Response of Dry Creek, California, to land use change, gravel mining and dam closure

Transcription

1 Erosion and Sedimentation in the Pacific Rim {Proceedings of the Corvallis Symposium, August, 1987). IAHS Publ. no Response of Dry Creek, California, to land use change, gravel mining and dam closure INTRODUCTION MICHAEL D. HARVEY, and STANLEY A. SCHUMM Earth Resources Department, Colorado State University, Fort Collins, Colorado 805, USA ABSTRACT A 21.5 km reach of Dry Creek, downstream of Warm Springs dam, was studied in order to determine its responses to land use change, gravel mining and.dam closure. Land use changes ( ) caused aggradation and then degradation, but the channel was in equilibrium by Gravel mining in Dry Creek and Russian River ( 's) reduced base level of Dry Creek by 3 m. The channel degraded by an average of 3.2m, and by 1984 mean channel width and depth, at bankfull stage, had increased from m and from m, respectively. Channel longitudinal profiles were not significantly different; therefore, the major response of Dry Creek to the imposed changes was an increase of width. Dam closure (1983J did not cause the expected further degradation of Dry Creek because of channel armoring, tributary contribution of sediment, emplacement of 3 grade-control sills, and the presence of bedrock outcrops in the channel. Since the 1850's the drainage basin and channel of Dry Creek, a tributary to Russian River, Sonoma County, California (Fig. 1) have been subjected to land use change, gravel mining and dam closure. The objective of this study was to describe the effects of the first two perturbations and to ascertain the probable effects of the third. Land use change commenced in the 1850's and at that time about half of the 562km 2 drainage basin was forested. By 1870 about 40% of the forest had been cleared and converted to grazing land, and this caused greatly accelerated slope erosion (Ritter and Brown, 1971). Gravel mining commenced in Russian River about 1900, but it was reported to have had little effect on Dry Creek until the 1950's when Coyote and Healdsburg dams were constructed on Russian River, thereby preventing the replenishment of the 451

2 452 Michael D.Harvey & Stanley A.Schumm mined gravels (Simons and Li, 1980). As a result about 3m of degradation occurred in Russian River, and this leered base level of Dry Creek. Fig i. Map showing location of study reach of Dry Creek. In 1983, Warm Springs dam was closed to form Lake Sonoma (Fig. 1). Closure of the dam isolated the upper 337km 2 of the basin from the lower 225km 2. The dam is located 21.5km upstream of the confluence with Russian River, and it is this reach of Dry Creek that is the subject of this paper. Data that were available for Dry Creek include historical accounts of the channel (Ritter and Brown, 1971) and seven surveys of the channel; 1940, 1964, 1974, 1976, 1980, 1981, Forty seven permanent ranges had been established along the study reach by 1964, but only of them were re-surveyed in each time period. Mean values of channel width, depth and width-depth ratio, at bankfull discharge, for the entire study reach, are presented in Table 1. Two terraces were observed and mapped. Terrace II is the higher and oldest terrace. The banks of the channel are composite (Thome and Tovey, 1981) with an average median (d5g) grain size ef 5.6mm in the lower bank and an average d5q of 0.07mm in the upper bank. CROSS SECTION ADJUSTMENTS Following European settlement in the 1850's, the forested acreage was reduced by about 40% (Ritter and Brown, 1971). This caused accelerated slope erosion and increased the sediment load of Dry Creek, which aggraded. Evidence for this is coarsegrained (Û5Q = 5.6mm) sands and gravels that overlie a buried soil, which represents the floodplain elevation of Dry Creek

