Influence of large woody debris on the morphology of six
central European streams
Jochem Kail
Institute of Ecology, Faculty of Hydrobiology, University of Essen, Universitätsstr. 5, D-45117 Essen, Germany
Received 8 July 2001; received in revised form 5 February 2002; accepted 18 October 2002
Abstract
The impact of large fallen trees on channel form is described for six short stream sections in central Europe influenced by
large woody debris (LWD sections), five of which are compared to nearby reference sections free of LWD (reference sections).
Three-dimensional models of streambed topography were generated by surveying cross-sections with a spacing of 1 per 1/15
channel width. Parameters derived from digital terrain models and cross-sections compared between LWD sections and
reference sections include the extent of pools, bars, and cutbanks, streambed and bank complexity, cross-sectional area, width,
depth, and cross-section complexity as described by Andrle's [Math. Geol. 26 (1994) 83] `angle-measurement-technique'
(AMT analysis), a measure of the deviation of a cross-section line from a straight line. Structural diversity is greater in LWD
sections at almost all spatial scales, particularly in terms of pool volume (Mann­Whitney U-test, p < 0.01) and cross-section
complexity described by median angle of AMT analysis (Mann­Whitney U-test, p < 0.05). Large pools are clearly associated
with large fallen trees and attain volumes up to 36 m3
. With the exception of the ratio of one LWD section where the fallen tree
is oriented parallel to flow, the ratio of pool volume to bed planimetric area ranges from 424 to 693 m3
/ha, which is in the upper
range reported for small, high-gradient streams in Oregon, NW America (229­755 m3
/ha) [Can. J. Fish. Aquat. Sci. 47 (1990)
1103]. Pool volume of LWD sections is strongly correlated to the blockage ratio (Spearman rank order correlation, rs = 0.93,
p < 0.01). Differences in channel morphology between the LWD sections and reference sections indicate a strong morphologic
control of large woody debris in these central European stream sections.
D 2002 Elsevier Science B.V. All rights reserved.
Keywords: Large woody debris; Channel morphology; Digital terrain models; Angle-measurement-technique; Central Europe
1. Introduction
The influence of large woody debris (LWD) on
stream channel morphology has been studied extensively
in small North American streams, but there are
few studies from other regions to allow intra-regional
comparisons. Keller and Swanson (1979) described
general morphological changes generated by LWD for
streams in the Pacific Northwest. Subsequent workers
from this region observed significant increases in bed
load transport after LWD removal (Beschta, 1979;
Bilby, 1981; Klein et al., 1987; MacDonald and
Keller, 1987; Smith et al., 1993) or the breakage of
single log jams (Mosley, 1981). Cross-section width
as well as depth and their variability are higher in
channel sections with high LWD loading (Keller and
Tally, 1979; Hogan, 1987; Fausch and Northcote,
0169-555X/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S0169-555X(02)00337-9
E-mail address: jochem.kail@uni-essen.de (J. Kail).
www.elsevier.com/locate/geomorph
Geomorphology 51 (2003) 207­223
1992; Nakamura and Swanson, 1993; Trimble, 1997;
Buffington and Montgomery, 1999). Murphy et al.
(1986), Carlson et al. (1990), and Fausch and Northcote
(1992) reported a strong correlation of pool
volume and LWD quantity, whereas Evans et al.
(1993) found the effect of wood on pool volume to
be masked by other geomorphic factors such as a
shallow base of bedrock that constrained pool depths.
Comparable investigations are rare for Europe,
where the impact of LWD is far less obvious. However,
because of very different stream characteristics, forest
communities, and management practices, significant
differences may be expected from results reported for
the North American streams. Gregory et al. (1985)
described significant influences of LWD dams on
channel morphology in small to medium sized streams
in S. England. Piégay et al. (1998) described changes in
channel topography and sedimentation in a sixth-order
river (Ain River, France) characterised by complex
LWD accumulations. LWD accumulations in the
French alpine river Drôme were found to be rare and
ephemeral, and therefore to have little effect on channel
morphology (Piégay et al., 1999). Except for one recent
study that quantified LWD-induced changes in channel
depth and width in two small central European streams
(Gerhard and Reich, 2000), little is known about the
impact of LWD on channel morphology in central
European streams and rivers.
This study describes small-scale channel morphology
of six central European stream sections influenced
by single, large fallen trees, and compares five of these
sections to reference sections free of large wood. The
streams studied can be classified as large streams or
small rivers and represent low-gradient meandering
lowland streams and lower mountain streams, two of
the most common stream types in the northwestern
part of Germany. Various parameters derived from
digital terrain models (extent of pools, bars, and
cutbanks; bed and bank complexity) and cross-sections
(area, width, depth, `angle-measurement-technique'
according to Andrle, 1994--a method to measure
the deviation of the cross-section line from a straight
line) are considered. These parameters are tested to
examine potential differences in channel morphology
between stream sections with and without large woody
debris.
This study focuses on small-scale channel morphology
instead of a reach-scale for three reasons.
First, morphological features associated with single
LWD pieces are easy to interpret. Second, the influence
of single, large fallen trees on channel morphology
is of special interest because single trees are
increasingly used in central European river rehabilitation
projects to enhance structural diversity. Third,
appropriate study stream sections with high LWD
loadings on a reach-scale are rarely found in central
European streams.
2. Study streams
Because of the long-lasting human impact and the
management of riparian forests, large fallen trees are
rarely found in central European streams. Stream
managers usually remove LWD from streams for
flood control reasons. However, some single logs
impacting streams can still be found in remote areas.
Six stream sections, most of which are located in
Northrhine-Westphalia (Germany) (Fig. 1), were selected
for this study. Three of the study stream sections
(Lippe, Berkel1, Berkel2) are located in the lowlands of
Northrhine-Westphalia, in low-gradient river plains
dominated by Holocene sediments and Quaternary
sands. Three streams (Ahr, Mo¨hne, Berg. Land) are
located in lower mountainous areas, primarily consisting
of argillaceous shale.
For five of the study streams, a section influenced
by one to three fallen trees (LWD section) and a
nearby reference section without LWD were selected
(for details concerning the selection of reference
sections, see Section 3.1). No comparable reference
section could be found within the short restored reach
of the Lippe River in which the LWD section is
located. Stream morphology up and downstream of
this reach is heavily modified by human and thus not
comparable to the restored reach. Investigations were,
therefore, restricted to the LWD section. The large
fallen tree, the impact of which was investigated in the
Berg. Land stream section, is located in the upper part
of a mid-channel bar. Because no similarly wandering
reference section could be found in the vicinity, the
channel on the right side of the bar was used as a
reference section. This was possible because the
obvious effects of the fallen tree are restricted to the
channel on the left of the bar. The study streams
(LWD sections and reference sections) are characterJ.
