrained ages. This allowed Demsey to determine the local value of scarp diffusivity coefficient. Adjacent Walker Lake is a remnant of Pleistocene Lake Lahontan. A chronosequence of piedmont soils was described and dated using volcanic ashes and materials for radiocarbon dating from piedmont alluvium, topographic position within the piedmont landscape assemblage, and diffusion-equation modeling of fault scarps. Alluvial surfaces notched by the prominent Lahontan highstand shoreline (Benson and Thompson, 1987a,b Adams and Wesnousky, 1999) are -13 ka, and alluvium deposited on the shoreline is younger than 13ka(Fig. 5.15). The soils-geomorphology work provided internally consistent results. Maximum scarp slope indeed becomes less with time, as described by the Bucknam and Anderson model of Figure 5.10. Parallel regressions A and C of Figure 5.16 indicate the Fault Scarps 179 Soil-profile site Soil profile development index Age of alluvial geomorphic surface Number of surface ruptures Fan graded to Lahontan shoreline 16.63 13-15 (a) 1 Lahontan shore platform 14.53 11-14 (a) 2 Lahontan shoreline inner edge 11.73 11-13 (a) 1 Fan deposited on shore platform 10.73 7-12 (a, b) 2 Stream terrace at Rose Creek 10.70 >7(b) 2 High recessional Lahontan shoreline 10.25 11-13 (a) 2 Fan with graben at Copper Canyon 7.70 2-5 (b) 1 Stream terrace at Rose Creek 6.01 2-5 (b) 1 Fan near Squaw Creek 4.06 2-5 (b) 1 Fan near Copper Canyon 2.27 2-5 (b) 1 Fan deposited on lake highstand 2.06 2-5 (b) 1 Stream terrace near Cat Creek 1.63 <3(b) none Fan with graben near Copper Canyon 1.31 2-5 (b) 1 Fan in topographically low setting 1.26 2-5 (b) 1 Surface deposited Inside Graben 1.14 <3(b) none Fan with dated tephra layer 0.66 <0.6 (c) none Piedmont adjacent to younqest scarps 0.4£> <3(b) none Stream terrace with dated tephra 0.32 <0.5 (c) none Oldest unfaulted alluvial surface 0.15 <3(b) none Fan in topographically low setting 0.24 <3(b) none Table 5.1 Wassuk Range, Nevada piedmont surfaces and soil profiles ranked by soil profile development index (Harden, 1952). Age estimates are based on relation to the Lake Lahontan highstand, fault-scarp modeling, and radiocarbon dating and tephra correlations. Table 1 of Demsey, 1957. 55 r L tc + 0.66H + 0.15 n is 57 * L 3 tc + 4.7H + 1.6 Apparent displacement, m range of maximum slope-scarp offset regressions for sets of fault scarps created by the most recent Ho-locene surface-rupture event in two segments of the Figure 5.17 Plots of relative scarp age (tc product) and apparent displacement for young fault scarps along the northern Wassuk Range. From Demsey (1957). A. Both ages of Holocene fault scarps. B. 13 ka Lake Lahontan shoreline. Wassuk Range piedmont. The plot for the prior event is beneath the other two and its intercept value describes a greater decrease of maximum scarp slope. Descriptions of the soil profiles on a wide range of alluvial surfaces by Karen Demsey, Oliver Chadwick, and Philip Pearthree provided the detail needed to assess several different diffusion-equation models. Distinct advantages of working in the Wassuk Range study area included having tephra to cross-check radiocarbon age estimates, the latest Pleistocene shoreline of Lake Lahontan, and soil-profile descriptions on surfaces whose ages could also be estimated by diffusion-equation modeling. The information for this tidy package of data is summarized in Table 5.1, 180 Chapter where the data has been ranked according to the soil-profile development index of Harden (1982, 1987; Harden et al., 1991). This is done by numerically rating those soil-profile characteristics that change quickly with the passage of time (but at different rates of course). These included percent clay in the B horizon, redness with increase of iron oxyhydroxides, and profile thickness. The relation between apparent scarp age and scarp height is shown in Figure 5.17 for the Wassuk Range scarps. Variations in local microclimate probably are minimal because the piedmont fault scarps face towards the east. The topographic profiles of piedmont fault scarps with small apparent displacements generally are modeled best with relatively low values of tC products compared to higher scarps of the same age. Numerical age control seems internally consistent as shown in a plot of relative scarp age and apparent scarp height (Fig. 5.17B) for the 13 ka highstand shoreline of Lake Lahontan. The relation between the diffusivity coefficient for the scarps of the Wassuk Range piedmont, C*w, and apparent displacement, Da, is: 5 C*w = (0.36 Da + 0.12) m2/ky (5.3) where = 0.76 and n is 15. The several approaches to diffusion-equation modeling for Wassuk Range fault and shoreline scarps provide different results (Table 5.2). The modeled ages of the Lake Lahontan shoreline range from 11 to 20 ka, which suggests that the diffusivity coefficient and/or corrections for scarp height may vary between local study areas with the Basin and Range Province. ! 2 3 4 Time, ka Figure 5.15 Spread of calculated fault-scarp ages based on diffusion-equation modeling of the youngest fault scarps along the northern Wassuk Range, west-central Nevada using different diffusivity coefficients. A. 1.1 mz/ky of Hanks and others (1954). F3. ( 0.145 H + 0.3) mz/ky of Pierce and Coiman (1956). C. ( 0.35 H + 0.06) mz/ky of Demsey (1957) . The C*w diffusivity coefficient seems most appropriate for the Walker Lake piedmont fault scarps because of the small standard deviations of the C*w dataset as compared to the large standard deviations of the datasets when using c and C* diffusivity coefficients (Fig. 5.18). The C*w diffusivity coefficient also agrees best with the independent numerical ages of the most recent surface rupture in the Wassuk Range Average Estimated 5carp Age, ky Scarp set Linear diffusion equation Cubic equation c = 1.1 m2/ky (a) C*=0.14(H)+0.3 m2/ky (b) c*w=0.35(H)+.06 m2/ky (c) 1 = 4 (d) North area 2.5 ±1.5 3.5 ±2.1 2.3 ±1.1 1.5 ±0.6 Segment A 2.4 ±1.5 3.7 ±1.9 2.4 ±1.0 1.4 ± 0.5 Segment S> 2.3 ±1.7 2.1 ± 1.2 1.0 ±0.1 Segment C 4.5 ± 2.3 2.5 ±1.3 1.5 ±0.3 South area Segment D 7.6 ± 1.7 4.6 ± 0.9 2.4 ± 0.7 Prior event 6.5 ± 3.7 12.3 ± 5.5 6.9 ± 2.3 5.6 ± 3.5 Lahontan shorelines 11.2 ±5.3 20.1 ± 5.5 13.2 ±3.6 15.0 ± 6.6 Table 5.2 Comparison of morphologic ages for Wassuk Range scarps using different degradation rate coefficients and diffusion equations. From Demsey, 1957. Sources area) Hanks and others (1954), b) Pierce and Coiman (1956), c) Demsey (1957), d) Andrews and FSucknam (1957). Fault study area. Karen Demsey's 1987 study indicates a need to critically evaluate and constrain the diffusiv-ity coefficient for each study area. We also need to consider how diffusivity coefficients vary with scarp height, microclimate, and alluvium mass strength. 