Catastrophic middle Pleistocene jökulhlaups in the upper Susquehanna River:
Distinctive landforms from breakout floods in the central Appalachians
R. Craig Kochel a,
, Richard P. Nickelsen a
, L. Scott Eaton b
a
Department of Geology, Bucknell University, Lewisburg, PA 17837, USA
b
Department of Geology & Environmental Science, James Madison University, Harrisonburg, VA 22807, USA
a b s t r a c ta r t i c l e i n f o
Article history:
Received 14 November 2008
Received in revised form 23 March 2009
Accepted 24 March 2009
Available online 6 April 2009
Keywords:
Jökulhlaup
Ice-damming
Susquehanna River
Middle Pleistocene
Valley and Ridge
Geomorphology
Widespread till and moraines record excursions of middle-Pleistocene ice that flowed up-slope into several
watersheds of the Valley and Ridge Province along the West Branch of the Susquehanna River. A unique
landform assemblage was created by ice-damming and jökulhlaups emanating from high gradient mountain
watersheds. This combination of topography formed by multiple eastward-plunging anticlinal ridges, and the
upvalley advance of glaciers resulted in an ideal geomorphic condition for the formation of temporary icedammed
lakes. Extensive low gradient (1°­2° slope) gravel surfaces dominate the mountain front
geomorphology in this region and defy simple explanation. The geomorphic circumstances that occurred
in tributaries to the West Branch Susquehanna River during middle Pleistocene glaciation are extremely rare
and may be unique in the world. Failure of ice dams released sediment-rich water from lakes, entraining
cobbles and boulders, and depositing them in elongated debris fans extending up to 9 km downstream from
their mountain-front breakout points. Poorly developed imbrication is rare, but occasionally present in
matrix-supported sediments resembling debris flow deposits. Clast weathering and soils are consistent with
a middle Pleistocene age for the most recent flows, circa the 880-ka paleomagnetic date for glacial lake
sediments north of the region on the West Branch Susquehanna River. Post-glacial stream incision has
focused along the margins of fan surfaces, resulting in topographic inversion, leaving bouldery jökulhlaup
surfaces up to 15 m above Holocene channels. Because of their coarse nature and high water tables,
jökulhlaup surfaces are generally forested in contrast to agricultural land use in the valleys and, thus, are
readily apparent from orbital imagery.
 2009 Elsevier B.V. All rights reserved.
1. Introduction
A jökulhlaup is a glacier outburst flood that usually occurs when a
glacially dammed lake drains catastrophically, or when there is a
catastrophic release of water from a glacier. Jökulhlaups are common
in Iceland, where subglacial volcanism triggers catastrophic episodic
releases of meltwater (i.e., Costa, 1988a; Waitt, 2002; Carrivick et al.,
2003; Carrivick, 2007). They also occur in nonvolcanic regions where
they tend to occur from the temporary damming of unglaciated
tributary valleys by glacial ice in the mainstream valley such as at
Hidden Lake, Alaska (Anderson et al., 2003). In special cases, major ice
lobes intersected mountain ranges to dam large rivers; a primary
example being the Channeled Scabland Floods associated with the
Glacial Lake Missoula jökulhlaups of the northwestern U.S.A. (Baker,
1973; Waitt, 1985). The geomorphic impact of jökulhlaups is typically
huge as illustrated by the paleoflood landforms of the Channeled
Scablands.
We report here on a geological setting and perhaps unique set of
fluvioglacial conditions created by the movement of glacial lobes up
tributary valleys to the West Branch Susquehanna River and by the
distinctive structural geology of the region that provided conditions
favorable for the formation of temporary ice-dammed lakes and
subsequent jökulhlaups in central Pennsylvania (Kochel et al., 2003).
Thin glacial lobes (200 m thick) flowed upslope into the Buffalo Valley
from the West Branch Susquehanna River and intersected a series of enechelon
sandstone ridges formed by eastward-plunging anticlines
(Fig. 1). Headwater streams emerging from the mountains between
the plunging folds were temporarily dammed by the ice. Jökulhlaups
were released as these temporary ice dams failed resulting in a
distinctive suite of landforms visible from orbital imagery (Fig. 1) and
whose legacy remains significant in the geomorphic evolution and land
use history of the region today.
Sediment-rich flows created from the jökulhlaups occurred
because of the interaction of periglacial and glacial processes. These
events created a series of elongated debris fans composed of matrixsupported
gravel resembling debris flow deposits, but these deposits
were emplaced on exceptionally shallow gradients that are well below
the slopes required for normal debris flow transport (Fig. 2). Postglacial
fluvial incision resulted in topographic inversion of some of the
jökulhlaup surfaces because their coarse sandstone fill armored them,
causing streams to downcut in softer shales along their margins. The
Geomorphology 110 (2009) 80­95
 Corresponding author. Tel.: +1 570 577 3032.
E-mail address: kochel@bucknell.edu (R.C. Kochel).
0169-555X/$ ­ see front matter  2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomorph.2009.03.025
Contents lists available at ScienceDirect
Geomorphology
journal homepage: www.elsevier.com/locate/geomorph
impact of the flows has been significant in the evolution of mountainfront
geomorphology in this section of the Valley and Ridge. These
catastrophic floods likely played a role in the evolution of macrotubulent
erosional development of the lower gorge of the Susquehanna
River located more than 150 km downstream. The middle
Pleistocene landforms produced by these floods still dominate the
geomorphology of the region today and are readily visible from orbital
imagery (Fig. 1).
2. Regional setting
2.1. Pleistocene geology of north-central Pennsylvania
2.1.1. Glaciation
The impact of Pleistocene climate changes have been significant in
north-central Pennsylvania, including numerous episodes of glaciation
and extensive periglacial activity on the ridges and hillslopes
(Peltier, 1949; Denny, 1956; Braun, 1989; Marchand and Crowl, 1991).
Late Pleistocene glaciers advanced over significant portions of the
Appalachian Plateau north of Williamsport (Fig. 1), but failed to cover
the ridges to the south in the Valley and Ridge. Middle Pleistocene
glaciers were more extensive but also failed to cover the ridges south
of Williamsport. Evidence of these earlier, more extensive ice
advances is widespread in the form of extensively weathered till,
subdued moraines, and glacial lake deposits in the West Branch
Susquehanna River (Peltier, 1949; Bucek, 1976; Marchand and Crowl,
1991). Investigations of weathering and soil development on these
sediments suggest a middle Pleistocene age for these deposits (Peltier,
1949, Marchand and Crowl, 1991). Intersection of these ice lobes with
Bald Eagle Ridge (the northernmost of the large Ridge and Valley
folds) temporarily dammed the West Branch Susquehanna River at
Williamsport, resulting in an extensive lake, Glacial Lake Lesley
(Bucek, 1976). Paleomagnetic evidence indicates this lake dates to
880,000 BP (Ramage et al., 1998), supporting the pedogenic
interpretations for a middle Pleistocene age for the till. More
extensively weathered tills have been observed in some locations
Fig. 2. Quaternary deposits in Buffalo Valley, Union County, Pennsylvania; including jökulhlaups, ice dammed lakes, till, and terraces. Jökulhlaup surfaces (orange) extend
downstream from the mountain-fronts of White Deer Creek (WD), Spruce Run (SR), Rapid Run (RR), North Branch Buffalo Creek (NB), and Laurel Run (LR). An extensive Late
Pleistocene debris fan (brown) occurs near the head of Buffalo Creek and a smaller one on Stony Run. Significant middle Pleistocene till deposits are shown (pink), while scattered
thin till occurs throughout the valley region. The approximate limit of the middle-Pleistocene ice (pink dashed line), which advanced from the ENE between the two branches of the
Susquehanna River, is modified from Marchand and Crowl (1991). Alluvium (yellow) is undifferentiated middle Pleistocene to Holocene because of the map scale; however, our
detailed mapping has identified several middle Pleistocene terraces and at least one post-jökulhlaup-age terrace along Penns Creek and Buffalo Creek.