Kail / Geomorphology 51 (2003) 207­223208
ised in Table 1. In order to further illustrate the
characteristics of LWD sections, a photo of the
Berkel1 LWD section is given in Fig. 2.
3. Methods
3.1. Experimental design
The impact of large fallen trees on small-scale
channel form is described by comparing channel
morphology of stream sections influenced by large
woody debris (LWD sections) with nearby reference
sections free of LWD (reference sections).
The LWD sections were demarcated based on the
extent of the large fallen tree(s) and the morphological
features (pools, bars, cutbanks, channel widening) in
the area of the fallen tree(s) that were visible or
detectable by wading. Because demarcation of morphological
features in the field was difficult, areas up
and downstream of these sections were mapped to
ensure complete portrayal of these morphological
features. Investigating longer stream sections would
enclose areas not influenced by the fallen tree(s) and,
thus, distort the results because parameters describing
channel morphology are related to bed planimetric
area.
LWD sections are some tens of meters in length and
are located in specific areas of the channel reach (e.g.,
a single riffle; half a meander wavelength; deeply
entrenched, straightened section). Therefore, comparable
stream sections free of large wood were chosen
as reference sections rather than randomly chosen
sections, which would enclose geomorphological features
different from the LWD sections. Because of the
high variability of channel conditions (e.g., riparian
vegetation, slope, discharge, bedrock confinement,
riprap), choosing a reference section as similar as
possible to the LWD section seemed more appropriate
than investigating a greater number of `reference
sections' (e.g., several riffles) to quantify the variability
of the specific channel area.
Fig. 1. Location of the study streams in Northrhine-Westphalia and Rhineland-Palatinate (location of the Ahr study stream), Germany. The
author is under a legal obligation not to exactly locate the Berg. Land stream section.
J. Kail / Geomorphology 51 (2003) 207­223 209
Table 1
Data on the investigated stream sections and the investigated large woody debris (LWD), RS = reference section, LWD = LWD section
Study streams Berkel1 Berkel2 Ahr Mo¨hne Berg. Landa
Lippe
RS LWD RS LWD RS LWD RS LWD RS LWD LWD
Stream characteristics
Catchment area (km2
) 247.5 247.5 538 275 180.5 1906
Mean section width (m) 12.0 17.8 15.4 16.8 22.7 31.3 18.1 20.3 18.1 16.0 40.5
Slope of water level (%) 0.05 0.07 0.05 0.05 0.3 0.1 0.2 0.2 0.1 0.6 0.04
Length
(section mapped, m)
28.0 50.5 33.5 28.5 41.8 67.0 38.0 54.0 30.0 30.0 44.0
Stream type lowland lowland lower
mountain
lower
mountain
lower
mountain
lowland
Bed material sand sand gravel/
cobble/
boulder
gravel/
cobble
gravel/
cobble
sand/silt/
marl
Bank material sand sand gravel/
cobble/
boulder
clay gravel/
cobble/
clay
sand/silt
Riparian vegetation sparse
poplar
sparse
poplar
sparse
poplar,
willow
sparse alder,
willow, ash
dense
beech,
oak, alder
sparse
willow
Section sinuosity straight straight straight meander straight bend
Peculiarity of section deeply
entrenched
deeply
entrenched
riffle meander near-natural bend
Bank line riprap
(bank line length %)
50 10 0 0 0 20 30 30 0 0 0
LWD characteristicsb
Date of inputc
1998 1995 01/1999 03/1998 not known 1997
Diameter at breast
height (cm)
75/65/65 50/75 115 40/60 125 75
Horizontal
orientation (j)d
100/115/115 85/0 75 90/65 95 0
Vertical orientatione
ramp/ramp/on
bed
ramp/bank on bed ramp/ramp above bed on bed
Individual tree
volume (m3
)f
7.19/4.14/3.55 2.13/2.44 5.4 1.1/5.14 12.34 9.68
Length/channel width >1 >1 f 0.7 >1 >1 f 0.3
Blockage ratiog
0.48 0.13 0.21 0.36 0.34 0.05
Root-wad anchored
in bank
no/yes/yes yes/no no yes/yes no cabled
LWD input natural natural natural natural natural restoration
project
Discharge (m3
/s)
Mean annual discharge
Mean low flow 0.368 0.368 0.458 0.847 0.472 7.09
Mean discharge 2.83 2.83 4.78 4.22 3.39 23.98
Mean high flow 31.4 31.4 74.2 42.1 46.0 110.4
Peak flow 59 59 149 96 112 178
Discharge since
LWD inputh
Mean low flow 0.471 0.347 mean
discharge
data not
available
0.782 date of LWD
input not known
6.84
J. Kail / Geomorphology 51 (2003) 207­223210
3.2. Field investigations
Topographic data were acquired in July/August
2000 using a Leica TCRA1103 electronic total station.
Some pools were not wadable and were mapped using
a small boat.
A preliminary investigation was carried out with the
aim to determine topographical survey point densities
necessary for an accurate description of mesoscale
stream morphology. The original data set of the Ahr
LWD stream section with a point density of 3.1 points/
m2
(which is assumed to represent mesoscale morTable
1 (continued)
Study streams Berkel1 Berkel2 Ahr Mo¨hne Berg. Landa
Lippe
RS LWD RS LWD RS LWD RS LWD RS LWD LWD
Discharge since
LWD inputh
Mean discharge 3.80 3.10 4.31 24.08
Mean high flow 43.2 32.7 39.6 113.2
Peak flow 48 48 51 178
Peak flow recurrence
interval (year)
10 10 2 5 10
a
The author is under a legal obligation not to exactly name the stream investigated.
b
LWD characteristics are given for individual trees (several values) or for entire LWD (single value).
c
Date of LWD input is assessed by means of a consultation of local stream managers, stream ecologists and residents.
d
Horizontal orientation according to Robison and Beschta (1990a) with 0j the root-wad pointing upstream, 90j perpendicular to flow and
180j the root-wad pointing downstream.
e
Vertical orientation was classified: ramp position, root-wad inside the channel and the other end supported on the opposite bank- on bed,
resting on the stream bed- bank, LWD lies between top of bank and mean water level parallel to flow-above bed, inside the bank-full
channel but completely above low-flow water level.
f
Tree volume inside bank-full channel.
g
Blockage ratio is the cross-sectional area blocked by LWD according to Gippel et al. (1996).
h
Discharge data since LWD input are not available for the year 2000 at the Berkel and Mo¨hne study streams.