5.3 Climatic Controls of Fault-Scarp Morphology Climate may change in either a temporal or spatial sense. Temporal changes range from seasonal fluctuations to the major shifts in air-mass circulation, precipitation, temperature, windiness, and cloud cover associated with shifts from glacial to intergla-cial climates of the late Quaternary. Spatial changes include microclimate differences on north and south sides of a valley and variations with altitude and latitude. This section summarizes the effects of microclimatic variations on scarp morphology (Tables 5.3 and 5.4) in order to illustrate the overall importance of climatic controls on hillslope processes including fault scarps. A study of the scarps of late Pleistocene stream terraces of the Big Lost River (Fig. 1.3) in Idaho by 2.2 ms eroded South North 2.2 rrf deposited O 5 m Loess Colluvium Talus Gravel Scarps 181 Pierce and Colman (1986) provided essential information about climatic controls on scarp evolution. Their study is fundamental to better understanding of hillslopes in a more general sense. Variables such as lithology, age, scarp height, altitude, and regional climate were held about constant in order to study the effects of microclimate on rates and types of slope processes. Hillslope microclimate at a given latitude is a function of orientation, height, and steepness. About 100 late Pleistocene scarps of piedmont stream terraces were surveyed. The strongly seasonal climate is arid to semiarid, is frigid (Table 2.1), and has more than 130 freeze-thaw cycles per year. Stream terraces are ideal for studying temporal changes in small hillslopes because each terrace riser is graded to the stable base level of the adjacent lower terrace tread. Stable terrace treads isolate terrace risers from the effects of further stream-channel down-cutting. The Idaho terraces were formed as a result of lateral erosion induced base-level falls by streams incised into glacio-fluvial outwash gravels deposited at about 15 ± 4 ka. Gravelly colluvium deposited at the scarp bases has a silty matrix that is derived mainly from loess that caps the gravels. Rates of scarp degradation are limited only by rates of transport processes. Thus the Idaho stream-terrace scarps are ideal for using diffusion-equation modeling to better understand climatic controls on scarp evolution. Slope orientation (aspect) has a profound influence on hillslope processes and shapes and is most obvious in asymmetric valleys that have markedly different vegetation on the north- and south-facing slopes (Melton, I960; Bull, 1964b, Figure 6; Carson and Kirkby, 1972; and Dohrenwend, 1978). Slope asymmetry may be especially prominent north of the Figure 5.19 Importance of microclimate and scarp height on Idaho scarps with differing orientations. Contrasts in the morphologies of north- and south-facing 5.5 m and 5.9 m high scarps of the same age. Erosion and deposition are relative to presumed 33.5° starting slope angle. From Figure 3 of Pierce and Colman (1956). A. 25° maximum slope for north-facing scarp with slower erosion and deposition. F3. 17° maximum slope for south-facing scarp with faster erosion and deposition. Scarp height, m South-facing North-facing South/North ratio 2 5.4 5.1 1.Ö 5 16.2 4.5 5.6 10 54 6.Ö 5.0 15 52 9.2 5.7 20 70 12 6.1 Table 5.3 Effects of scarp heights and microclimate on Idaho terrace scarps. (From Table 3 of Pierce and Colman, 1956.) Changes in degradation-rate coefficient (C* x 10'' m2/ky) with five heights of north-and south-facing scarps. 182 Chapter 5 January freezing isotherm because of more freeze-thaw cycles on south-facing slopes and longer duration of insulating snow blankets on the north-facing slopes (Russell, 1909). The striking contrast in scarp morphology jo for north- and south-facing slopes in the study area of Pierce and Colman is shown in Figure 5.19. Degradation of the scarp crest and aggradation of the scarp toe, after the attainment of the assumed starting angle of 33.5°, are four times larger for the south-facing as compared to the north-facing scarp of the same age. Contrasts in rates of fluvial processes on south and north-facing scarps are a function of scarp heights. For 2 m high scarps degradation rates on south-facing scarps are twice those on north-facing scarps, but for 10m high scarps the difference is fivefold instead of being twofold. The increase of degradation rate with increasing scarp height is largely a function of sufficient increase in slope area to cross a threshold to more effective erosional processes, such as slope wash and gully erosion. The effects of scarp height and microclimate on slope processes for the Idaho stream terraces are summarized in Tables 5.2A and B. These factors have a considerable influence on the value of the degradation-rate coefficient, C* (equation 5.2). The independent variable controlling fluvial contrasts between south- and north-facing scarps is solar radiation. South-facing scarps tend to be much more perpendicular to the suns rays than north-facing scarps. The percentage of potential solar-beam radiation on slopes of a given orientation and inclination - the solar index - can be interpolated from the work of Frank and Lee (1966). The solar index for 23° slopes at latitude 44° N will be only about 30% for north-facing scarps and about 58% for south-facing scarps. 