Fig. 1. Orbital image of north-central Pennsylvania showing the West Branch Susquehanna River. Note the series of eastward-plunging folds (dashed arrows) and the jökulhlaup
surfaces discussed here (white triangles -- WD = White Deer Creek, SP = Spruce Run, RR = Rapid Run, NB = North Branch Buffalo Creek, LR = Laurel Run). Landsat 7 image courtesy
of NASA/PASDA (Pennsylvania State Digital Access). Solid white lines are county boundaries.
81R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
82 R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
(Peltier, 1949; Marchand and Crowl, 1991) indicating there may have
been several middle to early Pleistocene ice advances down the West
Branch of the Susquehanna River.
Although these middle Pleistocene glaciers did not cover the ridges
south of Williamsport, they did extend several tens of kilometers
southward into the Valley and Ridge as relatively thin glacial lobes
along the West Branch Susquehanna River (Fig. 2). Extensive
weathered moraines occur in the valleys of Buffalo Creek and Penns
Creek, marking the upvalley extent of these ice lobes that were
sourced by the glacier moving south along the West Branch
Susquehanna River (Fig. 3). Middle Pleistocene till has been reported
at least 10 km south of the confluence of the branches of the
Susquehanna River at Sunbury (Peltier, 1949; Marchand and Crowl,
1991; Sevon and Braun, 1997). Widespread till and glacial landforms
in tributary valleys to the Susquehanna (such as White Deer Creek,
Buffalo Creek, and Penns Creek) show that portions of the Susquehanna
glacial lobe moved significant distances up these tributary
valleys (Fig. 2). These ice lobes resulted in significant drainage
adjustments and valley alteration in some locations. For example, a
lobe of the ice dammed the former channel of Penns Creek, depositing
a moraine just SW of the village of Mifflinburg (Fig. 2). Prior to this,
Penns Creek joined Buffalo Creek at the western edge of Mifflinburg.
The till now buries the former channel of Penns Creek, which was
detected in geophysical surveys (Fig. 4), and diverted Penns Creek
southward across a low divide into a new valley near Swengel. Four
major terraces occur along Penns Creek upstream of Swengel. The two
older terraces do not make the turn southward at Swengel but
continue eastward until they are lost in the outwash and moraines
deposited by the ice lobe that deflected Penns Creek. The two younger
terraces follow Penns Creek southward toward Selinsgrove and the
Susquehanna River, having formed later in the Pleistocene following
the drainage diversion. Studies of soil chronosequences (Malonee,
2001) are consistent with a middle Pleistocene age for the upper
terraces and a Late Pleistocene age for the lower ones.
Eventually, the upvalley-moving ice intersected the plunging folds
and headwater streams emanating from the ridges, where they
formed temporary ice dams and lakes at the mountain front (Figs. 1
and 2). Structural features lead to the unique presence of headwater
lakes dammed against middle Pleistocene glacial lobes. In this region
the north limb of the major Buffalo Valley synclinorium, trending N.
60° E., is the physiographic expression of the resistant Tuscarora
Sandstone, dipping SE into the valley containing less-resistant shales,
sandstones, and limestones of Silurian and Devonian age. On this SEdipping
limb there are five en-echelon minor anticlines with crest-tocrest
wavelengths of 3 km that plunge toward N. 70° E. The eroded
synclinal valleys in this system contain small NE-flowing streams of
varying discharge, some of which spawned significant lakes against
the glacial ice dams. Other SE-dipping limbs of major synclinoria in
the region do not contain the array of minor folds that have been so
important in creating the fluvial environment for the different-sized
Fig. 3. Middle Pleistocene terminal moraines in western Buffalo and Penns Creek. (A) View northeastward from Swengel of the short outwash plain and terminal moraine in the
distance was formed during the diversion of Penns Creek to the south. (B) Hummocky surface of boulder-rich terminal moraine west of Laurelton. (C) View of the orange soil and till
in the terminal moraines west of Laurelton (view is along the road shown in Fig. 3D). (D) Hummocky surface of terminal moraine complex west of Laurelton.
83R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
Fig. 4. Buried paleochannel of Penns Creek. The middle Pleistocene ice lobe, advancing upvalley from the east, blocked the former channel of Penns Creek NE of Swengel (moraine), diverting it southeastward from Swengel over a low divide,
and burying the former course of Penns Creek through this valley toward Buffalo Creek at Mifflinburg. (A) Seismic and electrical resistivity survey across the former valley of Penns Creek. View is to the south along the N to S ER Traverse line in
Fig. 4B. The former course of Penns Creek is in the swale near the center of the photo. (B) Map of the buried valley formerly occupied by Penns Creek. The two oldest (of four) terraces of Penns Creek stop abruptly as they head northeastward
into this valley from Swengel (Malonee, 2001), erased by the till deposited in the moraine and subsequent outwash. Two younger terraces along Penns Creek follow its current course to the south. Note the wide expanse of alluvium near
Swengel where this diversion took place. (C) Electrical resistivity survey running north-south across the buried valley of Penns Creek. The buried valley appears as the dark oval between 2 and 5 m depth and is buried by 2 m of till and
outwash sediment.
84R.C.Kocheletal./Geomorphology110(2009)80­95
ice-dammed lakes. Specifically, the next synclinorium to the north, the
White Deer synclinorium (which was also occupied by a glacial lobe)
did not have any small en-echelon folds that would have served as a
focus for ice-dammed lakes. We have found no evidence that
jökulhlaups were generated along the SE-dipping limb of the White
Deer synclinorium. Apparently, the ice was unable to completely dam
the stream emanating from this large synclinal valley. The glacial lobes
that flowed toward the SW upgradient into both the Buffalo valley and
the White Deer synclinorium valley were unable to surmount the
ridges to the west and north that stood over 300 m above the valley
floor. The elevations of the till observed in Buffalo, White Deer, and
Penns Creek valleys indicate that ice thickness in these Susquehanna
tributaries was 200­250 m.