3 Response to land use 453 prior to settlement. The thickness of this post-settlement alluvium ranges from m (average 0.9m). By about 1900 sediment delivery to the channel had decreased, and the channel incised until it was about 1.4m deep and channel width was about 10m (Cleveland and Kelley, 1977). The incision formed Terrace II which was 3.7m above the stream bed. Historical accounts indicate that the channel had attained a new state of equilibrium by about Table 1. Slopes of terrace and channel profiles and mean channel width, depth, and width-depth ratio at bankfull stage. Mean Mean Mean Profile Slope Width Depth Width/Depth No. of (std. dev.) (std. dev.) Ratio (std. dev.) Observations Cm) (m) (m) Terrace II Terrace I thalweg (2.7) 3.6 (0.5) 2.8 (1.0) thalweg (21.9) 4.0 (0.8) 21.6 (8.0) 1974 thalweg (30.5) 4.8 (1.3) 21.1 (9.2) 1976 thalweg (33.2) 5.1 (1.4) 19.8 (8.5) 1980 thalweg (35.1) 5.5 (1.2) 18.5 (7.7) 1981 thalweg (36.0) 5.5 (1.3) 18.6 (8.5) 1984 thalweg (36.0) 5.8 (1.3) 19.0 (10.2) Eighteen cross sections were surveyed along the study reach in At bankfull discharge mean channel width was 9.8m, mean depth was 3.6m, and mean width-depth ratio was 2.8 (Table 1). These data indicate that Dry Creek had reincised by 1940, but little channel widening had occurred. The cause of the incision is unknown, but Cleveland and Kelley (1977) and Simons and Li (1980) report that there was gravel mining in Russian River and in the lower reaches of Dry Creek since the turn of the century. The records from the Cloverdale gage (no. 4645J, which is located upstream of the study reach, show that Water Year 1938 was the year of record annual runoff (Cleveland and Kelley, 1977), which could be the cause of the incision. In the period from 1940 to 1984 the channel of Dry Creek continued to degrade and widen (Table 1). By 1984, mean channel width was 101.8m, mean depth was 5.8m, and mean width-depth ratio was 19. The incision that occurred prior to 1940 and up to 1984 formed Terrace I, which is an average of 4.6m above the 1984 bed of the channel. The cumulative effects of gravel mining in both Russian River and the lower reaches of Dry Creek and the construction of Coyote and Healdsburg dams on Russian River were responsible for 3m of base level lowering. Cleveland and Kelley (1977) and Simons and

4 454 Michael D.Harvey & Stanley A.Schumm Li (1980) reported that the base level lowering occurred in the mid-1950's, but they were unaware of the existence of the 1940 survey which shows that Dry Creek was degraded to an average depth of 3.6m at that time (Table 1), which is significantly greater than the 1.4m channel'depth in about Dry Creek was reported to be widening rapidly in the late 1950 s and early I960*s to the extent that it was the highest sediment yielding tributary to Russian River (2222 t Am 2 /year: Ritter and Brown, 1971). The Cloverdale and Geyersville (No. 4052) gage records show that this was a period of unusually high annual runoff (Cleveland and Kelley, 1977), and Simons and Li (1980) reported that the large storms were coincident with a number of severe fires in the watershed. The coincidence of the high annual runoff and fires caused Dry Creek to respond rapidly to the base level lowering in this time period. Comparison of the cross section profiles at cross section 41, that is 4km from the mouth of Dry Creek, shows that the response of Dry Creek to the incision was complex (Fig. 2). The 1940 thalweg was at a lower elevation than the 1964 thalweg, but the 1964, 1974 and 1984 channels were considerably wider than the 1940 channel. Widening of the channel increased the sediment supply and this caused aggradation. From 1964 to 1984 the cross section deepened. As Dry Creek widened the bankfull depth of the channel increased not only as a result of degradation, but also because the channel widened until the top bank was Terrace II (Figure 2). Therefore, in order to determine the actual depth of incision that was due to 3m of base level lowering, the elevation of Terrace II at each cross section was used as a common datum. The mean values and the standard deviation for 5 surfaces and time periods are shown in Table 2 and these data were used to construct Figure 3. On the basis of historical accounts (Ritter and Brown, 1971; Cleveland and Kelley, 1977) and the observation that a new floodplain had formed 1.4m above the bed within the incised channel of Dry Creek by 1984 (Harvey and Schumm, 1985), it was assumed that the depth of the pre-european settlement channel was also 1.4m. Begin et al. (1981) demonstrated that the ultimate effect of base level lowering by a certain amount is degradation along the entire length of the channel by the same amount and, therefore, 3m of the base level lowering should only have produced a 3m response in Dry Creek. The mean elevation of the bed of the channel in about 1900 was 3.7m below the Terrace II datum. By 1984, the mean elevation of the bed of the channel was 6.9m below the Terrace II datum, and this represents 3.2m of degradation (Fig. 3). Comparison of the minimum thalweg elevations at each of the cross sections indicates that the degradation had ceased by 1974 as far upstream as Yoakim Bridge (16.4km). An outcrop of Franciscan Formation was exposed in the bed of the channel in