30 0 North-facing scarps 9 = 20.8 Log h + 9.2 n is 24 ni 25 r2 + 0.95 SI a> 20 to 15 Hi ' South-facing scarps o 9 = 12.9 Log h +7.2 va 10 n is 52 E —\ r2 + 0.92 E 1 o is 1.5° 'x 5 TO ° 2 _L 3 10 Scarp height, h, in m 30 Figure 5.20 Importance of microclimate and scarp height on Idaho scarps with differing orientations. Relation between maximum scarp slopes and scarp heights for north-, and south-facing scarps. For a given height, the south-facing scarps degrade most rapidly and the effects of scarp orientation are more for higher than for lower scarps. From Figure 7 of Fierce and Colman (1956). Height (size) attributes Orientation (microclimate) attributes Soil wash more effective on south-facing, drier, less-vegetated slopes Soil wash increases with slope length and height Freeze-thaw cycles greater on steeper, south-facing slopes Efficiency of creep may increase with surface gradient Cohesion resistance to creep less on drier, less vegetated slopes Lesser infiltration and greater runoff on steeper slopes Winter snowpack protects north-facing slopes from winter erosion and freeze-thaw and enhances soil moisture and hence vegetation cover Steeper slopes drier because they intercept less precipitation per unit of horizontal area Table 5.4 Effects of scarp heights and microclimate on Idaho terrace scarps {from Fierce and Colman, 1966). 5carp height and microclimate attributes that may affect the C*degradation-rate coefficient. Attributes listed in order of estimated Importance. Fault Contrasts in solar index for south- and north-facing scarps are reflected in plants and soil profiles. High north-facing scarps have a prairie-grassland type of plant community with only 10% bare ground; the root-bound sod is resistant to erosion. High south-facing scarps have a desert shrub type of plant community with 70% bare ground; the cohesionless silty surficial soils are mantled with a layer of frost-heaved stones. Such profound microclimatic differences on scarp morphology have important implications for fault-scarp modeling. Regional as well as local climatic differences should also have a strong influence on the diffusivity coefficient. Several Pierce and Colman illustrations summarize the affects of microclimate on the Idaho fault scarps. Maximum scarp slope for a given age increases with scarp height, but increases more rapidly for north- than for south-facing scarps (Fig. 5.20). Slope angles for west-facing scarps occupy an intermediate trend. Volumes of material degraded from upper halves of the scarps (Fig. 5.11) increase with increasing scarp height, but much more rapidly for south- 35 0 5 10 15 Scarp height, h, in m Figure 5.21 Importance of microclimate and scarp height on Idaho scarps with differing orientations. Contrasts between volume removed from the upper halves of scarps and scarp heights for north- and south-facing scarps. From Figure 11 of Pierce and Colman (1956). Scarps 183 than for north-facing scarps (Fig. 5.21). The diffusivity coefficient, C* increases rapidly with increases in scarp height for south-facing scarps and slowly for north-facing scarps (Fig. 5.22). Pierce and Colman normalized all scarps to west-facing, where the solar index does not change appreciably with variations of scarp slopes. The line with C* values halfway between those of south- and north-facing scarps defines the relation between C* and scarp height for due-west facing scarps. Figure 5.22 has trends that can be compared with the solar index. The trends of the residuals of the diffusivity coefficient, C* result when values of C* are subtracted from the values along the normalizing line for each scarp height. For south-facing scarps, residual C* is positive and becomes more positive with increases in scarp height. For north-facing scarps, residual C* is negative and becomes more negative with increases in scarp height. These trends are similar to solar-index trends for south- and north-facing scarps at 44° N. 0 5 10 15 5carp height, h, in m Figure 5.22 Importance of microclimate and scarp height on Idaho scarps with differing orientations. Trends of the residuals of the diffusivity coefficient that remain after C* values are subtracted from the due west line that is normalized for the effects of variable scarp height. The negative and positive trends are similar to those for the solar constant. From Figure 13 of Pierce and Colman (1956). 184 Chapter 5 The Pierce and Coiman study clearly shows how careful one must be when using diffusion-equation modeling to estimate ages of single-rupture event fault scarps. Scarp morphologies may be a useful tool to estimate ages, but only if one accounts for the effects of climate and scarp height, both of which may have a profound influence on fault-scarp evolution even where alluvium mass strength is constant. Although the influence of variations in solar index is large, scarp height has a still greater influence on the diffusivity coefficient. Temporal changes in climate may have an equally profound effect on rates of evolution of fault scarps and on hillslope morphology (see chapters 2 through 5 of Bull, 1991). We live at a most interesting time in geologic history from a paleoclimatic viewpoint. Marine records reveal that the major gla-ciation and sea-level low at 18-21 ka and the subsequent Holocene warming and sea-level rise of 130 m represent by far the most extreme swing in climatic extremes during the past 100 ka. The key point in regard to fault-scarp studies is that rates of hillslope and scarp degradation, the diffusivity coefficient of equation 5.1, may have varied by more than tenfold during the past 20 ka. Mean degradation rates for Holocene fault scarps may be much different than the mean degradation rates for fault scarps that have been subject to the effects of both glacial and inter-glacial climates. Age estimates may be in error by a factor of two (Pierce and Coiman, 1986, p. 883) if the fundamental controls of scarp evolution are not taken into account for a particular study site. Diffusion-equation modeling, like much other modeling, has the drawback of being difficult to calculate precision and accuracy of the results. It should not be used as a sole way of dating the fault scarps of a study region. Larry Mayer (1984) was the first to emphasize caution. Results should note field measurement errors and residual unexplained variance of model results. Evaluation