Professional Documents
Culture Documents
IN M O U N T A I N RIVERS
By James C. Bathurst 1
Downloaded from ascelibrary.org by Illinois Inst of Technology on 06/24/13. Copyright ASCE. For personal use only; all rights reserved.
INTRODUCTION
625
FLOW CHARACTERISTICS
Bed slopes of mountain rivers are generally greater than 0.4% (see
Golubtsov, Ref. 15), but less than about 5%. At steeper slopes, channels
appear to develop a series of short pools and falls, a feature which is
not considered here (32). Bed material consists of gravel, cobbles and
boulders and the relative submergence, or ratio of depth, d, to sediment
size, D M , ranges from the region of large-scale roughness (d/DM < 1),
via intermediate-scale roughness (1 < d/Dm < 4), to the region of small-
scale roughness (d/D^ > 4) (6). Here d = mean flow depth and D ^ =
the size of median axis of the bed material which is larger than 84% of
the material. Significant sediment transport occurs only during large
floods, but may then have a considerable effect. Two-dimensional bed-
forms of the type observed in sand-bed channels are rare, although gravel
antidunes have been observed under exceptional conditions (7,32). On
a larger scale, pool/riffle series or alternating bars with a meandering
thalweg may develop (21). Froude numbers may attain high subcritical
or even supercritical values (22).
\fj A(gRSff2
in which Q = water discharge; A = flow cross-sectional area, g = ac-
celeration due to gravity; R = hydraulic radius; and Sf = friction slope.
Evaluation of / is potentially the biggest source of error in the slope-
area method since in mountain rivers variations in the value of /, of up
to 100%, can occur at a site as discharge and depth vary. Attempts to
predict these variations suffer from the lack of available resistance equa-
tions. Standard equations developed for rivers with gentler slopes and
finer bed material may not apply, especially in the region of large-scale
roughness (8,11,15). On the other hand those equations which have been
designed explicitly for mountain rivers tend to be either empirical and
valid only for high flows (1,15,33), or else theoretically based but cum-
bersome, based on restricted data and generally valid only for low and
moderate flows (3,6,22). While these offer insights into the flow resis-
tance processes, none is yet suitable for practical engineering use; and
it is clear that considerably more research into the hydraulics of moun-
tain rivers is needed. In view of this, the approach adopted here is to
fall back on and attempt to modify the semilogarithmic resistance equa-
tion derived from boundary layer theory and applied to gravel-bed rivers
of moderate slope. Several versions of this familiar relationship have been
626
DATA PREPARATION
628
Cross-
sectional Mean Relative Water
Downloaded from ascelibrary.org by Illinois Inst of Technology on 06/24/13. Copyright ASCE. For personal use only; all rights reserved.
629
All the data (based on mean reach values) are shown in Table 1.
Measurement Errors.—The most significant sources of error arose in
the measurement of water surface level and discharge. In the first case,
surface waves lapping against the surveying staff resulted in estimated
errors of up to ±5% in the depth values, and up to ±8% in water surface
slope. In the derivation of discharge values from the stage-discharge rat-
ing curves, difficulties were caused by the variations in discharge which
occurred during some of the surveys and by approximations inherent in
the rating curves. Time variations were dealt with by calculating a time-
averaged discharge, and in nearly all cases, the variations were less than
631
the Tromie sites, the errors may be larger since calculation of the higher
discharges involved extrapolation of the rating curve considerably be-
yond the maximum current-metered flow of 16.7 m 3 /s. Finally, the error
in measuring bed material size was estimated by the method of Hey and
Thorne (20) to be up to ±15%, while the error in measuring distance
was estimated to be up to ±1%.
From the foregoing, the range of errors in selected terms derived from
the basic quantities and appearing in the following review are by the
theory of errors ( 1 8 ) : - ( 8 / / ) ^ : ±12%; d/Dm :±16%; F : ± l l % in which F
= Froude number.
ANALYSIS
Hey (1979)
V _
©
- + Judd and Peterson 11969)
Downloaded from ascelibrary.org by Illinois Inst of Technology on 06/24/13. Copyright ASCE. For personal use only; all rights reserved.