Temporary ice-dammed lakes formed where the major tributaries
to Penns Creek and Buffalo Creek emerge from the mountain front in
this region. Significant are the lakes at the emergence of West Branch
Buffalo Creek, Rapid Run, Laurel Run, and Spruce Run (Fig. 2). Some ice
damming also occurred where White Deer Creek exits the mountain
front 10 km NE of Spruce Run. The ice dam at White Deer Creek,
however, does not appear to have fully blocked the stream. Thus, its
lake and subsequent jökulhlaups are not as extensive as they are for
some of the tributaries to the west in Buffalo Creek. It is unclear why
damming of White Deer Creek was less extensive, but may have been
related to the flow vector of the glacial ice toward the SW, slightly
bypassing this valley because of deflection toward the south around
the easternmost anticlinal ridge just to the north of White Deer. The
jökulhlaup events deposited extensive elongate debris fans along the
piedmont zones of these headwater streams, creating the bouldery
surfaces that are the subject of this paper.
2.1.2. Periglacial activity
Throughout the Pleistocene glacial intervals, extensive periglacial
activity affected the ridgetops and hillslopes of the Valley and Ridge
south of Williamsport (Peltier, 1950; Marsh, 1987; Braun, 1989, 1994;
Bunting, 2005). Extensive blockfields, stone stripes, and thermokarst
features are widespread across hillslopes in this region (Fig. 5).
Gelifluction and thermokarst decay processes were probably extensive
during these periglacial climates, delivering large volumes of
sediment to headwater tributary streams. Relict gelifluction lobes are
common as undulatory benches throughout the region. Debris fans at
the mouths of small headwater streams unaffected by direct ice
contact show significant terracing, likely the result of the influence of
gelifluction activity. Pleistocene debris fans are common throughout
central Pennsylvania (Berntsen, 1993). Terracing is particularly well
developed on debris fans with south-facing aspects throughout the
region.
3. Jökulhlaups
3.1. Jökulhlaup surfaces and landform geometry
Deposits from the jökulhlaups appear as smooth elongated, lowgradient,
bouldery, fan-like surfaces (linear forested zones in Figs. 1, 2,
6, 7 and 8) extending away from the mountain-front locations of
Fig. 5. Examples of periglacial features in the region. (A) Large blockfield on north-facing slope along Penns Creek west of Weikert. (B) Stone stripes along the upper slopes of Stony
Run watershed near Laurelton. (C) Blockfield on south-facing slope (north-facing slope also has a major blockfield) in Buffalo Gap. (D) Large late Pleistocene debris fan emanating
from Buffalo Gap, where the headwaters of Buffalo Creek exit the mountain front north of Hartleton. These slopes show the widespread abundance of periglacial colluvium typical of
the areas upstream of the ice-dammed lakes that likely contributed much sediment to the sediment-rich jökulhlaups.
85R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
major tributaries to Buffalo Creek and Penns Creek in Union County. It
is important to note that although the jökulhlaup deposits now occur
in both watersheds they were within a single watershed prior to the
middle Pleistocene glaciation that produced them. At that time, Penns
Creek joined Buffalo Creek along the western margin of Mifflinburg
(Figs. 1 and 2).
The elongate jökulhlaup surfaces extend between 5 and 9 km from
their mountain-front source and are up to 1 km in width (Fig. 6). In
most cases, the deposits fill the valley from one wall to the other and
are singular. However, the deposit from North Branch Buffalo Creek is
more complex and branches into two surfaces separated by up to 3 km.
Post-depositional modifications and stream captures have masked
some of the original extent of the North Branch surface that appears to
have at been formed by at least two phases of activity. Observations of
weathering indices do not reveal significant differences, suggesting
that the ages of the two arms of the North Branch surface are not
Fig. 6. Jökulhlaup surfaces viewed from the air in vertical air photos. Flow of both jökulhlaups was from NW to SE. (A) Rapid Run surface. (B) North Branch surfaces (2006 photos
courtesy PA-DCNR).
86 R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
Fig. 7. Isopach maps of typical jökulhlaup surface thicknesses based on water well data. Well logs and electrical resistivity surveys indicate that average thickness of the deposits is 10 m. (A) Rapid Run surface. (B) North Branch Buffalo Creek
surfaces. (C) DEM hillshade view of Rapid Run (RR) and the bifurcated North Branch (NB) jökulhlaup surfaces. (D) Model of the electrical resistivity survey across North Branch surface (line shown in Fig. 7B and photo 8B).
87R.C.Kocheletal./Geomorphology110(2009)80­95
greatly different. Many of the jökulhlaup surfaces are quite distinctive in
aerial and orbital imagery because they are continuous forested regions
extending from the mountains through an otherwise agricultural area
(Figs.1 and 6). Their coarse nature (locals refer to these surfaces as stone
runs) and high water tables have rendered them difficult-at-best for
farming and residential uses. Note the wide spacing of the contours in
Fig. 2 on the jökulhlaup surfaces. The axial gradient of these jökulhlaup
surfaces averages 1­2°, making them very distinctive in an otherwise
hilly topography (Fig. 8). This low gradient will be addressed later in
discussions related to emplacement processes.
3.2. Volumes of ice-dammed lakes and jökulhlaups
A combination of selected geophysical surveys, excavations, and
household well records shows that the jökulhlaup deposits average
about 7 to 10 m in thickness, with isolated pockets up to 20 m thick
(Fig. 7). Fig. 7D shows an electrical resistivity survey along the
northern segment of the North Branch Buffalo Creek deposit, which is
fairly typical of several such surveys that have been made in the region.
Because of the abundance of a fine-grained matrix in these sediments,
ground penetrating radar surveys proved relatively uninformative.
Table 1 provides estimates of the aerial extent and volume of these icedammed
lakes, peak discharge, and volume or the associated
jökulhlaup sediments. The lake on Rapid Run was the largest, having
a volume of nearly 1 km3
. Lake depths averaged between 70 and 80 m,
with depths at the ice dams averaging between 120 and 140 m.
Estimates of peak discharge are rough at best, based on comparisons to
historical observations of jökulhlaups. The duration for emptying the
lakes is thought to have been rapid. Eskilsson et al. (2002) showed that
the large jökulhlaup in 1996 at Skeiarsandur, Iceland drained in less
than 48 h. Peak discharge was estimated assuming most of the flow
was discharged in 48 h, following a hydrograph similar in shape to the
one observed by Eskilsson et al. (2002). Discharge estimates are in the
Fig. 8. Jökulhlaup surface morphology. (A) Oblique aerial view of part of the northern segment of the North Branch surface (foreground) and Rapid Run surface (background).
(B) Electrical resistivity survey across North Branch surface at the location shown in Figs. 7B and 8A. (C) View across the Rapid Run surface near Forest Hill. Here, the surface is 1.2 km
wide. (D) View upslope of the North Branch surface just north of Mifflinburg near Johnstown.
Table 1
Estimated dimensions of the ice-dammed lakes and associated jökuhlaups.