5 Response to land use at Lambert Bridge, which is located 10.3km upstream of the mouth (Simons and Li, 1980). Further degradation of the bed of the channel occurred between 1974 and 1981 in the reach upstream of Yoakim Bridge, but no significant change in bed elevation occurred between 1981 ano Aggradation and degradation as a L A \l 50 ' 40 CROSS-CHANNEL DISTANCE (m) XS4I \ R' sill ; 1 1/ 1 // ' // / / / \ \ // //!9Q4 \ \ v j V \ 1974 I9E CROSS-CHANNEL OISTANCE (fl) Fig. 2. Profiles at cross section 41. Table 2. Mean depths of each surface below Terrace II datum. Profile Mean (m) Std. Dev. On) No. of Observations Terrace I thalweg thalweg thalweg thalweg result of the downstream migration of gravel waves caused up to 0.5m of elevation change locally. The erosional activity of the tributaries to Dry Creek within the study reach attest to the headward progression of incision in Dry Creek which lowered their base levels. The tributaries that are located in the lower

6 456 Michael D.Harvey & Stanley A.Schumm reaches of Dry Creek have already responded substantially whereas those that are located in the upper reaches are still actively deepening and widening. LONGITUDINAL PROFILE ADJUSTMENTS Nine thalweg and terrace profiles were identified in Dry Creek, and their slopes were obtained from least-squares regressions (Table 1). The use of a least-squares regression model to describe the profiles of an alluvial stream may appear unusual since the slopes are generally best described by a power function (Hack, 1957). However, the study reach of Dry Creek is located downstream of the headwaters zone which is responsible for the exponential form of the profile. The terrace and thalweg profiles document the response of Dry Creek to change. TERRACE II with Post Seulement Alluvium T _ Pre_-J850_Floodolai Pre-1850 Thalweg ond TERRACE I o DISTANCE FROM LEFT END POINT (m) Fig. 3. Schematic cross section showing the chronological evolution of the channel. Distances below Terrace II are mean values for the individual surfaces in the study reach. Comparison of the slope data in Table 1 indicates that regardless of the changes in the controlling variables, and regardless of whether Dry Creek was aggrading or degrading, the slope of Dry Creek has not varied by more than 4.3% between any two surfaces. Tests for parallelism of the profiles showed that the slopes of the nine profiles were not statistically different (a = 0.05J. The changes in slope have been insignificant when compared to the cross-sectional changes and, therefore, even when