' - / // / _
- -
- 4 ' » °' ' -
- -
- -
"~
.?#P* • -
— ''^y* Hey equation
1 A / 1 1 11 1 1 1 1 l i l 1l l l
0.1 0.5 1 5 20
Relative Submergence d/D„.
FIG. 2.—Variation of (8//)1/2 with Relative Submergence, d/D 84 , for Field Sites of
this and other Studies
1 1 I | II I I | I I | 1 lll|
u
„ 13
- SITES
Slope Site
A 0.005 Findhorn
In 12 - # 0.005 Tweed
-
T3 D 0.010 Ettrick (Brockhoperigl
fl1 - • 0.010 Logan River 32 (Judd
J /7
t
ii 10
9
- v 0,011
and Peterson 19691
Tromie 2 /[
,
1
1 s"'
/*''
2 8
o 0.012 Bear River dl 1 s^' /
(Virmani 19731 / } 1/ ' /
7 / s
c
6
//
5 / / In '/ /
/ //I '
/ // r '
§ s
I3 - 1
/ /*"
Hey equation
~
In -
« 2 , < y
\ 1 i i i i 1 i i 1 1 1 1 11
cr
1 0.5 1 5 20
Relative Submergence d/D„,
FIG. 3.—Variation of (8//)1/2 with Relative Submergence, d/DM, for Principal Field
Sites in Fig. 2
A Williams (19781 /A / /
^L 9
e e
/ ® /%/
e
oo 8 * /Z • -
/
/ a . DD» * °°
y/A
7
t
c
= 6 ' e./ x —
/ ©
/ A / /,
hu
— '
/
'
s\s
* 0/ / / a
a
#®
•
^
A
-
—
3 - f / As // —
I , /V ^ \ A
A
tfl
'« 2 ' /
'// Hey equo tion
en
1 Mill 1 I I I i i i i1 i i i 1 i i n
0 0.5 5 10 50
Relative Submergence d/0.„
FIG. 4.—Variation of (8//)1/2 with Relative Submergence, d/D 50 , for Field Sites
Thought to Have Nonuniform Flows, Using Data from other Studies
12 1 1 1 | 1 1 1 1 | 1 1 1 |. M I I
_ DATA SOURCE •'
U)
"O o CSU (Bathurst et al 1981 I
o>10
X • EPFL (Bathurst et al 1983) /
~
/ * / /
~L 8 66 percent / / £
%
| 7 - margin /,' % \y /
6 -
c
.2 5 -
u
§ 4
LL
3 -
o - 1>?*v
2
§ £/°——-Hey equation
™ 1
a>
FIG. 5.—Variation of (8//)1/2 with Relative Submergence, d/D M , for Flume Data
for river data with d/D90 s 5. However, in collecting their data, Charlton
et a l , specifically excluded regions of rapids (riffles) which implies that
their field sites were in pools. Thus the reason for Eq. 2 overpredicting
(8//) 1/2 is probably connected with the effect of a nonuniform bed profile
in which downstream controls such as bars maintain the depth at a value
higher than that which would be maintained by boundary resistance
only in a uniform flow. This idea is supported both by Hey (19) and by
Fig. 4 in which numerous data for flows with nonuniform bed profiles
634
(by less than the size of the bed material) could be quite sufficient to
cause the reported trend below the line of Eq. 2. It is important, there-
fore, to select slope-area sites carefully on the basis of uniform bed pro-
file and to survey the cross sections in detail sufficient to account for
the projections of the bed material.
Second, Fig. 3 shows that much of the scatter relative to the line of
Eq. 2 for the higher flows included in Fig. 2 results not from random
scatter but from at-a-site variations of (8//) 1/2 with d/D 84 which do not
parallel the line of Eq. 2. At sites where such deviations occur, it is no-
ticeable that they tend to begin in the region of intermediate-scale rough-
ness (1 < d/DM < 4), which suggests that the region is one in which
the effects of large-scale roughness are drowned out and resistance falls
to values compatible with small-scale roughness. Measured values of
(8//) 1/2 may then be up to 66% higher than values predicted by Eq. 2.