Fan Approx.
lake
Approx.
lake
Approx.
lake
Peak Mean
sediment
Fan Fan
sediment
volume
(km3
)
depth
(m)
area
(km2
)
discharge
(cms)
size
(mm)
slope volume
(km3
)
Rapid Run 0.973 90 6.61 16,500 210 0.0129 0.07
North Branch 0.215 60 3.56 4800 200 0.0135 0.09
Spruce Run 0.27 70 3.41 3000 230 0.0075 0.03
Laurel Run 0.73 80 6.43 11,500 180 0.0148 0.03
Stony Run 0.09 50 1.71 2100 190 0.042 0.008
Peak discharge estimated based on assumption of draining the lakes in 24 h, using the
jökuhlaup breakout model summarized in Costa (1988a) and following a hydrograph
similar to that observed for the 1996 jökuhlaup in Iceland by Eskilsson et al. (2002).
88 R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
range expected using jökulhlaup breakout models based on the
relationship between lake volume and dam height shown as a
compilation of studies by Costa (1988a). The ice-dammed lakes were
not large, nor were their paleoflood discharges exceptional, but, the
volume of coarse sediment deposited and its influence on the
geomorphology of the region was significant. The summation of peak
flows from the jökulhlaups described here exceeds that of the largest
historical flow recorded 150 km downstream on the Susquehanna
River (at Holtwood) in 1972 from Hurricane Agnes.
3.3. Rapid Run surface
The largest and best preserved of the jökulhlaup surfaces is the one
in Rapid Run (Figs. 1, 2 and 6). Most of this surface is forested with
spotty low-density home development. The Rapid Run surface
extends for more than 6 km until it merges with the upper of two
terraces of Buffalo Creek at the village of Cowan (Figs. 2 and 6). There,
sediment influx appears to have caused significant aggradation of
Buffalo Creek, as evidenced by a notable flattening of its gradient for
several kilometers upstream (Fig. 9A). The volume of coarse gravel
input from Rapid Run appears to have played a major role in the
creation of a thin fill terrace downstream for at least 4 km along
Buffalo Creek. The Rapid Run deposit is widest (1.2 km) near its
upstream emergence from the mountains, where it attains full width
abruptly, and narrows to 0.8 km just before it merges with Buffalo
Creek. The Rapid Run surface is strewn with sandstone boulders and is
exceptionally smooth (Fig. 8C). Rapid Run and a series of small,
subparallel tributaries have barely incised into its surface. Intense
rainfalls typically result in the inundation of much of the surface as
these channels rapidly flow over their low banks of b1 m.
3.4. North Branch Buffalo Creek surface
The surfaces formed by the jökulhlaups from North Branch Buffalo
Creek are volumetrically the largest of the deposits described here
(Fig. 7B and C). However, because of their complex history and postdepositional
evolution, they are more discontinuous than the Rapid Run
surface. The North Branch surface appears to have had at least two major
episodes of formation, resulting in a northern and southern arm,
separated by a broad area (several kilometers) of exposed bedrock with
minor post-glacial drainage development and scattered till (Figs. 2, 6B,
7B, C). The northern segment is the most distinct and continuous,
extending in a 9-km E­SE arc from the mountain front to its confluence
with Buffalo Creek just north of Mifflinburg. Like its counterpart on
Rapid Run, the North Branch surface is quite bouldery, of low gradient,
and supports a system of shallowly incised braided channels during high
flow. The northern segment is mostly forested, has a few low-density
home developments, and is drained by the braided North Branch Buffalo
Creek channels which have barely incised into its surface. The southern
segment of the North Branch surface is less well defined and flows south
and then SE until it joins the upper terrace of Buffalo Creek a few
kilometers west of Mifflinburg (Figs. 2 and 7B, C). Because of the middle
Pleistocene age of both segments, it is difficult to discern differences in
the weathering of clasts or soil development between the northern and
southern flows. The southern segment appears to be the oldest,
supported by its position 2­3 m above the northern segment at its
origin and its more dissected nature related to post-glacial stream piracy
and erosion. Most of the southern segment is also forested, but lacks a
well-defined stream like the other jökulhlaup surfaces. This segment
was probably the path of an earlier North Branch jökulhlaup, which later
shifted eastward to form the northern segment. Sediments from the
southern segment can be traced eastward across a paleochannel of
Buffalo Creek where it rejoins the northern segment at the village of
Johnstown (Figs. 2, 7 and 8), 1.5 km upstream from its confluence with
Buffalo Creek. Some of the sediment from the southern segment also
flowed directly into Buffalo Creek west of Mifflinburg. This region has
been extensively modified by post-glacial erosion and drainage
development since the middle Pleistocene. As a result, the original bed
of the southern jökulhlaup now occupies minor divides in the region in
some of its downstream reaches. Evidence of the former valley floor are
the deposits of rounded Tuscarora Sandstone boulders as thin terraces
on these low divides. Post-glacial streams have incised along the
margins of the bouldery fill, finding it easier to erode the soft shales of
the valley floor bedrock.
3.5. Other jökulhlaup surfaces
Although less extensive, similar elongate bouldery jökulhlaup
surfaces occur downstream from the mountain fronts along White
Deer Creek, Spruce Run, and Laurel Run (Figs.1 and 2). The White Deer
jökulhlaup surface is relatively narrow, extending along the lower
6 km of the stream. No till or other direct evidence of the middle
Plesitocene ice has been observed at elevations close to those where
tills are found in tributaries to the west. Therefore, even though the
White Deer watershed is larger than Spruce Run, Rapid Run, and North
Branch Buffalo Creek watersheds, its jökulhlaup surface is relatively
small. Apparently, the ice was unable to form a significant or longenough-lasting
dam to create a large lake here. There is, however,
evidence of two jökulhlaup events in White Deer Creek. A small
remnant of an older surface occurs near the upstream edge of the
deposit along the north side of the creek. The older deposit is about
4 m above the more extensive younger jökulhlaup surface.
Fig. 9. (A) Longitudinal profile of Buffalo Creek. The moderately-sloping segment is the
large debris fan where the creek exits the mountain-front. Note the slight shallowing of
gradient between North Branch and Rapid Run, and the slight steepening downstream.
(B) Longitudinal profiles of tributaries, and related features. Quaternary debris fans are
common in the region, normally having slopes between 6 and 12° (Berntsen, 1993),
while jökulhlaup surfaces average 1­2°.
89R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
The Spruce Run surface (Figs. 1 and 2) is significantly narrower
than the Rapid Run or North Branch surface because it occurs along a
reach incised into more resistant sandstone bedrock along the plungeout
of one of the anticlines. Upstream of the sharp bend in the surface,
the deposit appears to be between 0.5 and 1 km wide, but much of it is
masked by a large reservoir. The lower kilometer of the Spruce Run
surface widens again to more than 0.5 km before it merges with the
upper terrace of Buffalo Creek at the village of Mazeppa.
The jökulhlaup surfaces along Laurel Run near the western margin
of the study area (Figs. 1 and 2) are much smaller and are tributary to
the now separate Penns Creek watershed. These surfaces are typically
b0.25 km wide, but extend several kilometers downstream to Penns
Creek. The smaller size of these surfaces reflects the significantly
smaller drainage basin areas of these tributaries located along the
terminal morainal belt from the glacial lobe.