of possible errors is discussed by Coiman, 1987; Mayer, 1987; Nash, 1987; and Avouac, 1993. The remarkable Pierce and Coiman study will encourage workers in other regions to carefully appraise the effects of climate. In the next section we shall see that lithologic control is an equally important variable affecting rates of fault-scarp degradation and diffusion-equation modeling age estimates. Recent studies suggest caution in using the standard diffusion-equation approach. Pelletier et al. (2006) in their study of prehistorical Lake Bonneville shoreline scarps in Utah concluded that neither scarp orientation nor microclimatic controls had a discernible effect on diffusivity values. Nash and Beaujon (2006) prefer a power-function instead of a linear diffusion model when examining how scarp gradient affects downslope flux of debris. They found an ideal flight of late Quaternary degradation terraces to test the linear diffusion model, which was not the best predictor of scarp morphology. Instead, scarp morphology, and changes related to scarp height, is modeled best when downslope flux of debris is proportional to slope gradient raised to the 3.4 power, results that are similar to the 3.0 power-function conclusion of Andrews and Bucknam, 1987. Modelers should be cautious in using initial results that provide reasonable comparisons, because better fits to the data may be possible. David Nash continues to make pioneering advances in the modeling of scarps, and readers are encouraged to take advantage of his web site*. 5.4 Lithologic Controls of Fault-Scarp Morphology Changes of alluvium mass strength that occurs with time or in space are complications for paleoseismolo-gists estimating ages of fault scarps with diffusion-equation models. Sandy piedmont gravel is not a truly uniform lithology. Seemingly minor spatial variations in cohesiveness may be a function of small amounts of clay, and of spatial variations in the abundance and size of boulders. Temporal changes in sandy gravel occur as pedogenic clay and calcium carbonate accumulate in soil-profile horizons. All these variables may affect rates of scarp degradation. An important question is "how wide a range in the variables of climate and lithology can be tolerated and still have a valid comparison of fault-scarp degradation rates between different study areas?" Variation in resistance of earth materials to erosional processes that change fault scarps is equally as important as climate and scarp height in determining scarp morphology. Two types of field areas are * Nash's computer program SlopeAge 112 for morphologic dating can be accessed at http://homepages.uc.edu/%7Enashdb/ SLOPEAGE/slopeage.htm. Enter the observed profile and assumed initial profile data for the entire scarp profile and the program calculates fc,, which is the age of the hillslope, t, times the diffusivity, c . Fault Scarps 185 Alluvial gravel Sandy lake gravels Figure 5.23 Lithologic controls of fault-scarp morphologies, Black Rock fault, northwestern Nevada. A. Maximum scarp slopes and topographic profiles of 1.4 ka fault scarps in alluvial gravel, lacustrine gravel, and lacustrine silty clay with scarp heights of 3.6, 3.5, and 4.0 m. Figure 6 of Dodge and Grose (1950). B. Regression lines of maximum scarp slope and scarp height for different deposits. From Figure 4 of Dodge and Grose (1950). Angular alluvial gravels, y = 15.29 + 29.73x, n is 15, rz is 0.50. Sandy lake gravel, y = 10.70+ 24.4Sx, n is 12, rz is 0.53. Lake clays, y = 9.36 + 24.11x, n is 11, rz is 0.55. used here for evaluation of lithologic controls; both are of single-rupture event normal fault scarps of late Holocene or historical age. Some surface ruptures cut materials that range in texture from gravel to clay. An example is a Holocene fault scarp in the Black Rock Desert of northwestern Nevada (Fig. 5.23) that was studied by Dodge and Grose (1980). A second approach is to study fault scarps in sandy gravels that have changed their resistance to erosion through the actions of pedogenic and fluvial processes. An example is the Pitaycachi fault in Mexico (Fig. 5.24) where pedogenesis between very infrequent earthquakes makes it excellent for studies of lithologic control of fault-scarp morphology. 5.4.1 Fault Rupture of Different Materials Dodge and Grose (1980) made regressions of maximum slope and scarp offset for single-rupture event fault scarps in lithologies that range from alluvial -0.2 0 0.2 OA 0.6 03 log of scarp offset (H, in meters) gravel to lake clay, which were ruptured along the Black Rock fault in northwestern Nevada at about 1.4 ka. Scarps in less resistant materials have topographic profiles that appear older than scarps of the same age and height in more resistant basin-fill lithologies (Fig. 5.23A). Slopes are not as steep and scarps have undergone more crestal rounding in lake clays than in angular gravelly alluvium. Regressions of maximum scarp slope angle and scarp height for these three types of deposits (Fig. 5.23B) suggest the magnitude of this lithologic control. The lesser slope for the lake-clay regression, as compared to the alluvial-gravel regression, also suggests that clays degrade more rapidly than alluvial gravels. Maximum slopes for 1-2 m fault scarps are about 23° in alluvial gravel but only 15° in lake clays, and maximum slopes for 3.5 m fault scarps are about 31° in alluvial gravel but only 19° in lake clays. Watters and Prokop (1990) realized that diffusion-equation modelers prefer initially cohesionless soils, but these are not present at many field sites. They compared fault-scarp characteristics with standard geotechnical engineering properties of shear strength, bulk density, Atterberg limits, and particle-size distribution. Laboratory testing of slightly cohesive materials involved carving large, undisturbed, approximately homogeneous, blocks out of fault scarp free faces in Dixie Valley, west-central Nevada. These 3.1 Normal-fault displacement of alluvial surface by the 1557 surface rupture, in meters STUDY SITES OTHER THAN STREAMS BJ Bolsa y~| Escarpa Vieja C~| Huella de los Caballos A| Escarpa Antigua P~| Dos Escarpas J~| Javelina F~| Facet 