Such systematic deviations are less evident for the flume data (Fig. 5)
where equivalent at-a-site variations (for flows with a given bed material
and slope) tend to plot in a more restricted range.
From the foregoing, it is clear that patterns of flow resistance behavior
for high-gradient, boulder-bed rivers do not parallel those for lowland
rivers, at least as expressed by the semilogarithmic type of equation.
Modification of Eq. 2 is, therefore, necessary if it is to represent the ob-
served variations. One proposal, in which allowance is made for the
projections of the boulders into the flow, has been put forward by
Thompson and Campbell (30), but it does not apply here since it has
the effect of reducing the predicted value of (8//) 1 / 2 below that of Eq. 2
rather than raising it, as required. Instead, examination of the data sug-
gests that the most likely causes of the observed variation in resistance
are processes connected with bed material size distribution and channel
slope, since the river sites with the more rapid rates of change of resis-
tance [Tromie 2, Ettrick (Brockhoperig)] tend to have wider size distri-
butions (cr S 0.35) and steeper slopes (S = 1%), whereas the less rapid
rates of change are found at sites (Tweed and Findhorn) with narrower
size distributions (a s 0.31) and gentler slopes (S ~ 0.4%). The possible
dependence of at-a-site variations in flow resistance on these two pa-
rameters is, therefore, examined in more detail.
Effect of Bed Material Size Distribution.—This is likely to affect both
the relative roughness area, AW/AT (the proportion of the total cross-
sectional area occupied by significantly projecting boulders), and the
roughness concentration or spacing (the proportion of the bed area oc-
cupied by significantly projecting boulders). The two factors are ob-
viously related and together provide a three-dimensional picture of bed
material disposition in the channel.
Relative roughness area significantly affects flow resistance by deter-
mining the degree to which the flow is funnelled between projecting
elements (6,17,30). A connection with bed material size distribution and
635
(Brock hoperig)
v 0.490 Tromie 2
8 0.4
<
- ^ -
in
S 0.3
c:
- «\ -
en
o
- \ -
* 0 . 2 - o \ _
Of
_ s\.
"5 A W / A T = 0.275
» 0.1
- 0.375 l o g l d / D a 4 )
\ o _
0 i i I 1 , 1 111 1 I i 1 I I II
0.1 0.5 1 5 10
Relative Submergence d/D84
in which the range of a = 0.247 - 0.49 (Eq. 3). For d/Du > 4, a different,
less rapid rate of change is evident. This corresponds to the region of
small-scale roughness in which AW/AT (s5%) does not change much and
in which the effect of flow funnelling is not significant. Thus, the im-
portance of the region of intermediate-scale roughness as one of tran-
636
637
not completely explained by this correlation and the data are not suffi-
cient to allow any dependency on slope to be quantified.
Sediment Transport and Flow Resistance.—Many studies have in-
vestigated the interaction of sediment transport, bedform development
and flow resistance for sand-bed channels (e.g., Ref. 29), but rather less
is known about gravel- and boulder-bed channels. Both Griffiths (16)
and Charlton, et al. (11), have proposed, on the basis of field studies,
that separate flow resistance relationships should be used for moving
and stationary beds. On the other hand, several flume studies have shown
that for gravel-bed channels with high gradients, the preferred two-di-
mensional bedforms are the transitional plane bed and the antidune (7,32),
neither of which is likely to increase very significantly flow resistance
due to form drag (29). Three-dimensional bed features such as alternate
bars also do not seem to produce a much higher resistance through form
drag (at least at sediment-moving flows) (21,27), although the consid-
erable sediment transport associated with their movement and forma-
tion could increase resistance (7).