Fig. 2. Downstream view at the Berkel1 LWD section at mean flow. Two large fallen trees in the foreground are mainly submerged at mean
flow; tree with root-wad in the background is located above mean flow water level.
J. Kail / Geomorphology 51 (2003) 207­223 211
phology) was progressively thinned to a density of 0.5
points/m2
. Terrain models were computed for each
data set and compared to the original surface (Fig. 3).
No clear limit was noticeable, but errors increase
rapidly when point densities are < 1 points/m2
. Therefore,
survey points were measured at a distance of
f 1/50 channel width ( f 0.3 m) in cross-sections
with a spacing of about 1/15 channel width (maximum
spacing < 1 m) to ensure point densities >1 points/m2
.
Topographic breaks in slope of particular geomorphological
importance (e.g., bank top, cutbanks, extent of
pools) were measured separately.
In some zones, measurement was not possible
(e.g., areas covered by dense vegetation, debris accumulations
or wood accumulations). Therefore, point
density varies slightly between LWD and reference
sections of most study streams. Differences in point
densities between LWD sections and corresponding
reference sections are less than 10%, except the
Mo¨hne stream (29%). In the Mo¨hne LWD section,
dense overhanging limbs of riparian trees that could
not be cut partly covered the stream and hindered
measurement of a larger number of survey points,
resulting in a comparatively low point density. Point
densities range from 1.4 to 3.1 points/m2
depending
on the cross-section spacing.
The circumference of the large fallen trees was
measured at several points of the stem and the main
limbs using a measuring tape. Approximate date of
input and input mechanism (natural, restoration project)
were determined by means of a consultation of
local stream managers, stream ecologist, and residents.
Water surface slope was determined by hydrostatic
levelling. Due to the afflux caused by the large fallen
trees, slope at some LWD sections is high compared
to the corresponding reference sections. Therefore,
channel sections both upstream and downstream of
the sections investigated were included in water level
measurements, in order to describe mean channel
slope rather than the drop in water level associated
with the large fallen trees.
3.3. Terrain models
Three-dimensional terrain models were computed
from the field data using the GIS ``ArcView 3DAnalysť'
to describe the stream morphology of
LWD sections and reference sections (Fig. 4). Surfaces
were created as triangulated irregular networks
(TINs) following Lane et al. (1994) and Milne and
Sear (1997) using the topographic breaks measured
separately in the field.
The extent of pools and bars was determined
according to Beebe (1997). Parts of the stream bed
at least one standard deviation below the mean depth
are defined as pools. Conversely, parts of the stream
bed at least one standard deviation above the mean
depth of the streambed are defined as bars. Pool and
bar volume was computed for single morphological
features using the ArcView 3D-Analyst tool ``Area
Fig. 3. Terrain model surface error (per unit stream area) against survey point density.
J. Kail / Geomorphology 51 (2003) 207­223212
and Volume Statistics''. Pools and bars were classified
according to Church (1992).
Nonvegetated areas of the bank steeper than 65j
and above the mean water level were defined as
cutbanks. The outlines of these cutbanks usually correspond
to topographic breaks measured in the field.
Because the length of the bank is highly dependent
on the scale considered (Andrle, 1994), these linear
features were standardized to the accuracy of the field
data, which is the same at all stream sections (point
density about 0.3 m at bank-top break lines).
3.4. Cross-sections
The terrain models indicated cross-sections measured
in the field were not exactly perpendicular to the
channel, especially in zones of high structural diversity
caused by LWD where the channel direction could not
be accurately determined in the field. Because some
channel-related parameters (e.g., cross-section width)
depend on the exact perpendicular orientation of the
cross-sections, they were not based on measuring
points but derived from the terrain models with a
spacing of 1 m (corresponding to accuracy of field
data, 16­53 cross-sections per stream section). For this
purpose, the ArcView extension ``Profile Extractor''
was used. In addition, the following parameters were
calculated: area (cross-sectional area), horizontal
Fig. 4. Contour map of Berkel1 LWD section three-dimensional model. Height above/below mean streambed height is given in meters.
Streambed area one standard deviation above/below mean depth of streambed (0.5 m) is defined as bar and pool, respectively. Pool is located
downstream of the first large fallen tree, which is located on the streambed. Sketch of the large fallen trees is not to scale; flow is from left to right.
Fig. 5. Measurement of cross-section angle for AMT-Analysis.
J. Kail / Geomorphology 51 (2003) 207­223 213
Table 2
Qualitative and quantitative description of morphological features present at LWD sections (LWD) and reference sections (RS); absolute size of
morphological features and volume of pools/bars related to bed planimetric area and area of cutbanks related to section length are given; based
on the close proximity to the large fallen trees, their shape and visually observed flow patterns, some morphological features are classified as
`clearly associated with LWĎ
Stream section Morphological
feature
Volume
(m3
)
Planimetric
area (m2
)
Maximum
depth/height
(cm)
Surface
area (m2
)
Pool/bar
volume
(m3
/ha)
Cutbank area
(m2
/100 m)
Clearly
associated
with LWD
Berkel1 LWD pool 16.8 51.4 75 593.5 +
side bar f 2 f 21 f 22 53.0
cutbanks 20.7 86.6 +
13.4 56.1 +
3.3 13.8 +
Berkel1 RS pools 0.4 5.3 20 20.1
0.3 6.4 17 15.1
0.3 12.1 11 15.1
0.06 3.5 3 3.0
0.03 1.3 9 1.5
0.006 0.8 1 0.3
side bar f 0.5 f 12 f 31 20.1
Berkel2 LWD pools 7.2 26.6 60 247.6 +
5.6 19.5 89 192.6 +
mid-channel bar 3.1 16.9 41 106.6 +
cutbank 30.7 107.7 +
Berkel2 RS pools 1.1 23.3 7 41.1
0.4 10.4 11 14.9
0.2 7.2 2 7.5
mid-channel bar 0.1 4.8 8 3.7
Ahr LWD pool 35.9 163.9 54 423.8 +
mid-channel bars 2.9 28.3 23 34.2 +
2.6 19.5 66 30.7 +
1.4 19.1 29 16.5 +
0.2 6.1 10 2.4 +
0.04 1.8 6 0.5
Ahr RS pool 15.9 116.7 56 226.1
side bar f 4 f 24 f 38 55.5
Mo¨hne LWD pools 59.6 143.1 98 692.7
0.5 11.4 16 5.8
point bar f 49 f 154 f 87 568.3
cutbank 94.8 171.4
Mo¨hne RS pools 20.2 97.3 53 382.8
4.5 21.3 59 85.3
point bar f 12 f 43 f 76 223.6
cutbank 67.6 177.4
Berg. Land LWD pool 11.8 37.3 70 516.2 +
mid-channel bar 0.2 3.6 12 8.7
side bar f 3 f 8 f 104 109.4 +
island 7.7 31.7 44 336.9
Berg. Land RS pool 8.0 37.3 65 321.1
mid-channel bar 0.08 5.6 4 3.2
island 9.6 25.7 60 385.3
Lippe LWD pools 9.3 96.5 38 58.0
3.5 44.2 25 21.8
1.7 21.1 21 10.6 +
0.5 16.5 10 3.1
0.3 11.2 7 1.9
J. Kail / Geomorphology 51 (2003) 207­223214
length (width), and maximum depth. Cross-section
depth values, calculated with a spacing of 0.1 m by
the extension ``Profile Extractor'', were used to calculate
mean depth. Channel dimensions were determined
for the cross-sections at bank-full stage, which can be
defined as the point where a break in the slope of the
banks occurs and water begins to flow onto the floodplain
(Wolman and Leopold, 1957).