10 km Chapter 5 United States of America Republic of Mexico | | Pediment and basin fill ^§§;i;\ Bedrock —• 1557 normal fault scarp Figure 5.24 Map showing informally named study sites (in squares) and heights of 1557 scarps in m (bold lettering). Modified Figure 1 of Bull and Pearthree, 1955. Fault Scarps 187 2 O Holocene Late Pleistocene Mid-Pleistocene Š * 4 V-5 J=./?'i= 2 6 10 14 10 Meters Concealed ty Coíluvíum 22 26 30 Figure 5.25 Valley-fill stratigraphy exposed in the fault scarp at the Javelina site. Comparison of the maximum hues of soils and fill units listed in Tables 5.4 and 5.5 suggest that fill 6 is a Holocene aggradation event, fill 5 was deposited during the late Pleistocene, and fills 1-4 are early to mid-Pleistocene in age. Figure 13 of Bull and Pearthree, 1955. engineering soil characteristics influence the ratio of total scarp height to the amount of free face present. Assessing lithologic influences on fault-scarp processes and morphology cannot be done with standard engineering tests for common alluvium with ubiquitous cobbles and boulders, or for pedogenic overprinting by the formation of calcic and argillic soil-profile horizons. So, next we attempt to devise other ways of appraising lithologic controls on fault-scarp morphology. 5.4.2 Lithologic Controls on an 1887Fault Scarp The unusual Pitaycachi fault is interesting to earth scientists and planners. It is unusual because the great Sonoran earthquake of 1887 occurred in a vast region otherwise devoid of historical surface-rupturing earthquakes. The nearest historical surface rupture is a distant 600 km to the west in California. It is interesting because the 75 km long rupture is exceptional for normal faulting events. The 1887 surface rupture occurred in the San Bernardino valley of Sonora, Mexico (Fig. 5.24) which lies on the western fringe of the semiarid Chihuahuan Desert. The central 35 km of the 1887 scarp bounds the Sierra de San Luis, an embayed tectonic activity class 4 mountain front. Sandy to bouldery piedmont gravel is derived from granitic and volcanic rocks, and limestone. Di- verse alluvial materials and soil-profile characteristics result in variable alluvium mass strength. Piedmont surfaces are gently sloping, generally being 1° to 3°. The pediment immediately upslope from the Pitaycachi fault at the Javelina site is 0.5-1 km wide, and the pediment embayment formed on carbonate rocks farther north is several km wide. Valleys have been incised into the pediment only to be backfilled by climate-change induced aggradation events. Aggradation of valleys cut into a dissected pediment upstream from the Pitaycachi fault has occurred repeatedly at the Javelina site (Fig. 5.25, Table 5.5). Correlation of the Pleistocene—Holocene stratigraphic contact exposed in the scarp face with that exposed in the fanhead trench downslope from the fault indicates an 1887 apparent throw of 3.1 m. Before 1887 the fault scarp was only 2.5 to 3.5 m high. But unlike most western North America sites, the penultimate surface rupture is very old. Instead of an earthquake recurrence interval of 0.1 to 10 ky, it is on the order of 200-400 ky. Many, extremely infrequent but very large, prehistorical earthquakes have occurred here, dating back into the Tertiary. Valley-fill stratigraphy underscores the rarity of tectonically induced downcutting events. Five valley fills exposed in the Pitaycachi fault scarp face at the Javelina site represent episodes of about 2 to 4 m of downcutting and subsequent backfilling (Fig. 5.25). The fills have a great variety of lithologies, colors, and sedimentary structures (Table 5.5). These reflect different combinations of climate-controlled changes in weathering and episodes of stripping of the upstream hillslope sediment reservoir, plus dia-genetic changes after the aggradation events. The older fills have redder hues, whereas the younger are grayer. The time span required for the numerous climatic changes associated with the six periods of valley degradation and aggradation, and the subsequent diagenetic reddening and cobble weathering, exceeds Chapter 5 Figure 5.22 Aggradation event Lithology Dry Color of Weathered Alluvium Weathering of Grano-diorite Cobbles 6 Sand, silty Brown (7.5YR 4/3) Unweathered cobbles to incipient iron oxide stains along fractures in solid cobbles 5 Gravel, clayey Bright reddish brown (5YR 5/6) to bright brown (2.5YR 5/6) Iron oxide stains along fractures in solid to punky cobbles 4 Sand, clayey cross-bedded Reddish brown (2.5YR 4/6) No cobbles present 3 Gravel, clayey Reddish brown (5YR 4/6 to bright brown (2.5YR4/6) Funky to grussy cobbles 2 Gravel, sandy Dull orange (7.5YR 7/4) Funky to grussy cobbles 1 Gravel, clayey Bright reddish brown (5YR 5/6) to reddish brown (2.5YR 4/6) Grussy cobble remnants Table 5.5 Characteristics of Quaternary valley fills at the Javelina site. See Figure 5.22 for the interrelations between these valley fills and their estimated ages. 400 ky. This time span includes only the 1887 and the penultimate surface-rupture events. The penultimate event was similar in size to the 1887 event, where the height of the much eroded scarp is still preserved on nearly flat flights of ancient stream terraces. Few paleoseismologists would have spotted the trace of an active fault on even gently sloping terrain, where the scarp of the penultimate earthquake is obscure at best. The banner photo on the first page of this chapter and Figure 5.26 are examples. The 1887 apparent displacements at the Facet site range from 2.5 to 4 m. An obvious bevel of the fault-scarp crest is no longer apparent because most of the semiarid hillslope has been slowly eroded to almost a uniform slope . Presumably, a spectacular record of many late Cenozoic events is contained locally in the bedrock fault scarps of the two facets of more resistant rock at both sides of Figure 5.26. Fig ure 5.26 Aeria I view of the Facet site showing hillslope planation between the penultimate and 1557 surface-rupture events. The 10-35 m high wedges of relatively more resistant lava flows and carbonate cemented colluvium immediately upslope from the 2.5 to 4 m high 1557 scarp record large amounts of cumulative late Cenozoic uplift. The lack of pre-1557 fault scarps between these isolated remnants of former hillslopes indicates an exceptionally long time span between surface-rupture events. Figure 11 of Bull and Fearthree, 1955. 