The surface in Stony Run (Fig. 2) has a slightly higher gradient
ranging from 4 to 6° (Fig. 9B) and may have formed as a simple debris
fan fed by gelifluction sediments, without having an ice-dammed lake
and jökulhlaup. The lower 2 km of the Stony Run surface abruptly
levels to 1­2°, and appears similar to other jökulhlaup surfaces in the
region before it grades into the Laurel Run jökulhlaup surface at the
north edge of the village of Laurelton.
3.6. Sedimentology
Information on the sedimentology of the jökulhlaups deposits
comes from a variety of sources, including observations of terrace and
bank exposures; shallow excavations for homes, highway drainage
ditches and pipelines; water well records; and isolated geophysical
surveys, including electrical resistivity, ground penetrating radar, and
seismic refraction. The jökulhlaup deposits are generally matrixsupported
bouldery gravels with indistinct stratification (Fig.10). Rare,
poorly developed imbrication was noted in some exposures, but
overall no fabric developed in these sediments. Similar matrixsupported
sediments occur in jökulhlaups deposits formed on a 2­3°
gradient along the southern coast of Iceland (Fig. 10D). Recent studies
Fig. 10. Sediments in the jökulhlaup deposits. (A) Bouldery surface of the Rapid Run jökulhlaup upstream of Cowan. (B) Trench excavated in the North Branch Buffalo Creek surface
north of Mifflinburg. Note the poor sorting and matrix support. (C) Sediment piled from a home foundation excavated in North Branch Buffalo Creek surface north of Mifflinburg.
(D) Similar matrix-supported sediment from the 1996 jökulhlaup deposits of Skeidarsandur, Iceland.
90 R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
in Iceland have discussed the occurrence of matrix-supported gravels
formed by fluidized flows similar to those described here (Kudsen and
Russell, 2002; Marren, 2002; Rushmer, 2006; Carrivick, 2007) Clasts
are composed primarily of rounded to subrounded very resistant
Tuscarora Sandstone with minor amounts of Bald Eagle Sandstone.
Clast size varies, but most average 15 to 25 cm in intermediate axis
diameter. Larger clasts with intermediate diameters between 35 and
40 cm are not uncommon. Matrix sediment is a mix of sand and clay.
Much of the clay appears to be pedogenic, giving the deposits (as well
the till of similar age) a distinctive orange color with B horizon color in
the 5 YR to 2.5 YR hue range. Clast size is somewhat greater for the
most of the upstream deposits to the west (Laurel Run and Stony Run).
Although no stratification has been recognized in any of the
excavations, some foundation trenches reveal two distinctive deposits
of similar nature separated by a moderately well-developed paleosol.
No fine-grained units have been observed in any of the excavations,
some of which penetrated 3 m into the gravels, suggesting that the
jökulhlaup sediments were deposited by single, rapid sedimentation
events.
3.7. Transport flow mechanics/emplacement processes
Deposition of the sediments in these surfaces appears to be from
breakout floods, or jökulhlaups, related to catastrophic failure of small
mountain-front lakes dammed by ice from the glacial lobe that was
advancing westward up Buffalo Creek valley. The exact nature of the flow
mechanics and emplacement of the sediment is less certain. The
sedimentology of the jökulhlaup deposits is consistent with that of debris
flow, probably at the water-rich end of the spectrum approaching that of
sediment-rich hyperconcentrated flow (Costa,1988b). The difficulty with
a debris flow emplacement process is the low gradient present on these
landforms. Debris flow runout generally ceases at slopes of 7­8° because
of the deformation dynamics of these Bingham flows (Johnson, 1970;
Costa, 1988b; Ritter et al., 2006). The steepest of the surfaces attains a
gradient of about 4°, but the larger surfaces have slopes b2°.
Clearly the jökulhlaups were highly sediment-charged. The
abundance of landforms reflecting active periglacial processes on
hillslopes in the area (blockfields, stone stripes, gelifluction lobes, and
thermokarst features) is strongly suggestive that large volumes of icerich
sediment were delivered to the temporary ice-dammed lakes
prior to their breakout floods. It is less clear whether catastrophic
breakout floods could have transported the sediment with the
rhealogy of a debris flow on the low valley gradients. Discussions
with Richard Iverson (U. S. Geological Survey, personal communication,
2003) confirmed that it may have been possible to develop high
basal pore-water pressures downstream of the ice dams because of
permafrost during the glacial advance. This situation, combined with a
likely high volume of ice in both the lake water and ice incorporated
along the flow route from the catastrophically disintegrating glacial
dam, may have provided a mechanism for rapid emplacement of these
sediment-charged, slushy flows on these low gradients. It is also
possible that warming during the glacial disintegration may have
triggered widespread hillslope water and sediment additions to the
lakes in the form of flows similar to the slush avalanches described in
high latitude periglacial regions (Rapp and Nyberg, 1981; Nyberg,
1989). In any case, the jökulhlaup flows appear to have been similar to
viscous debris flows but moving on an extraordinarily low gradient.
Although subaqueous debris flows (density currents) can also occur
on similar low gradients, this process is unlikely here as an
emplacement process given the topographic setting because these
surfaces are downstream of the lakes.
Another emplacement process that could explain transport of
coarse-grained matrix-rich sediment on such gentle gradients is
related to the occurrence of a major flood wave as the ice-dam
abruptly collapsed. Significant flood waves have been described
during ice-dam breakouts around the world (Costa, 1988a) and in
accounts of major dam breaks with direct witnesses, such as the
Johnstown, PA flood of 1889. Elevated energy gradients associated
with significant flood waves can readily transport large volumes of
coarse-grained sediment on relatively gentle gradients. Given that the
ice-dammed lakes described here had depths ranging from 100 to
150 m at the ice-dams, it is quite likely that sudden collapse of the ice
dam would have resulted in flood wave heights N10 m, capable of
transporting the sediment-rich flows on low-gradient surfaces.
4. Downstream impacts
4.1. Interaction with mainstem streams
The volume of sediment contributed to the mainstem Buffalo Creek
and Penns Creek during these jökulhlaup events was enormous. The
largest, and highest, terraces along Buffalo Creek grade downstream
from the jökulhlaup surfaces. This sediment appears to have remained
as a lag following the breakout flows, forming a thin fill terrace. Major
terraces along both streams can be tied directly to gravels sourced from
the jökulhlaups. The channels of Buffalo and Penns Creeks are
dominated by the sandstone gravel supplied from these events
today, even though the downstream two-thirds of their watersheds
are underlain by shale and limestone bedrock. Competence studies and
point bar observations show that most of the gravel is entrained during
bankfull to slightly overbank flows (Forsburg and Kochel, 2007). The
longitudinal profile (Fig. 9A) of Buffalo Creek reflects some of the
impact of the gravel contributed by North Branch and Rapid Run. The
reach between these two tributaries is notably flatter than upstream
and downstream segments, probably from aggradation and partial
damming by the jökulhlaup sediment inputs to the mainstream.