3.5. AMT analysis
To describe the complexity of cross-sections at
different spatial scales, the angle-measurement-technique
(AMT) was used following Andrle (1994) and
Nestler and Sutton (2000). An Avenue script was
written to perform AMT analysis in ArcView using
the cross-sectional data computed by the extension
``Profile Extractor''. A starting point A along the
cross-section is randomly chosen. The point of intersection
B between a line of length S beginning at point
A and the cross-section is calculated. This process is
repeated beginning at point B. The angle between the
two lines is calculated (Fig. 5). For each scale S, a
sample of 500 angles is stored, which was found to be
sufficient to produce minimal error while still keeping
computational time to a reasonable level (Andrle,
1994). The mean angle describes the extent to which
the cross-section deviates from a straight line at the
given scale S. More complex cross-sections, therefore,
have greater mean angles.
Because mean angle increases markedly with the
entrenchment of the channel, AMT analysis was
restricted to the streambed. Otherwise, differences in
entrenchment of the streams would mask differences
in streambed morphology. The influence of the large
fallen trees on channel entrenchment was not investigated.
Only values of S greater than the accuracy of
the field data (survey point spacing in cross-sections
of about 0.3 m) were used for analysis.
4. Results
4.1. Terrain models
Large pools are associated with the large fallen
trees in all LWD sections except the Lippe (Table 2).
Median pool volume (m3
/ha) is higher in the LWD
sections than in reference sections (Mann­Whitney
U-test, p < 0.01, n = 29). Differences are greatest in the
LWD sections of the lowland streams Berkel2 and
Berkel1, which have pool volume of 7 to 11 times that
of the reference sections. Several small pools are
present in the reference sections, and one to two large
pools in the LWD sections. In spite of the later date of
LWD input, pool volume is markedly higher in the
Berkel1 LWD section, which is possibly due to the
higher blockage ratio (Table 1).
Differences are less apparent in the mountain
stream sections (1.5 to 2 times the pool volume of
reference sections), where bend scour pools (Mo¨hne),
trench pools (Berg. Land), and a deep thalweg (Ahr)
are present in the reference sections.
Volume of sidebars and point bars is largely
dependent on the delineation of riverbed and banks.
Small differences in the extent of riverbed and banks
result in large differences in bar volume. Therefore,
only clearly identified bars (mid-channel bars) are
considered. Mid-channel bars that are discernible in
the terrain models are restricted to LWD sections of
the Berkel2, Ahr, and Lippe. The mid-channel bar
volume is 29 times higher in the Berkel2 LWD section
Table 2 (continued)
Stream section Morphological
feature
Volume
(m3
)
Planimetric
area (m2
)
Maximum
depth/height
(cm)
Surface
area (m2
)
Pool/bar
volume
(m3
/ha)
Cutbank area
(m2
/100 m)
Clearly
associated
with LWD
Lippe LWD pools 0.1 4 5 0.6
0.08 6.1 5 0.5
0.04 3.4 3 0.2
0.02 2.1 3 0.1
point bar f 28 f 215 f 27 172.7
mid-channel bar 5.6 61.2 28 34.9 +
cutbank 26.1 59.0
J. Kail / Geomorphology 51 (2003) 207­223 215
compared to the corresponding reference section. In
the Ahr, several bars formed downstream of the large
fallen tree. The same is true for the zone between the
LWD and the outer bank at the Lippe. No mid-channel
bars (Berkel1, Mo¨hne) or bars of marginal extent
(Berg. Land) are present in the other stream sections.
The occurrence of cutbanks is restricted to LWD
sections of the lowland sand bed streams (Berkel1,
Berkel2, Lippe) and the outer bank of the stream
sections at the meandering lower mountain stream
(Mo¨hne). Cutbank area is nearly the same in the
Mo¨hne study sections indicating that the large fallen
trees at the LWD section did not increase cutbank
area.
Increase in streambed surface area compared to
planimetric area indicates topographic complexity of
the streambed. High values indicate a rough streambed
surface and, therefore, high form drag (Buffington
and Montgomery, 1999). All streams except the
Mo¨hne show an increased stream bed surface area
compared to planimetric area in the LWD sections
(Fig. 6). In the Mo¨hne, meander morphology leads to
a high value at the reference section. The differences
are most obvious in the lowland streams (Berkel1
four times, Berkel2 seven times higher than reference
sections) and the lower mountain stream Ahr (fourfold
higher).
Increase in bank line length compared to section
length is a measure of bank line complexity. Bank line
length is higher at all LWD sections compared to the
corresponding reference sections (Fig. 7) except for
the Mo¨hne stream section where a large curve in the
downstream part of the reference section lengthens the
bank considerably. By far the highest values are found
at the Berg. Land stream, which is probably due to the
near-natural condition of this reach. Differences
between LWD sections and reference sections are
due to the channel widening induced by LWD (lowland
stream Berkel1 and Berkel2, respectively, 2- and
1.6-fold higher in LWD section) and a curve in the
shoreline caused by the uprooting of the tree and
further bank erosion at high flow (lower mountain
stream Berg. Land, 1.7-fold higher than reference
section). The cause of the increase in bank length at
the Ahr LWD section (1.6-fold higher) is not apparent.
