9500

• UCREFRP Catalog Number: 9500
• Author: Pitlick, J. and R. Cress
• Title: Downstream Changes in the Channel Geometry of a Large Gravel Bed River
• USFW Year: 2002
• USFW - Doc Type: American Geophysical Union
• Copyright Material: YES

34-4 PITLICK AND CRESS: DOWNSTREAM CHANGES IN CHANNEL GEOMETRY
Table 2. Summary Information for Main Stem Gauging Stations on the Colorado River
USGS Drainage Mean Annual Annual Suspended
Location Number Area, kmz Discharge, m3/s Sediment Load, t/yr
Glenwood Springs 9072500 11,800 64 0.37 x 106
Cameo 9095500 20,800 111 1.53 x 10'
State Line 9163500 46,200 180 3.43 x 106
Cisco 9180500 62,400 200 5.43 x 106
"The station at Glenwood Springs is located upstream of the study area; locations of other stations are shown in
Figure 1. Annual discharges and sediment loads were estimated on the basis of flow records from 1934-1997, except
at the State Line gauge, where the records begin in 1952 (from Pillick and Cress [2000]).
Subsurface samples were taken by collecting 100-150 kg
of the sediment underneath the surface layer. These samples
were nearly always large enough to ensure that the coarsest
particle represented no more than 5% of the total weight.
The coarse fraction of this sediment was sieved in the field,
while the line Fraction was sieved in the laboratory. A total
of 78 surface samples and 27 subsurface samples were taken
in the 260-km study reach.
[12] The average channel slope of the 10 individual
reaches was measured with a global positioning system
(GPS) supplemented in a few places by measurements from
topographic maps. Readings of the water surface were taken
every 0.8-km with a mapping-grade GPS (Trimble Path-
finder Pro-XR). These data were subsequently corrected
with differential post-processing techniques, yielding verti-
cal positions with errors of f0.5-0.3 in. These errors tend to
be random and are small in comparison to the total drop in
elevation through most reaches (25-50 m). As a further
check on accuracy, we compared slopes derived from the
GPS measurements with those derived from topographic
snaps and found that they were essentially the same.
[13] The hypothesis stated in the introduction implies that
gravel bed rivers adjust their bank-full width B, depth H, and
slope S to transport bed load at shear stresses slightly above
the threshold for motion. If so, the width, depth, slope and
grain size should change simultaneously downstream to give
a constant bank-full dimensionless shear stress Tb. The
dimensionless shear stress, or Shields stress, is defined as
r* - T [(p, - p) o D]- 1, where T - p o H S is the bank-full
shear stress, h,. and p are the densities of sediment and water,
respectively, g is the gravitational acceleration, and D is the
grain size. We formulated individual values of Tb for each
cross section using the measured bank-full depth, the reach-
average slope, and the reach-average median grain size of
the surface sediment, D50. Reach-average values of S and
Dip were used largely for practical reasons: we could not
measure the surface grain size or slope at every cross section,
thus we combined measurements in each of the 10 reaches.
In our experience reach-average slopes do not differ greatly
from local slopes at bank-full flow [Pitlick and Van Steeter,
1998], and variations in grain size have as much to do with
the local topography and sedimentology of individual bars as
with systematic downstream fining or tributary inputs.
[14] Standard statistical tests were used to evaluate the
significance of regression relations between In transfonned
values of grain size, width, depth, T, and -rb, and distance
downstream. We examined trends within individual reaches
as well as overall trends. T test comparisons were used to
assess differences in local versus regional trends in grain size
and to assess the correlation between downstream changes in
grain size and shear stress. Analysis of covariance was used
to examine differences in channel geometry between alluvial
and quasi-alluvial reaches [Kleinbaum and Kupper, 1978].
Field Observations and Results
4.1. Channel Slope
[15] The longitudinal profile of this portion of the Colo-
rado River is made up of four concave segments separated by
three breaks in slope (Figure 3). The breaks in slope occur in
P
U
0
at

160
140 ..? ? ?? ?, ?, ? •
120
1 c u
i.
° a
16
100 LO
4,0?
! o
00
C., t - o
80
0.6
0.4
C
° 0.2
to
0.0
a
Gi L
10 -0.2
-0.4
Figure 2. Trends in (a) water surface width and (b)
average bed elevation at USGS gauging stations near
Cameo, Colorado (open circles), and near Cisco, Utah (solid
circles). Water surface widths were taken directly from
discharge notes; bed elevations were computed by taking
the difference between the observed gauge height and the
mean flow depth.
1984 1988 1992 1996 2000