6 4 0 0.6° - 0.2 m 1637 r/ 2.4° rupture^ 3.6° 0 40 50 120 160 Distance, m 200 240 Figure 5.27 Topographic profile of an ancient multiple-rupture event fault scarp. The 1557 surface rupture is the 0.2 m increment in 3.2 m high scarp in clayey gravels at the Bolsa site. From Figure 2 of Bull and Pearthree, 1955. Fault Scarps 189 The northern termination of the Pitaycachi fault scarp has much smaller displacements for both the 1887 and prehistorical events. The sequence of small late Cenozoic surface ruptures is recorded locally by darker soil on the footwall block. Virtually no fault scarp is present (Fig. 8, Bull and Pearthree, 1988). Many small surface ruptures at the Bolsa site have created a broad, curving scarp crest rising above a miniscule 1887 offset (Fig. 5.27). Pediments were created during the Tertiary when denudation rates exceeded low rates of uplift, and perhaps before the creation of extensional stresses responsible for Plio-Pleistocene normal faulting. This low tectonic intensity landscape-evolution model characterizes parts of Arizona (Pearthree and Calvo, 1987) and New Mexico (Machette, 1986). The scarp of the Pitaycachi fault rupture is ideal for evaluation of lithologic controls on fault-scarp degradation in gently sloping bouldery alluvium with varying degrees of soil-profile formation. Rupture time is known and throw is fairly uniform. Free Stage Diagnostic Morphology Distribution of CaCO^. Calcic 5oiie Thin, discontinuous coatings on pebbles, usually on undersides. Coatings sparse to common " Continuous, thin to thick coatings on tops and bottoms of pebbles. Coatings common, some carbonate in matrix, but matrix still loose. III Massive accumulations between clasts, becomes cemented in advanced form. Essentially continuous dispersion in matrix Pedogenic Calcretes IV Thin (<0.2 cm) to moderately thick (1 cm) laminae in upper part of horizon. Thin laminae may drape over fractured surfaces. Cemented platy to weak tabular structure and indurated laminae. Horizon is 0.5-1 m thick. V Thick laminae (>1 cm) and small to large pisolites. Vertical faces and fractures are coated with laminated carbonate (case-hardened) surface). Indurated dense, strong platy to tabular structure. Horizon is 1-2 m thick. VI Multiple generations of laminae, breccia, and pisolites; recemented. Many case-hardened surfaces. Indurated and dense, thick strong tabular structure. Horizon commonly is >2 m thick. Table 5.6 Stages in the morphogenetic sequence of soil carbonate accumulation In gravelly alluvium (after Machette, 1955, p. 5). 190 Chapter 5 faces are still present on fault scarps formed in cohesive materials. Our measurements were made 94 to 99 years after the great Sonoran earthquake of 1887. Macroclimate is similar throughout the region. The west-facing aspect of the fault scarps eliminates most local microclimatic variations (Section 5.3). A key factor is that at least 200 ky elapsed between the prior surface rupture event and the 1887 event. Although multiple-rupture event fault scarps are present locally, a great variety of alluvium is ruptured only by the 1887 event. Soil-profile characteristics were used to estimate the approximate ages of the alluvial geomorphic surfaces. The Las Cruces, New Mexico soils study (Bachman and Machette, 1977; Gile and Grossman, 1979; Gile et al., 1981), 250 km to the northeast, is relevant because the climate, parent materials, topography, and biota are similar to those of the Pitaycachi area. Rates of input of atmospheric dust are important in the genesis of argillic and calcic soil horizons, and morphologic stages of calcic soil-profile horizons (Table 5.5) are essential for dating and correlation. The central two-thirds of the 1887 scarp is 2.5 to 4 m high, but south of Rio de Bavispe and north of Arroyo de los Embudos the scarp typically is only 0.5 to 1.5 m high. The 1887 fault trace is especially prominent where graben formation or subsidence due to Assuring of the downthrown block has concentrated local runoff and promoted dense vegetation. Apparent throw can be determined at many localities by measuring the vertical displacement of soil-horizon boundaries, alluvial strata, or terrace treads across the fault. Scarp heights approximate maximum displacements on these gentle piedmont slopes. Representative scarp heights are noted on Figure 5.24, but throw may be less than scarp heights where local subsidence or warping of the hanging-wall block materials has occurred. Our goal here is to quantify how different earth materials influence types of geomorphic processes, and the amounts and styles of fault-scarp degradation and aggradation. The scope of work done was threefold. First measurements of various earth materials and geomorphic properties were made in a partially successful attempt to quantify differences in erodibility. Topographic profiles of fault scarps were surveyed in order to compare amounts of degradation and aggradation of scarps in different lithologies. Third, new descriptive techniques were developed to better relate fault-scarp morphologies to lithologic controls. These include indices that describe scarp- crest sinuosity and the proportion of free face present a century after surface rupture. 5.4.2.1 Geomorphic Processes Both tectonic and fluvial processes shape fault scarps. Tectonic processes include rupture of the land surface and alluvium to create an obvious topographic discontinuity. This initial fault scarp may be quickly changed immediately by aftershock seismic-shaking events. Initial fault-scarp morphologies also are dependent on materials strength and cohesiveness that control the initial proportions of free face and debris slope that are present. A general lithologic sequence of increasing tendency to develop and preserve free faces is loose sand, gravelly sand, sandy gravel and clayey sand, clayey gravel, and cemented gravel and bedrock. New scarps in weak materials may consist only of debris slopes or fluvial processes may remove vestiges of free faces in a year. Stronger scarp materials for a given scarp height have a larger initial free-face component that requires > 100 years to erode by fluvial processes that also construct the debris and wash slopes. Lithologic controls of fluvial processes include: 1) Amount of silt and clay in the alluvium. Fine particles tend to plug the interstices between larger particles as part of the process of raindrop splash on bare soil; thereby reducing infiltration and increasing runoff for a given rainfall event. 