In order to study the influence of sediment transport, the variation of
(8//) 1/2 with d/D50 for flume (7,25) and river flows (14,16) with sediment
transport is shown in Fig. 7. D50 is used because Griffith's data (16) do
not include values for D ^ , and the line of Eq. 2 is accordingly based on
6.8 Dso rather than 3.5 D M . Only data for channels with slopes greater
5 10
Relative Submergence
FIG. 7.—Variation of (8//)1/2 with Relative Submergence, d/D50, for Fiume and
River Flows with Sediment Transport, Using Data from other Studies
638
The rather negative conclusion from the foregoing analysis is that the
complicated nature of the flow resistance processes, coupled with the
lack of available data, currently prevent the development of a satisfac-
tory practical method for predicting flow resistance in steep mountain
rivers. In particular, Eq. 2, while applying in an approximate way, can-
not easily be extended to cover all conditions in such rivers. Application
of the results of this study is, therefore, restricted to an empirical ap-
proach in which the likely limits to the resistance variation are deline-
ated, and the flow resistance variation itself is calculated from the avail-
able data (Fig. 8).
In Fig. 8, the envelope of available data refers to data for uniform flow.
(The diagram is based on data from this and other studies for flows at
slopes exceeding 0.4%; roughness scale and Eq. 2 are included for com-
parison.) For d/Dsi < 1 (large-scale roughness), the data lie in a relatively
restricted range and the average rate of change of (8//) 1/2 with d/DM is
parallel with the line of Eq. 2. For 1 < d/DBi < 6 (essentially interme-
diate-scale roughness), the range widens and, on average, the flow re-
639
9 - envelope
/
' sediment —
FIG. 8.—Summary Diagram for Calculation of (8//) 1/2 as Function of d/DM for Uni-
form Flows
sistance changes more rapidly with relative submergence. For both these
regions, data points for nonuniform or over-bank flows are likely to fall
below the envelope. For d/D8i > 6 (small-scale roughness), the pattern
of variation is unclear, but on average the rate of change appears to level
off. A possible reason for this is that flows with higher values of (8//) 1/2
simply do not occur in mountain rivers: Generally, the steeper the
channel gradient, the lower are the values of d/DM and (8//) 1/2 at which
bankfull and flood conditions are attained. Certainly there are few pub-
lished values of (8//) 1/2 greater than about 12 for channels with slopes
exceeding 1%.
There are three simple but different ways of quantifying the variation.
The first is to plot three straight-line equations along the center of the
envelope—one for each roughness scale. Given the empirical approach,
this is logical, but it has the disadvantage of setting the flow resistance
constant for d/DM > 6, a pattern which conflicts with theory. The second
method is to adopt the Hey equation (Eq. 2). However, while this equa-
tion provides a reasonable average value of (8//) 1/2 for d/DM > 6, it lies
along the bottom of the envelope for shallower flows. The third ap-
proach, adopted here, is to combine the first two, defining an equation
which is parallel to Eq. 2, but which lies nearer the center of the en-
velope. The equation suggested here lies along the center of the region
of large-scale roughness and along the center of the flume data in
Fig. 5
dicted by Eq. 5, (8//) t 1/2 , relative to the observed value, (8//)J /2 , was
J calculated as [(8//)J/2 - (8//) c 1/2 ]/[(8//)J /2 ]. For the 44 data points, the
i mean error is about - 1 3 % . The root mean square error, (2e 2 /n) 1/2 , is
J about 34%, in which e = the individual error; and n = the number of
1
data points. It should be noted, however, that the data are not well
Downloaded from ascelibrary.org by Illinois Inst of Technology on 06/24/13. Copyright ASCE. For personal use only; all rights reserved.
\ CONCLUSION
ACKNOWLEDGMENTS
Considerable and generous help with the selection of field sites and
the provision of discharge data was given by: Mr. J. M. Storey (Nor-
thumbrian Water Authority), Mr. C. S. Stevenson (Tweed River Purifi-
cation Board), Mr. J. A. Reid (Tay River Purification Board), Mr. K. Grib-
ben and Mr. D. Fraser (Northeast River Purification Board) and Mr. T.
) Inglis (Highland River Purification Board). Miss A. Whitcombe, Mr. D.