4.2. Cross-section parameters
Considering the quartiles of all values, a distinct
increase in variability of cross-sectional area was
Fig. 6. Increase in bed surface area compared to planimetric area (%). RS = reference section, LWD = LWD section.
J. Kail / Geomorphology 51 (2003) 207­223216
noted at the Berkel1 LWD section (Fig. 8A). Here,
widening of the channel caused by bank erosion and a
deep scour pool increased the area of single crosssections
dramatically. Differences are less pronounced
at the Berkel2, Ahr, and Berg. Land streams and are
marginal at the Mo¨hne. Median cross-sectional area is
greater at the lowland sand bed LWD stream sections
Berkel1, Berkel2, and at the lower mountain stream
Berg. Land (Mann­Whitney U-test, p < 0.01).
Differences in variability of cross-sectional area
between LWD sections and corresponding reference
sections at the lower mountain streams Ahr and Berg.
Land are due to the wide range of channel depth
values. In addition to channel depth, higher variability
of stream width at the LWD sections causes differences
in cross-sectional area variability at the lowland sand
bed LWD stream sections Berkel1 and Berkel2 compared
to corresponding reference reaches (Fig. 8B­D).
At the Mo¨hne stream, only the variability of stream
width is considerably higher at the Mo¨hne LWD
section compared to the reference section (Fig. 8B).
Median cross-section width is higher at all LWD
sections compared to the corresponding reference
sections, except the Berg. Land stream (Mann­Whitney
U-test, p < 0.01). Here, at the right side of the midchannel
bar where the reference section is located,
widening of the channel in the lower part increases
stream width.
Variability of maximum depth can be considered to
be a measure of thalweg complexity. Differences
between LWD sections and reference sections are
most striking at the lower mountain streams (Ahr
and Berg. Land), considerable at the Berkel stream,
and small at the Mo¨hne stream (Fig. 8C). This is also
true for the variability of mean depth, which indicates
that pools in some LWD sections cover a large part of
the cross-sections and are not restricted to a narrow
thalweg (Fig. 8D). Median cross-section depth is
greater at the lowland sand bed streams Berkel1,
Berkel2, and at the lower mountain stream Berg. Land
(Mann­Whitney U-test, p < 0.05).
The differences in variability of cross-section
parameters between LWD sections (sample A) and
reference sections (sample B) was tested using the
interquartile coefficient as a measure of dispersion.
Variability of cross-section area and cross-section maximum
depth is higher at the LWD sections (Mann­
Whitney U-test, p < 0.05, n = 10), whereas variability
of cross-section width, mean depth, and width/depth
ratio show no significant difference.
Variability and median of width/depth ratio do not
differ between LWD sections and corresponding referFig.
7. Increase in bank line length compared to section length (%). RS = reference section, LWD = LWD section.
J. Kail / Geomorphology 51 (2003) 207­223 217
ence sections, except the Ahr stream, in spite of an
evident change in channel morphology. Low variability
of width/depth ratio at the LWD sections is probably
due to the simultaneous increase in width and depth
because of bank and bed erosion. Hence, width/depth
ratio is considered to be an inappropriate measure to
describe the effects of LWD on channel morphology at
the stream sections investigated in this study.
4.3. AMT analysis
Median angle of each spatial scale S was calculated
based on cross-sectional mean angle data for each
stream section (Fig. 9). The median angle of LWD
sections is significantly greater compared to the reference
sections at all scales, with the exception of the
30-, 40-, 70-, and 80-cm scales (Mann­Whitney Utest,
p < 0.05).
Stream sections may be roughly grouped into three
categories (Fig. 9). Group A consists of Ahr, Berkel1,
and Berkel2 reference sections and the Lippe LWD
section, which all have a relatively flat stream bed.
Median angle is low (2­4j) and decreases slightly at
larger scales. Group B consists of the Berg. Land
reference section and shows an increase in median
angle up to almost 6j at the scale of 80 cm and a
slight decrease at larger scales. Group C consists of
the Mo¨hne LWD section and reference section as well
as the Berg. Land, Ahr, Berkel2, and Berkel1 LWD
sections.
Median angle increases with scale at the Mo¨hne
stream sections, whereas values seem to approximate
to angles ranging from about 7­9j at the other study
sections at larger scales. Increase of median angle at
lower scales is highest at Berkel1 and Berkel2 LWD
sections. Differences between these two curves at
Fig. 8. Variability and median of cross-section parameters (A: area; B: width; C: maximum depth; D: mean depth) at reference sections (RS) and
LWD sections (LWD). Min­Max, 25­75%, and median are shown; n = number of cross-sections investigated, * = significant differences
between LWD section and corresponding references section (Mann­Whitney U-test, p < 0.01), + = significant differences between LWD
section and corresponding reference section (Mann­Whitney U-test, p < 0.05).
J. Kail / Geomorphology 51 (2003) 207­223218
larger scales are probably due to the larger and deeper
pool at the Berkel1 LWD section.
In comparing LWD sections with the corresponding
reference sections, greater median angles at the
LWD sections are statistically significant for the
Berkel1, Berkel2, Ahr, and Mo¨hne stream sections
at all scales. Differences in the near-natural Berg.
Land stream sections are statistically significant only
for the largest scales investigated (280, 290, and 300
cm), where median angle is low at the reference
section (Mann­Whitney U-test, p < 0.05).
This is probably due to the overall form of the
pools, with a circular pool at the LWD section and a
narrow trench pool at the reference section (short axis
of the trench pool parallel to cross-sections). Therefore,
lines AB/BC of AMT analysis can span the
entire pool at larger scales, resulting in lower mean
angles at the reference section.
4.4. Relationship between stream morphology, LWD,
and stream characteristics
The relationship between stream morphology (pool
volume, pool area, mid-channel bar volume, midchannel
bar area, cutbank area, related to bed planimetric
area; maximum pool depth, bed and bank line
complexity, median angle of AMT analysis over all
scales S), LWD characteristics (blockage ratio, channel
volume blocked by LWD volume, horizontal
orientation, vertical angle, mean height above bed,
diameter at breast height, time since LWD input), and
stream characteristics (slope, catchment area, width,
power per unit width) was assessed by Spearman
correlation analysis because of the nonnormal distribution
and low number of cases (n = 6).