2) Transmissivity rates for unsaturated flow in porous media. 3) Shear stresses needed, for a given slope, to entrain sand and gravel by flowing water. 4) Abundance and type of pedogenic clay that forms cohesive argillic soil horizons in sandy gravel parent materials. It is subject to shrink-swell processes and may decrease in strength when wetted. 5) Abundance and distribution of pedogenic cementation by oxides of iron, aluminum, and silicon, and by calcium carbonate. 6) Cobbles and boulders increase the alluvium mass strength of the deposits and may accumulate as a sur-ficial lag deposit, or form a desert pavement that armors alluvial surfaces. 7) Soil characteristics that influence the type and density of vegetation whose leaves and stems intercept rain and whose roots increase soil shear strength. 8) Particle-size distribution that favors bioturbation by roots and rodents. Fault Scarps Free faces and debris slopes are the dominant topographic elements of the single-rupture event fault scarps formed by the 1887 earthquake. There has been little modification of the upslope and downslope original surfaces of the Pitaycachi fault scarp in a century. Incipient rounding of the scarp crest by sheetflow and rillwash has steepened a few meters of the upslope original surface from 2° to 3°- 5°. Downslope original surfaces are little changed. Scarps in cohesive gravelly soils provide a scattering of cobbles, and some reach the downslope original surface (Fig. 5.28A, B). Sandy gravel is more 191 Figure 5.26 Views of 1667 Pitaycachi fault scarps in different materials. A. Holocene sandy gravel in 3.3 m high scarp near Arroyo Pitaycachi. Weakly cohesive alluvial-fan deposits still have a free face, below which is a bouldery debris slope. F3. Late Pleistocene sandy gravel with strongly developed argillic and weakly developed calcic soil horizons at the Dos Escarpas site. Post 1667 free face retreat has made a debris slope that is progressively coarser-grained in the downslope direction (Fig. 5.27). Hat for scale. C. The 4 m high scarp In massive carbonate-cemented gravel south of Arroyo Capadero Is dominated by a large free face. Blocks of cemented alluvium lie on the debris slope. Figure 3 of Bull and Pearthree, 1966. 192 Chap conducive for deposition of a few centimeters as a new alluvial slope that consists of freshly deposited sand derived from the adjacent debris slope and free face. Post-1888 streamflow has notched the fault scarp and deposited alluvial fans (Fig. 2.12). Free-face slopes reflect lithologic controls, including the soil-profile characteristics summarized ter 5 in Table 5.6. The Pitaycachi fault dips 50° to 80° at the land surface, and the slope of the initial free face probably was 60° to 90° in clayey and cemented lithologies. Free faces are virtually unmodified in carbonate-cemented limestone gravel (Fig. 5.28C), but are no longer present in clayey gruss derived from quartz monzonite. Free faces are still common in Bt soil horizon Soil Site Age Thick- Maxi- Chro- Maxi- CaC03 CaC03 Scarp Tectonic classfica- ness mum ma2 mum depth stage height4 offset4 tion1 (cm) Hue2 % clay3 (cm) (m) (m) Typic Arroyo Holocene Not old enough for a Bt horizon None 1.5* torriorthent Hondo Ustollic Embudos Late 120 5 YR 6 25 140 1 2* 1.5 to haplargid Pleistocene 3.0 Ustollic Embudos 320 5YR 6 25-30 350 1 2* 1.5 to haplargid 3.0 Typic Huella de trun- 5 YR b 25-30 >90 lll-IV 3.1* haplargid Caballos cated Ustollic Arroyo 93 2.5 YR b 33 71 ll-lll haplargid Hondo Ustollic Embudos Mid- 90 2.5 YR 6 60 66 1 4 to 7 4 to 6 haplargid Pleistocene Ustollic Javelina >105 2.5 YR b 45-50 6 to b 4 to 6 haplargid Petrocalcic Arroyo Early >163 2.5 YR b 45 6b lll-IV 15 9 to 13 ustollic Hondo Pleistocene haplargid Petrocalcic &o\ea 200 2.5 YR b 50 210 IV 3 to 6 ustollic haplargid Eecar^a Ustollic 116 2.5 YR b 50 44 1 14 to 16 vieja haplargid Petrocalcic Arroyo Late 54 2.5 YR 6 40 10 V paleargid Hondo Tertiary Eecar^a VI 43 39 antigua Not described * Surface ruptured only by 1&&7 event 1 Taxonomy is from Soil Survey Staff (1975). 2 Munsell 1992 notation. 3 Gravelly parent materials have /ess than 5% clay. 4 Scarp height is the vertical measurement from base to crest. Displacement (apparent net vertical throw) is measured from offset etratiqraphic unite and soil horizons, or more commonly by projection of the slopes of the upper and lower original surfaces to the fault trace._ Table 5.7 Summary of soil-profile characteristics of the Pitaycachi fault Terrace chronosequence. Fault Scarps 193 sandy gravel but have retreated 2 to 8 m upslope from the fault trace. Debris slopes typically are 24° to 38° and have been sites of net erosion, not deposition. Amounts of debris-slope retreat can be estimated where roots of mesquite trees are exposed. 