Aggradation at the mouth of Rapid Run may have partially dammed
Buffalo Creek, resulting in up to 30 m of fill. Sediment poured into
Buffalo and Penns Creeks, triggering significant aggradation and a
minor decline of their axial gradients. Evidence for temporary ponding
of Buffalo Creek upstream of the tributaries affected by jökulhlaups
includes its exceptionally wide valley floor, as compared to the
upstream and downstream reaches. Additionally, in the reach between
Rapid Run and North Branch Buffalo Creek, the bedload in the main
channel appears finer than upstream and downstream (Fig. 9A). This
section of the reach is where gradient is slightly lower, suggesting that
there may have been some partial damming by the sediment from
Rapid Run causing enhanced mainstream aggradation. Buffalo Creek
appears underfit along most of its course through its broad valley along
this reach. Note in Fig. 9B the low gradient characteristic of the
jökulhlaup surfaces (North Branch, Rapid Run, Spruce Run, and Laurel
Run) compared to steeper profiles for debris fans (Buffalo Creek at
Buffalo Gap and Stony Run fans). These debris fans, with gradients
between 5° and 8° appear to have formed by normal hyperconcentrated
flow and debris flow processes unrelated to ice-dammed lakes
and jökulhlaups. Soils on the Stony Run surface are similar in age to the
middle Pleistocene till and jökulhlaups. The very large fan emanating
from Buffalo Gap (where Buffalo Creek emerges from the mountain
front) appears to have formed subsequent to these events, in the Late
Pleistocene, based on soil and weathering observations. The large
debris fan at Buffalo Gap (Figs. 2 and 5D) appears to have prograded
over the middle Pleistocene till.
4.2. Lower Susquehanna River gorge
The lower 40 km of the lower Susquehanna River in the
Pennsylvania Piedmont Province is deeply incised through a narrow
bedrock gorge cut into resistant metamorphic rocks of Pre-Cambrian
age. The bedrock canyons, known locally as the Susquehanna Deeps
(Matthews, 1917; Thompson and Sevon, 1999) of the Lower
Susquehanna gorge were scoured below sea level by catastrophic
macroturbulent flood flows in the Pleistocene (Kochel and Parris,
91R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
2000; Reusser et al., 2004). Evidence of these megafloods in the lower
gorge includes large potholes, longitudinal grooves, hydraulically
trapped erratic boulders from sources 100 km upstream, and multiple
bedrock strath terraces (Fig. 11). It is likely that the upstream
jökulhlaups, and much larger breakout flows associated with the
draining of Glacial Lake Lesley above Williamsport(the volume of Lake
Lesley was 10 to 15 times larger than these lakes combined),
contributed significantly to the catastrophic paleofloods that were
integral in scouring the bedrock channels of the lower gorge. Although
the bedrock straths have been cosmogenically dated as Late
Pleistocene (Ruesser et al., 2004), it is possible that the major bedrock
erosion features within the lower gorge were cut earlier and then
modified by Late Pleistocene flows. Step-backwater modeling based
on erratics and high-level slackwater sediments in the lower gorge
(Kochel and Parris, 2000; Hubacz, 2009) documented paleoflood
flows up to 142,000 m3
/s (5,000,000 ft3
/s); five times the largest
historical flow observed from Tropical Storm Agnes in 1972 (Moss and
Kochel, 1978).
5. Post-glacial landform evolution and topographic inversion
The most prominent impact of the jökulhlaups on post-glacial
landscape evolution of the region has been the armoring of the valley
floors with the resistant bouldery sandstone gravel. Post-glacial
hydrologic regimes have been largely unable to mobilize the jökulhlaup
sediments and unable to significantly incise these flow surfaces. The
primary channels of the tributaries to jökulhlaup surfaces have been
able to incise their channels b1 m. As a result, these surfaces have
relatively high water tables and contain wetlands and wet forest
ecosystems. Although the Pleistocene sediment is no longer mobilized
by modern floods, high magnitude events readily inundate the
jökulhlaup surfaces. Post-glacial incision has occurred in areas between
the tributaries and along their margins, resulting in the inversion of
topography of some of the major jökulhlaup surfaces. The most
prominent of these is along the upper half of the Rapid Run surface
near the village of Forest Hill (Fig. 12). There, post-glacial incision by
Muddy Creek into the softer shales and thin sandstones of the
Bloomsburg Formation has resulted in perching of the Rapid Run
jökulhlaup surface about 15 m above the floor of Muddy Creek. Postglacial
streams have found it easier to downcut into the softer bedrock,
having been unable to entrain the resistant Tuscarora sandstone
boulders of the jökulhlaup surfaces. This amount of downcutting is
consistent with studies of regional denudation rates for the upper
Susquehanna River region. Erosion rates have been estimated at about
2 cm/1000 y usingmodernsuspended sedimentdata (Judson and Ritter,
1964) as well as long-term stratigraphic evidence from the Cenozoic
sedimentary record (Sevon, 1999). Incision by post-glacial drainage has
been slightly less than the 18 m expected, but certainly within the
general range.
Inversion of topography along piedmont areas from armoring by
glacial outwash composed of resistant clasts has been well documented
along the Beartooth Front in Montana (Ritter, 1967). Ritter (1967)
illustrated the importance of this process in the evolution of
mountain-front geomorphology where Pleistocene outwash terraces
along Rock Creek and other streams emanating from the Beartooth
Mountains decrease in age as they step down, with the oldest terrace
Fig. 11. Pleistocene paleoflood features in the lower gorge of the Susquehanna River near Holtwood. (A) Orbital view of the central and lower Susquehanna Valley and the Holtwood
Gorge (courtesy, NASA). (B) Large potholes and bedrock straths cut in resistant pre-Cambrian schist by macroturbulence during the paleofloods. (C) Large pothole with boulder
erratics trapped inside. Some of the erratics (like diabase) were sourced near Harrisburg, while some of the conglomerates originated from up to 50 km farther upstream.
92 R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
now forming major watershed divides as topography has been
inverted (Fig. 13). Similar to the case in Pennsylvania, post-glacial
streams were unable to downcut through the armored gravel surfaces
of the outwash but were able to more readily incise through softer
bedrock in the Bighorn Basin. In Pennsylvania, with time, this process
will continue to elevate the former valley floors covered by jökulhlaup
sediment, increasing the prominence of their legacy on landform
evolution in the Buffalo Creek watershed.
6. Conclusions
The very unique interaction of an upvalley-flowing, thin glacial
lobe with high gradient mountain streams along eroded plunging
anticlines of the Appalachian Valley and Ridge resulted in a series of
sediment-rich jökulhlaups, normally reserved to high latitude
mountainous streams and alpine glaciers. The geomorphic circumstances
that occurred in tributaries to the West Branch Susquehanna
River during early to middle Pleistocene glaciation is extremely rare
and may be unique in the world. Small lakes formed upstream from
the mountain-front ice dams and likely filled with water and sediment
from active periglacial denudational processes upstream in their
watersheds. Failure of the ice dams resulted in catastrophic breakout
floods as the ice advance waned. The jökulhlaups produced major
piedmont landforms that extend as elongated fan-like surfaces
composed of poorly sorted, matrix-supported gravel that are
sedimentologically similar to debris flow sediments. However, the
gradient of these surfaces generally ranges from 1 to 2°, far too low for
normal debris flow processes. Enhanced basal pore-water pressures
may have developed from the interaction of glacial meltwater and
permafrost. This, in combination with catastrophic release of icedammed
lakes as large flood waves, produced a special set of
hydrogeomorphic conditions that facilitated the emplacement of
these sediment-rich jökulhlaups along mountain-front tributaries to
Buffalo and Penns Creeks in a series of catastrophic flow events.