Pool volume of LWD sections is strongly correlated
to the blockage ratio (rs = 0.93, p < 0.01), which
indicates that blockage ratio is one important parameter
determining the hydraulic and, therefore, morphological
influence of large fallen trees, as stated by
Gippel et al. (1996). Not surprisingly, channel width is
strongly correlated to catchment area (rs = 0.98, p <
0.01). No other correlations were found to be significant
for the variables examined in this study.
5. Discussion
5.1. Power of parameters to describe change in
channel morphology
Differences between LWD sections and reference
sections were described using a wide range of parameters,
derived from both terrain models and crossFig.
9. Median of mean angle of cross sections at scales S ranging from 30 to 300 cm. Study sections in legend are listed according to the mean
angle over all scales S. RS = reference section, LWD = LWD section.
J. Kail / Geomorphology 51 (2003) 207­223 219
sections (extent of morphological features; bed and
bank complexity; cross-sectional area, width, maximum
depth, mean depth; AMT analysis; see Section
4). However, only differences in distinct morphological
features such as pools, bars, and cutbanks could
be detected. This is due to the experimental design of
comparing LWD sections with reference sections.
Long-term studies (e.g., survey of several years) are
necessary to examine less evident morphological
features, such as large, flat, depositional areas.
5.2. LWD characteristics
It can be assumed that nearly all trees investigated
are located where they entered the channel and
changed little in position because either (1) they are
still anchored in the bank with their root-wad, (2) the
cutbank caused by the uprooting of the tree remains in
close vicinity to the tree, or (3) the position of the tree
is known because it was placed in the stream within the
scope of a restoration project or was observed by local
stream managers, ecologists, or residents. Moreover,
most trees that entered the channel naturally are
oriented nearly perpendicular to flow (deviation from
perpendicular F 25j), which indicates that trees did
not rotate at high flows (e.g., Q2 to Q10 floods which
have been recorded since the wood entrance). In
addition, remapping of the tagged points on trees in
2001 revealed no change in position. Only one tree can
be considered to be driftwood (Berkel1, tree oriented
parallel to flow). Therefore, it can be inferred that the
impact of LWD on channel morphology changed little
over time and LWD characteristics listed in Table 1
represent LWD conditions that influenced channel
morphology since LWD input. However, wood and
debris accumulations trapped by the trees could have
formed and disappeared or changed during floods, and
have transiently increased or changed blockage ratio
and the impact on channel morphology.
Besides anchoring of root-wads in stream banks,
stability of trees investigated that naturally entered the
channel is enhanced by length of trees compared to
channel width which is greater than or equal to twothirds
of the channel width. Flume studies of Braudrick
and Grant (2000, 2001) showed that presence of rootwads,
length, and diameter of trees increase the stability
of logs. Bryant (1983) and Lienkaemper and
Swanson (1987) observed that trees considerably longer
than channel width result in relatively stable wood
pieces. Gurnell and Gregory (1995) also observed that
deciduous trees, which fall into the channel, are often
anchored in the bank by their root-wad.
LWD mass in European streams is low compared
to wood loading in North America, but we can expect
that it could be comparable in reaches where the
human impact is reducing (Elosegi et al., 1999;
Piégay et al., 1999; Hering et al., 2000; Diez et al.,
2001). In central European streams similar to those
investigated in this study, large fallen trees of comparable
size are extremely rare (Hering et al., 2000).
The main reasons for the low LWD loading in the
streams investigated are sparsely vegetated banks
(Fig. 2) and the removal of LWD by stream managers.
Even in nature reserves stream managers are under a
legal obligation to remove LWD if it is considered to
be a flood risk to works downstream. Due to changes
in EG agricultural policy and nature conservation
laws, extensive farming on floodplain areas becomes
more common. Therefore, in some exceptional cases,
large fallen trees are left in the channel. Considering
the impact of the large fallen trees investigated on
channel morphology, it can be assumed that channel
morphology of these streams is far from that which
characterizes the potential natural state.
5.3. Comparing observed scour patterns with those
described in literature
The pool at the Ahr LWD section is located
directly upstream and to the side of the large fallen
tree that lies perpendicular to flow in the middle of the
channel. Mid-channel bars consisting of fine gravel
accumulated downstream of the tree. This scour
pattern is very similar to those described by Abbe
and Montgomery (1996) for LWD jams at the apex of
bars in a large alluvial river. This is possibly a typical
scour pattern at LWD obstructions located on the
streambed in the middle of the channel, either nearly
perpendicular or parallel to flow, if peak flows do not
overtop the obstruction.
Cherry and Beschta (1989) and Hilderbrand et al.
(1998) observed that different scour patterns depend
on angle to flow and vertical angle of logs. Scour at
the Berkel1 LWD section occurs downstream of one
of the large fallen trees, which is oriented perpendicular
to flow (Fig. 4). This scour pattern can be
J. Kail / Geomorphology 51 (2003) 207­223220
classified as a plunge pool according to terminology
of Robison and Beschta (1990b) and corresponds to
the scour pattern described by Hilderbrand et al.
(1998) as perpendicular dam. Moreover, the pool in
the Berg. Land LWD section can be described as an
underflow pool (Robison and Beschta, 1990b). No
other scour pattern associated with the single, large
fallen trees investigated in this study corresponds to
those described by the authors mentioned above.
5.4. Assessing the morphological influence of single,
large fallen trees investigated
Pool volume of the LWD sections investigated is
well within the upper range of pool volume found in
some small, high-gradient streams in Oregon, NW
America, where pool volume ranged from 229 to 755
m3
/ha (Carlson et al., 1990). Single, large fallen trees
can, therefore, be considered to be capable of increasing
pool volume locally to an extent comparable to
North American conditions even in low-gradient central
European streams.
Differences between LWD sections and reference
sections are most striking in the lowland sand bed
Berkel stream. This is true not only for bed morphology
(e.g., pool volume, bed complexity) but also for stream
bank morphology (cutbank area, bank complexity,
variability of cross-section width) and cross-section
complexity (AMT analysis). Some rare habitat types
(e.g., deep pools, which are used as rearing habitat for
certain fish species; Fausch and Northcote, 1992;
Spalding et al., 1995; Young, 1996) are restricted to
the immediate vicinity of large fallen trees.