5.4.2.2 Scarp Materials Alluvium mass strength characteristics of fault scarps should be described in both a surface and a subsurface sense. The diverse materials ruptured by displacement along the Pitaycachi fault are summarized in Tables 5.6 and 5.7. Several characteristics were measured in an attempt to describe alluvium mass strength numerically. Seismic velocities increase with increases in cementation and abundance of unweathered boulders and cobbles of rhyolite. Seismic velocities are a fairly good tool for describing deeper subsurface materials but are not appropriate for describing loose finegrained materials that are subject to rockfalls, rillwash, and sheetflow Data obtained with a spring-loaded penetrometer seem reliable for describing weakly cohesive sandy and silty alluvium, but not for other lithologies. Gravels could be tested only by penetrating into pockets of the materials between cobbles. Argillic soil horizons with more than 40% clay were difficult to test. Even when moist, the smallest penetrometer head was unable to penetrate these tough soils at the moderately low design applied stresses of this standard geotechnical engineering tool. The K constant in the Universal Soil Loss Equation (Wis-chmeir and Smith, 1965, 1978) was quite low whenever gravel was abundant. David Taylor improvised a test that worked well. He concentrated on measuring resistance to shear of both gravelly and nongravelly surficial materials. By noting the mean depth of penetration during horizontal movements of 6 kg weighted steel tines he made estimates of the shear strength of surficial soils. These shear-strength estimates provided a consistent measure of variations of surficial materials strength that correlated well with the penetrometer data of loamy soils and with scarp morphologic characteristics such as maximum slope and scarp-crest sinuosity (Fig. 5.29). The Taylor rake test succeeded because it measured materials characteristics in a depth increment relevant to geomorphic processes. This attempt to devise a numerical multiparameter materials strength index for fault scarps in alluvium was unsuccessful (Table 5.7). Most parameters were inconsistent or were not appropriate for describing lithologic controls of the main fluvial processes responsible for changing scarp morphologies. Quantitative multi-factor indices that describe erodibility of gravelly piedmont alluvium need more testing before use in fault-scarp studies. For now, we relate fault-scarp characteristics to three broad lithologic classes (Table 5.7). Lithologic resistance to erosion increases greatly in sandy gravel with the passage of time. Sur- E 50 12 3 4 Scarp crest sinuosity Figure 5.29 Relation of shear-strength index of surficial materials to sinuosity of fault scarp crests in different materials. Type of alluvium Particle-size distribution % Silt % > 2 mm Shear-strength index, mean penetration, mm Penetrometer, mean kilonew-tons/m2 Mean seismic velocity, m/sec Mean K factor of the Universal Soil-Loss Equation Grussy 1 to 7 6 to 15 41 210Ö 335 0.03 Cobbly 7 to 27 4 to 57 29 4663 354 0.04 Carbonate cemented 8 to 58 30 to 65 20 5390 326 0.05 Table 5.6 Average alluvium mass strength characteristics of alluvial materials ruptured by the Pitaycachi fault in 1667 at 39 fault scarps. 194 Chapter ficial lag gravels accumulate and protect the soil from erosion by rainsplash and sheetflow. Concurrent increases in shear strength are caused by additions of pedogenic clay and carbonate in argillic and calcic horizons. Erosion of the 1887 scarp resulted in different materials being present on different topographic elements. Figure 5.30 is a sample of particle-size distributions of free-face and debris-slope materials. The free-face source material is finer grained and more poorly sorted than the debris-slope surficial materials. Cobble size increases downslope, and there is an abrupt transition to fine materials at the scarp base that coincides with an incipient wash slope. 5 5.4.2.3 Scarp Morphology Topographic profiles of the 1887 fault scarps were surveyed normal to the fault trace to evaluate scarp height, maximum scarp slope, and scarp width. Scarp-crest sinuosity and free-face index also were measured. Each of these topographic characteristics is affected by soil profile and alluvium-mass-strength characteristics. Replication accuracies of a given scarp profile were determined for each survey method and as many as 20 surveys along a single transect were made to compare variations between different surveyors. Mayer (1984) concluded that site selection and careful calibration of survey methods are important. Profiles for 2 to 3 m high scarps vary greatly as a function of lithology. A century of erosion has reduced scarps in clayey sand to 8° to 20° slopes without free faces. Free faces that appear unchanged typify scarps in carbonate-cemented gravels. They may persist for millennia. Scarps underlain by different materials evolve at different rates so may become more similar, or different, with the passage of time (Peter Knuepfer, personal communication). Scarps in soft, sandy alluvium and clayey gravel are clearly different after a century of erosion, but this distinction will decrease with the passage of time as scarps converge to a common morphology at different lith-ologically controlled rates. More than 10 to 100 ky may be needed for two lithologic classes of scarps to become indistinguishable. Carbonate-cemented free faces are even more persistent (Table 5.8). They may still be present after 1,000 years. Morphologic descriptions of young fault scarps should also include parameters that 1) involve both horizontal dimensions and 2) the proportions of topographic elements (Fig. 5.2) present. Here 2 4 ft 16 32 64 12ft Abundance of particle sizes in phi size classes, mm Figure 5.30 Comparisons of particle-size distributions in the free face and the upper and lower parts of the debris slope at the Dos Escarpas site. See Figure 5.25F3 for a view of the sample site. we describe scarp-crest sinuosity and the proportion of free face still present in scarp height. Free-face retreat undermines boulders and results in a narrow crest compared to scarps without free faces. At the Dos Escarpas site, scarp-crest sinuosity was measured by placing a loosely draped tape along the break in slope that defines the scarp crest (Fig. 5.31). Figure 5.31 View of survey tape laid out along a segment of the crest of the 1SS7 ecarp at the Dos Escarpas site. Instead of being straight, the white tape has a sinuosity of 1.4. Fault Scarps 195 Type of alluvium Maximum elope, ° Mean slo Scarp width, m Scarp crest sinuosity Mean 1 G Mean 1 G Mean 1 G Mean Range Grussy 20 Ö 10 4 7.4 1.9 3.Ö 2.9 - 4.6 Cobbly 44 15 29 Ö 4.Ö 1.9 2.4 1.4-3.7 Carbonate cemented 50 13 37 10 4.3 1.3 1.3 1.1-1.5 Table 5.9 Morphologic characteristics of the 1557 scarps of the Pitaycachi fault as related to the degree of cementation and relative abundances of gravel and clay in the scav