Valley fills from these jökulhlaups averaged 7 to 10 m, producing
gently-sloping, wide, smooth surfaces between 6 and 9 km in length
and averaging 0.5 to 1 km in width. These landforms are prominent
today, nearly a million years after their formation, and are readily
visible from orbital imagery because of their forest cover. The surfaces
remain forested chiefly because of their bouldery nature, and a high
water table occurs because of poor drainage and limited incision by
surface streams. Although most of the Pleistocene sediment is no
longer mobilized by modern floods, high magnitude events readily
inundate the jökulhlaup surfaces.
Fig. 12. Inversion of topography of the Rapid Run surface near Forest Hill, north of Mifflinburg. (A) Northeast to southwest topographic profile from the post-glacial Muddy Creek
Valley (incised into relatively soft shale and sandstone bedrock) across the more resistant, boulder-strewn Rapid Run jökulhlaup surface. Rapid Run has been unable to incise into the
bouldery deposits. Formerly the valley floor, this surface now forms a modest divide between Rapid Run and Muddy Creek. Sediments in Muddy Creek are local sandstone and shale
and are much more angular than the rounded Tuscarora Sandstone boulders in the jökulhlaup. (B) View of the change in elevation along the northeastern margin of the surface. The
floor of the photo (near the truck) is the floodplain of post-glacial Muddy Creek. Bedrock can be seen along the left side of the road, bordering the edge of the elevated jökulhlaup
surface.
93R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
Fig. 13. Inverted topography of older terraces along Rock Creek near Red Lodge, Montana along the front of the Beartooth Mountains mapped by Ritter (1967). (A) Air photo showing
Rock Creek and Mesa Bench. Mesa bench currently forms a major divide between Rock Creek and Bear Creek, but was formerly the middle Pleistocene valley floor of Rock Creek. (B)
Oblique view looking north along the direction of the arrow in Fig. 13 A, showing the inverted topography of Mesa Bench.
94 R.C. Kochel et al. / Geomorphology 110 (2009) 80­95
The legacy of the jökulhlaups has considerably impacted the
nature of the evolution of the piedmont region by post-glacial
drainage in a manner similar to that described by Ritter (1967)
along the Beartooth Front in Montana. Post-glacial streams have not
had the competence to mobilize the jökulhlaup sediment. Thus, the
tributaries have incised only negligibly into the bouldery fill. Postglacial
streams whose headwaters are valleyward of the ridges have
been able to downcut into softer valley bedrock (thin, less competent
sandstones and shales), leaving the jökulhlaups surfaces higher and
developing an inversion of topography.
A paleosol seen in some deeper excavations suggests that
jökulhlaups may have occurred episodically throughout the early
and middle Pleistocene. This interpretation is consistent with the
work of Peltier (1949) on Susquehanna River terrace chronosequences,
description of multiple Pre-Wisconsin tills by Marchand and
Crowl (1991), and the occurrence of remnants of higher jökulhlaup
surfaces along the southern branch of the deposits from North Branch
Buffalo Creek. It is unclear whether Penns Creek itself was dammed by
ice to create a major lake in the canyon to the west of the study area,
but the presence of high terraces with early-to-middle Pleistocene
soils and similar sedimentology suggests that that may have occurred
in a similar fashion. These catastrophic breakout flood events
undoubtedly contributed to the rapid evolution and down-cutting of
the lower gorge of the Susquehanna River, some 150 km downstream.
Acknowledgements
This paper is dedicated to the memory and legacy of the late Louis
C. Peltier who pioneered glacial and periglacial geomorphic research
along the Susquehanna River. His insights were exceedingly progressive
and his attention to detail was amazing. We had the fortune of
benefiting from his visit to the field during the early stages of this
work and numerous insightful discussions in subsequent years.
We thank the students and faculty of Bucknell's Geomorphology
and Environmental Geophysics classes for their assistance in running
several of the electrical resistivity, ground penetrating radar, and
seismic surveys. Many landowners in the Buffalo Valley granted us
access to their property, for which we are appreciative. Tim Diggs
helped in the comparative fieldwork in Iceland. Carl Kirby kindly
provided the hillshade figure.
References
Anderson, S.P., Walder, J.S., Anderson, R.S., Kraal, E.R., Cunico, M., Fountain, A.G.,
Trabants, D.C., 2003. Integrated hydrologic and hydrochemical observations of
Hidden Creek Lake jökulhlaups, Kennicott Glacier, Alaska. Journal Geophysical
Research 108, 4­1­4-19.
Baker, V.R., 1973. Paleohydrology and sedimentology of Lake Missoula Flooding in
Eastern Washington. Special Paper 144. Geological Society of America, Boulder, CO,
p. 174.
Berntsen, J.P. 1993. The morphology and sedimentology of alluvial fans and debris flow
fans in the Ridge and Valley Province, central Pennsylvania. Senior Thesis,
Department of Geology, Bucknell University, Lewisburg, PA.
Braun, D., 1989. Glacial and periglacial erosion of the Appalachians. In: Gardner, T.W.,
Sevon, W.D. (Eds.), Appalachian geomorphology. Geomorphology, vol. 2, pp. 233­256.
Braun, D.D., 1994. The ubiquitous boulder colluvium of eastern Pennsylvania, a relict of
periglacial mobilization of the entire land surface. In: Braun, D.D. (Ed.) Late
Wisconsinan to pre-Illinoian(G) Glacial and Periglacial Events in Eastern
Pennsylvania. Guidebook for the 57th Field Conference, Friends of the Pleistocene
Northeastern Section, May 20-22, Hazleton, Pennsylvania. Open-File Report, U. S.
Geological Survey 94-434, pp. 34-36.
Bucek, M.F., 1976. Pleistocene geology and history of the West Branch Susquehanna
River valley near Williamsport, Pennsylvania. Ph.D. Dissertation, Pennsylvania State
University, University Park, PA.
Bunting, P. 2005. Emplacement Processes of Quaternary Blockfields in Central
Pennsylvania. Senior Thesis, Department of Geology, Bucknell University, Lewisburg,
PA.
Carrivick, J.L., 2007. Hydrodynamics and geomorphic work of jokulhlaups (glacial
outburst floods) from Kverkfjoll Volcano, Iceland. Hydrological Processes 21,
725­740.
Carrivick, J.L., Tweed, A.J., Russell, F.S., 2003. Glacier outburst floods (jokulhlaups) from
Kverkfjoll, Iceland; flood routeways, flow characteristics and sedimentary impacts.