In the lower mountain streams, morphological
channel changes caused by large fallen trees are
pronounced, but less evident on the stream banks.
This is probably due to resistance of bank material and
low entrenchment (e.g., cutbank area increases with
channel entrenchment). Meander morphology and
local geomorphic controls such as local geology are
likely to mask the influence of LWD on channel form,
as suggested by Evans et al. (1993) and Hilderbrand et
al. (1997).
The effect of the single, large fallen tree on channel
morphology at the Lippe (by far the largest study
stream) is low compared to those on the other study
streams. This is probably due to low blockage ratio
(0.5%), which depends on stream size and the orientation
of the log parallel to the banks. Nevertheless,
two distinct morphological features (small pool, small
mid-channel bar) are clearly associated with the large
tree in the Lippe channel (Table 2).
Considering the extent of morphological features at
LWD sections compared to reference sections, the
effect of the large fallen trees on channel morphology
is evident in most study streams (Table 2, Figs. 6­9).
Although sample size of paired sections is small
(n = 5) and reference sections vary in structural diversity,
the differences between LWD sections and reference
sections are statistically significant for some
parameters (pool volume, median angle of AMT
analysis).
However, morphological changes were not observed
directly. Therefore, differences in structural
diversity between LWD sections and reference sections
could partly be caused by morphological differences
that existed prior to LWD input. Comparability
of LWD sections to reference sections is limited by
differences in bank line riprap (Berkel1, Ahr) and
slope (Ahr, Berg. Land). Bank line riprap at the
Berkel1 and Ahr reference sections consists of loose
boulders and building rubble, which possibly hinders
lateral erosion. However, lateral erosion does not
occur at the Berkel2 reference section, which is free
of bank line riprap and comparable to the Berkel1
stream sections.
Five of the study streams were remapped in 2001.
Provisional results show that considerable changes in
channel morphology occurred in all LWD sections,
except for the Berg. Land stream, indicating that
channels are still adjusting to the presence of the large
fallen trees. Some of the channel features that were
present at the first mapping period in 2000 developed
(e.g., pools got deeper, sidebars expanded), but others
diminished (e.g., pools filled, mid-channel bars
eroded). Although channel morphology before the
first mapping in 2000 and prior to LWD input is not
known, and morphological changes observed over a
1-year period may not be representative in the longer
term, it is hypothesized that there is no clear trend
towards an equilibrium state of channel morphology.
Dynamic feedback between flow produced by the
large fallen trees and channel morphology may result
in changing trends of morphological development.
Because changes in channel morphology are highly
dynamic (as remapping in 2001 suggests), and morJ.
Kail / Geomorphology 51 (2003) 207­223 221
phological differences described are strongly dependant
on channel conditions (channel morphology, discharge,
sediment supply) and LWD characteristics,
transferability of the results is limited. However, given
similar channel conditions and LWD characteristics,
differences between LWD sections and stream sections
free of large wood of the same order of magnitude
are to be expected in central European rivers.
Acknowledgements
This study has been supported by the German
Environment Foundation (Deutsche Bundesstiftung
Umwelt). I gratefully acknowledge the review of this
paper by J.M. Buffington, R.A. Marston, D.R.
Montgomery, H. Piégay, F.J. Swanson, D. Hering,
and the technical support by Leica Geosystems. I
thank S. Hertrich and C. Bremser for field assistance.
References
Abbe, T.B., Montgomery, D.R., 1996. Large woody debris jams,
channel hydraulics and habitat formation in large rivers. Regulated
Rivers: Research and Management 12, 201­221.
Andrle, R., 1994. The angle measurement technique: a new method
for characterizing the complexity of geomorphic lines. Mathematical
Geology 26 (1), 83­97.
Beebe, J.T., 1997. Fluid patterns, sediment pathway and wood obstructions
in the Pine River, Agnus, Ontario. PhD Dissertation,
Department of Geography and Environmental Studies, Wilfrid
Laurier University, Ontario, Canada.
Beschta, R.L., 1979. Debris removal and its effects on sedimentation
in an Oregon Coast Range stream. Northwest Science 53
(1), 71­77.
Bilby, R.E., 1981. Role of organic debris dams in regulating the
export of dissolved and particulate matter from a forested watershed.
Ecology 62 (5), 1234­1241.
Braudrick, C.A., Grant, G.E., 2000. When do logs move in rivers?
Water Resources Research 36 (2), 571­583.
Braudrick, C.A., Grant, G.E., 2001. Transport and deposition of
large woody debris in streams: a flume experiment. Geomorphology
41, 263­283.
Bryant, M.D., 1983. The role and management of woody debris in
West Coast salmonid nursery streams. North American Journal
of Fisheries Management 3, 322­330.
Buffington, J.M., Montgomery, D.R., 1999. Effects of hydraulic
roughness on surface textures of gravel-bed rivers. Water Resources
Research 35 (11), 3507­3521.
Carlson, J.Y., Andrus, C.W., Froehlich, H.A., 1990. Woody debris,
channel features, and macroinvertebrates of streams with
logged and undisturbed riparian timber in northeastern Oregon,
USA. Canadian Journal of Fisheries and Aquatic Sciences 47,
1103­1111.
Cherry, J., Beschta, R.L., 1989. Coarse woody debris and channel
morphology: a flume study. Water Resources Bulletin 25 (5),
1031­1036.
Church, M., 1992. Channel morphology and typology. In: Calow,
P., Petts, G.E. (Eds.), Rivers Handbook--Hydrological and
Ecological Principles. Blackwell, Oxford, UK, pp. 126­143.
Diez, J.R., Elosegi, A., Pozo, J., 2001. Woody debris in North
Iberian streams: influence of geomorphology, vegetation, and
management. Environmental Management 28 (5), 687­698.
Elosegi, A., Diez, J.R., Pozo, J., 1999. Abundance, characteristics,
and movement of woody debris in four Basque streams. Archiv
fur Hydrobiologie 144 (4), 455­471.
Evans, B.F., Townsend, C.R., Crowl, T.A., 1993. Distribution and
abundance of coarse woody debris in some southern New Zealand
streams from contrasting forest catchments. New Zealand
Journal of Marine and Freshwater Research 27, 227­239.
Fausch, K.D., Northcote, T.G., 1992. Large woody debris and salmonid
habitat in a small coastal British Columbia stream. Canadian
Journal of Fisheries and Aquatic Sciences 49, 682­693.