Lapping and Mr. A. Flower (Institute of Hydrology and Coventry Poly-
technic) were cheerful and excellent field assistants under adverse con-
ditions. Part of the work on this paper was carried out while the writer
was visiting the Ecole Polytechnique Federale de Lausanne, Switzerland,
and he therefore thanks Prof. W. H. Graf for the generous provision of
641
APPENDIX.—REFERENCES
642
Division, ASCE, Vol. 105, No. HY4, Proc. Paper 14500, Apr., 1979, pp. 365-
379.
20. Hey, R. D., and Thome, C. R., "Accuracy of Surface Samples from Gravel
Bed Material," Journal of Hydraulic Engineering, ASCE, Vol. 109, No. 6, Proc.
Paper 18047, June, 1983, pp. 842-851.
21. Jaeggi, M. N. R., "Formation and Effects of Alternate Bars," Journal of Hy-
draulic Engineering, ASCE, Vol. 110, No. 2, Proc. Paper 18580, Feb., 1984, pp.
142-156.
22. Judd, H. E., and Peterson, D. F., "Hydraulics of Large Bed Element Chan-
nels," Report No. PRWG 17-6, Utah Water Research Lab., Utah State Uni-
versity, Logan, Utah, 1969.
23. Keulegan, G. H., "Laws of Turbulent Flow in Open Channels," Journal of
Research of the National Bureau of Standards, Vol. 21, Research Paper RP1151,
Dec, 1938, pp. 707-741.
24. Limerinos, J. T., "Determination of the Manning Coefficient from Measured
Bed Roughness in Natural Channels," Water Supply Paper 1898-B, United States
Geological Survey, Washington, D.C., 1970.
25. Meyer-Peter, E., and Mxiller, R., "Formulas for Bed-Load Transport," Pro-
ceedings of the Second Meeting of the International Association for Hydraulic Struc-
tures Research, Stockholm, Sweden, Appendix 2, June, 1948, pp. 39-64.
26. Moore, R. C , "Modeling Hydraulic Roughness in Steep Mountain Streams,"
thesis presented to Colorado State University, at Fort Collins, Colorado, in
1980, in partial fulfillment of the requirements for the degree of Master of
Science.
27. Parker, G., and Peterson, A. W., "Bar Resistance of Gravel-Bed Streams,"
Journal of the Hydraulics Division, ASCE, Vol. 106, No. HY10, Proc. Paper 15733,
Oct., 1980, pp. 1559-1575.
28. Rouse, H., "Critical Analysis of Open-Channel Resistance," Journal of the Hy-
draulics Division, ASCE, Vol. 91, No. HY4, Proc. Paper 4387, July, 1965, pp.
1-25.
29. Simons, D. B., and Richardson, E. V., "Resistance to Flow in Alluvial Chan-
nels," Professional Paper 422-J, United States Geological Survey, Washington,
D.C., 1966.
30. Thompson, S. M., and Campbell, P. L., "Hydraulics of a Large Channel Paved
with Boulders," Journal of Hydraulic Research, International Association for
Hydraulic Research, Vol. 17, No. 4, Delft, The Netherlands, 1979, pp. 341-
354.
31. Virmani, J. K., "The Relationship between Channel Forming Flows and the
Cross-Section Shape, Slope and Bed Materials in Large Bed Element Streams,"
thesis presented to Utah State University, at Logan, Utah, in 1973, in partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
32. Whittaker, J. G., and Jaeggi, M. N. R., "Origin of Step-Pool Systems in
Mountain Streams," Journal of the Hydraulics Division, ASCE, Vol. 108, No.
HY6, Proc. Paper 17160, June, 1982, pp. 758-773.
33. Williams, G. P., "Bank-Full Discharge of Rivers," Water Resources Research,
Vol. 14, No. 6, Dec, 1978, pp. 1141-1154.
34. Wolman, M. G., "A Method of Sampling Coarse River-Bed Material," Trans-
actions, American Geophysical Union, Vol. 35, No. 6, Part 1, Dec, 1954, pp.
951-956.
35. Zevenbergen, L. W., "Flow Resistance of Boulder-Bed Streams," thesis pre-
sented to Colorado State University, at Fort Collins, Colorado, in 1983, in
partial fulfillment of the requirements for the degree of Master of Science.
643