XVI INQUA congress; Shaping the Earth; a Quaternary perspective Congress of the
International Union for Quaternary Research, vol. 16, p. 148.
Costa, J.E., 1988a. Floods from dam failures. In: Baker, V.R., Kochel, R.C., Patton, P.C.
(Eds.), Flood Geomorphology. Wiley, New York, pp. 439­463.
Costa, J.E., 1988b. Rheologic, geomorphic, and sedimentologic differentiation of water
floods, hyperconcentrated flows, and debris flows. In: Baker, V.R., Kochel, R.C.,
Patton, P.C. (Eds.), Flood Geomorphology. Wiley, New York, pp. 113­122.
Denny, C.S., 1956. Surficial geology and geomorphology of Potter County, Pennsylvania.
U. S. Geological Survey Professional Paper, 288.
Eskilsson, C., Arnason, J.I., Rosbjerg, D., 2002. Simulation of the jokulhlaup on
Skeidararsandur, Iceland, in November 1996 using MIKE 21 HD. In: Snorrason, A.,
Finnsdottir, H.P., Miss, M. (Eds.), The extremes of the extremes: extraordinary
floods, vol. 271. IAHS-AISH Publication, pp. 37­43.
Forsburg, A., Kochel, R.C., 2007. Adjustments of stream channel morphology to land use
in Buffalo Creek tributaries, central Pennsylvania. Geological Society of America
Abstracts with programs. 39, no.1, 43.
Hubacz, D. 2009. Catastrophic paleofloods on the lower Susquehanna River; An analysis
of slackwater deposits in Holtwood Gorge, Pennsylvania. Senior Thesis, Department
of Geology, Bucknell University, Lewisburg, PA.
Johnson, A.M., 1970. Physical processes in geology; a method for interpretation of
natural phenomena; intrusions in igneous rocks, fractures and folds, flow of debris
and ice. Freeman, San Francisco, 577 pp.
Judson, S., Ritter, D.F., 1964. Rates of regional denudation in the United States. Journal of
Geophysical Research 69, 3395­3401.
Kochel, C.R., Parris, A., 2000. Macroturbulent erosional and depositional evidence for largescale
pleistocene paleofloodsin the lowerSusquehanna bedrock gorge near Holtwood,
PA. Geological Society of America Abstracts with Programs, vol. 32, pp. A­28.
Kochel, R.C., Nickelsen, R.P., Eaton, L.S., 2003. Elongated low-gradient debris fans:
Jökulhlaups resulting from tributary damming by Pre-Wisconsinan glacial ice lobes
moving up Buffalo Creek valley in central Pennsylvania. Geological Society of
America Abstracts with Programs. 35, no. 6, 335.
Kudsen, O., Russell, A.J., 2002. Jökulhlaup deposits at the Asbyrgi Canyon, northern
Iceland; sedimentology and implications for flow type. In: Snorrason, A.,
Finnsdottir, H.P., Moss, M. (Eds.), The Extremes of the Extremes; Extraordinary
Floods, vol. 271. IAHS-AISH Publication, pp. 107­112.
Malonee, J.D., 2001. Pleistocene paleochannel abandonment of Penns Creek in Union
County, Pennsylvania. Senior Thesis, Department of Geology, Bucknell University,
Lewisburg, PA.
Marchand, D.E., Crowl, G.H., 1991. Surficial geologic map of parts of Union and Snyder
counties, Pennsylvania. U. S. Geological Survey, Miscellaneous Investigations Series
I-2051.
Marren, P.M., 2002. Criteria for distinguishing high magnitude flood events in the
proglacial fluvial sedimentary record. In: Snorrason, A., Finnsdottir, H.P., Moss, M.
(Eds.), The Extremes of the Extremes; Extraordinary Floods, vol. 271. IAHS-AISH
Publication, pp. 237­241.
Marsh, B., 1987. Pleistocene pingo scars in Pennsylvania. Geology 15, 945­947.
Matthews, E.B., 1917. Submerged "deeps" in the Susquehanna River. Geological Society
of America Bulletin 28, 335­346.
Moss, J.H., Kochel, R.C., 1978. Unexpected geomorphic effects of the Hurricane Agnes
storm and flood, Conestoga drainage basin, south-eastern Pennsylvania. Journal of
Geology 86, 1­11.
Nyberg, R., 1989. Observations of slushflows and their geomorphic effects in the
Swedish mountain area. Geografiska Annaler. Series A: Physical Geography 71,
185­198.
Peltier, L.C., 1949. Pleistocene terraces of the Susquehanna River, Pennsylvania. Pa. Geol.
Survey, 4th ser., Bulletin G23, 187 pp.
Peltier, L.C., 1950. The geographic cycle in periglacial regions as it is related to climatic
geomorphology. Annals of the Association of American Geographers 40, 214­236.
Ramage, J., Gardner, T.W., Sasowsky, I.D.,1998. Early Pleistocene Glacial Lake Lesley, West
Branch Susquehanna River valley, central Pennsylvania. Geomorphology 22, 19­37.
Rapp, A., Nyberg, R.,1981. Alpine debris flows in northern Scandinavia: Morphology and
dating by lichenometry. Geografiska Annaler. Series A: Physical Geography 64,
183­196.
Reusser, L., Bierman, P., Pavitch, M., Zen, E-an., Larsen, J., Finkel, R., 2004. Rapid Late
Pleistocene incision of Atlantic passive margin river gorges. Science 305, 499­502.
Rushmer, E.L., 2006. Sedimentological and geomorphological impacts of the jokulhlaup
(glacial outburst flood) in January 2002 at Kverkfjoll, northern Iceland. Geografiska
Annaler. Series A: Physical Geography 88, 43­53.
Ritter, D.F., 1967. Terrace development along the front of the Beartooth Mountains,
southern Montana. Geological Society of America Bulletin, 78, 467­483.
Ritter, D.F., Kochel, R.C., Miller, J.R., 2006. Process Geomorphology, 4th ed. Waveland
Press, Long Grove, IL. 560 pp.
Sevon, W.D., 1999. Geologic history, Cenozoic. In: Schultz, C.H. (Ed.), The Gology of
Pennsylvania. Special Publication, Geological Survey of Pennsylvania, pp. 450­455.
Sevon, W.D., and Braun, D.D., 1997, Glacial Map of Pennsylvania. Pennsylvania
Geological Survey, Map 59.
Thompson, G.H., Sevon, W.D., 1999. The Susquehanna River deeps; an erosional enigma.
Geological Society of America, Abstracts with Programs. 31, no. 7, 49.
Waitt Jr., R.B., 1985. Case for periodic, colossal jökulhlaups from Pleistocene glacial Lake
Missoula. Geological Society of America Bulletin 96, 1271­1286.
Waitt Jr., R.B., 2002. Great Holocene floods along Jokulsa a Fjollum, North Iceland. In:
Martini, I., Baker, V.R., Garzon, G. (Eds.), Flood and megaflood processes and
deposits; recent and ancient examples. Special Publication of the International
Association of Sedimentologists, vol. 32, pp. 37­63.
95R.C. Kochel et al. / Geomorphology 110 (2009) 80­95