Gerhard, M., Reich, M., 2000. Restoration of streams with large
wood: effects of accumulated and built-in wood on channel
morphology, habitat diversity and aquatic fauna. International
Review of Hydrobiology 85 (1), 123­137.
Gippel, C.J., O'Neill, I.C., Finlayson, B.L., Schnatz, I., 1996. Hydraulic
guidelines for the re-introduction and management of
large woody debris in lowland rivers. Regulated Rivers: Research
and Management 12, 223­236.
Gregory, K.J., Gurnell, A.M., Hill, C.T., 1985. The permanence of
debris dams related to river channel processes. Hydrological
Sciences Journal 30 (3), 371­381.
Gurnell, A.M., Gregory, K.J., 1995. Interactions between semi-natural
vegetation and hydrogeomorphological processes. Geomorphology
13, 49­69.
Hering, D., Kail, J., Eckert, S., Gerhard, M., Meyer, E.I., Mutz, M.,
Reich, M., Weiss, I., 2000. Coarse woody debris quantity and
distribution in central European streams. International Review
of Hydrobiology 85 (1), 5­23.
Hilderbrand, R.H., Lemly, A.D., Dolloff, C.A., Harpster, K.L.,
1997. Effects of large woody debris placement on stream channels
and benthic macroinvertebrates. Canadian Journal of Fisheries
and Aquatic Sciences 54, 931­939.
Hilderbrand, R.H., Lemly, A.D., Dolloff, C.A., Harpster, K.H.,
1998. Design considerations for large woody debris placement
in stream enhancement projects. North American Journal of
Fisheries Management 18 (1), 161­167.
Hogan, D.L., 1987. The influence of large organic debris on channel
recovery in the Queen Charlotte Islands, British Columbia,
Canada. Proceedings of the Corvallis Symposium Erosion and
Sedimentation in the Pacific Rim. IAHS Publ. No. 165, Corvallis,
OR, pp. 343­353.
Keller, E.A., Swanson, F.J., 1979. Effects of large organic material
on channel form and fluvial processes. Earth Surface Processes
4, 361­380.
Keller, E.A., Tally, T., 1979. Effects of large organic debris on
J. Kail / Geomorphology 51 (2003) 207­223222
channel form and fluvial processes in the coastal redwood environment:
adjustments of the fluvial system. In: Rhodes, D.D.,
Williams, G.P. (Eds.), Proceedings Tenth Annual Geomorphology
Symposium, Dubuque, IA. Kendall/Hunt Pub., Dubuque,
IA, pp. 169­196.
Klein, R., Sonnevil, R., Short, D., 1987. Effects of woody debris
removal on sediment storage in a northwest California stream.
Proceedings of the Corvallis Symposium, Erosion and Sedimentation
in the Pacific Rim. IAHS Publ. No. 165, Corvallis, OR,
pp. 403­404.
Lane, S.N., Chandler, J.H., Richards, K.S., 1994. Developments in
monitoring and modelling small-scale river bed topography.
Earth Surface Processes and Landforms 19, 349­368.
Lienkaemper, G.W., Swanson, F.J., 1987. Dynamics of large woody
debris in old-growth Douglas-fir forests. Canadian Journal of
Forest Research 17, 150­156.
MacDonald, A., Keller, E.A., 1987. Stream channel response to the
removal of large woody debris, Larry Damm Creek, northwestern
California. Proceedings of the Corvallis Symposium, Erosion
and Sedimentation in the Pacific Rim. IAHS Publ. No. 165,
Corvallis, OR, pp. 405­406.
Milne, J.A., Sear, D.A., 1997. Modelling river channel topography
using GIS. International Journal of Geographical Information
Science 11 (5), 499­519.
Mosley, M.P., 1981. The influence of organic debris on channel
morphology and bedload transport in a New Zealand forest
stream. Earth Surface Processes and Landforms 6, 571­579.
Murphy, M.L., Heifetz, J., Johnson, S.W., Koski, K.V., Thedinga,
J.F., 1986. Effects of clear-cut logging with and without buffer
strips on juvenile salmonids in Alaskan streams. Canadian Journal
of Fisheries and Aquatic Sciences 43, 1521­1533.
Nakamura, R., Swanson, F.J., 1993. Effects of coarse woody debris
on morphology and sediment storage of a mountain stream
system in western Oregon. Earth Surface Processes and Landforms
18, 43­61.
Nestler, J., Sutton, V.K., 2000. Describing scales of features in river
channels using fractal geometry concepts. Regulated Rivers:
Research and Management 16, 1­22.
Piégay, H., Citterio, A., Astrade, L., 1998. Ligne de débris ligneux
et recoupement de méandres, exemple du site de Mollon sur la
rivie`re ďAin (France). Zeitschrift fu¨r Geomorphologie 19,
187­208.
Piégay, H., Thévenet, A., Citterio, A., 1999. Input, storage and
distribution of large woody debris along a mountain river continuum,
the Drôme river, France. Catena 35, 19­39.
Robison, E.G., Beschta, R.L., 1990a. Characteristics of coarse
woody debris for several coastal streams of southeast Alaska,
USA. Canadian Journal of Fisheries and Aquatic Sciences 47,
1684­1693.
Robison, E.G., Beschta, R.L., 1990b. Coarse woody debris and
channel morphology interactions for undisturbed streams in
southeast Alaska, USA. Earth Surface Processes and Landforms
15, 149­156.
Smith, R.D., Sidle, R.C., Porter, P.E., 1993. Effects on bedload
transport of experimental removal of woody debris from a forest
gravel-bed stream. Earth Surface Processes and Landforms 18,
455­468.
Spalding, S., Peterson, N.P., Quinn, T.P., 1995. Summer distribution,
survival and growth of juvenile coho salmon under varying
experimental conditions of brushy instream cover. Transactions
of the American Fisheries Society 124 (1), 124­130.
Trimble, S.W., 1997. Stream channel erosion and change resulting
from riparian forests. Geology 25 (5), 467­469.
Wolman, M.G., Leopold, L.B., 1957. River flood plains: some observations
on their formation. US Geological Survey Professional
Paper 282-C, 107 pp.
Young, M.K., 1996. Summer movements and habitat use by Colorado
River cutthroat trout (Oncorhynchus clarki pleuriticus) in
small, montane streams. Canadian Journal of Fisheries and
Aquatic Science 53 (6), 1403­1408.
J. Kail / Geomorphology 51 (2003) 207­223 223