7 Response to land use 457 conditions allow, a river does not necessarily change its gradient (Haible, 1980; Knighton, 1984), although the channel gradients on the terraces may have been different due to differences in channel planform. The parallelism of the 1940 and later profiles, which were the result of base level lowering, may be explainable by the results of Begin et cd. [1981) who demonstrated that base level-induced degradation along a channel will form a new profile with essentially the same form as the original. Five profiles of Dry Creek are shown in Figure 4. The 1964 thalweg profile is at the same elevation as the 1984 bar-top profile, which represents the new floodplain. The actual 1984 beo profile is also shown, and it provides an indication of the variability along the profile even though the coefficient of determination (r z = 0.99) for the least squares regression profile was very high. Three grade-control sills that were emplaced in Dry Creek in 1981 are also shown on the figure. 220 ZOO 5 1 / ISO ISO " ^^V-*^ Ï-" ^ ^ ^ ^ _ ^ ^ ' LAMBERT ^ l^- T > >: -^-^>- ^ \ j S 13 Jr -' a 7 jr^" x^-.: *^\~"~' l "CHEEK i^ ^-rfj 00- LEGEMO 60- L *ST SOUAHES Fir Of 9 RANSE NUMBER \ CSËEK 40- RUSSIAN STATICMm > io IS STATION ((! i 10 S ) Fig. 4. Terrace and channel profiles of Dry Creek. EFFECTS OF DAM CLOSURE Warm Springs dam was closed in 1983, and as a result peak discharges for events with a return period of 2 years, or greater, were reduced by a minimum of 70% (Simons and Li, 1980). Closure of the dam also effectively removed 337km 2 of sediment contributing area to the study reach. The possible consequences of this reduction of both peak discharge and sediment load on the study reach of Dry Creek are (Williams and Wolman, 1984); (1) degradation and (2) reduction in channel size.

8 458 Michael D.Harvey & Stanley A.Schumm Degradation Degradation is most likely to occur in the reach immediately downstream of the dam if the size of the bed material is such that armoring does not occur (Livesey, 1965). Grain-size distribution parameters for the 16 samples of bed material that were collected along the study reach in 1984/1985 suggest that the bed of the channel is armored for at least 3km downstream of the dam (d84 range: 32-45mm). It is concluded that further significant degradation of Dry Creek is also unlikely to occur because of the following: (1) the outcrop of Franciscan Formation is providing local base level control for the channel upstream of Lambert Bridge (10.3km), and outcrops of cemented sands and gravels that overlie Franciscan Formation were observed in the lower part of the channel banks at distances of 11.6 and 18.2km upstream, (2) three grade-control sills were installed in the channel in 1981 between 5.5 and 6.2km (Fig. 5), and (3) the major tributaries to Dry Creek are supplying significant quantities of sediment as a result of base-level lowering-induced erosion. Following dam closure in 1983, discharges equivalent to a 10-year event were released from the dam. Comparison of the cross section data between 1981 and 1984 indicated that no significant degradation occurred as a result of these releases. Further, HEC-6 modeling of the study reach of Dry Creek for 50 years of controlled flows indicated that no degradation would occur in Dry Creek (R. Nishio, U.S. Army Corps of Engineers, written communication). Reduction of Channel Size Reduction of the size of the channel due to reduction in the peak discharges is unlikely because Dry Creek is an incised channel that has deepened and widened considerably as a result of the base level lowerinq (Table 1). However, in common with many incised channels that overwiden beyond their required hydraulic capacity (Schumm e_t a^. 1984; Harvey and Watson, 1986), a new floodplain has formed 1.4m above the bed of the channel, within the incised channel of Dry Creek. The margins of the floodplain are clearly defined by vegetation (e.g. willows) the growth of which is being promoted by both the reduction in peak discharges and the water availability due to the low-flow releases from the dam, which has changed Dry Creek from an ephemeral to a perennial-flow channel (Williams and Wolman, 1984). Widening of Dry Creek since the 1940's has resulted in numerous attempts to prevent further erosion by the installation of many different types of bank protection measures (McBride and Strahan, 1983). Simons and Li (1980) considered that improper placement and poor design of many of these structures was responsible

9 Response to land use 459 for much of the erosion of Dry Creek, but on the whole the measures have been successful in preventing further bank erosion along most of the study reach (Harvey and Schumm, 1985). CONCLUSIONS The channel and drainage basin of Dry Creek have been subjected to three major perturbations since the mid-1800's; land use change, gravel mining and dam closure. Land use change that involved the reduction of forest cover by about 40% increased the sediment load to Dry Creek, and it initially aggraded, as is evidenced by an average of 0.9m of coarse-grained sands and gravels that overlie a buried soil that represented the pre-european settlement floodplain. Subsequently Dry Creek degraded, and this formed Terrace II (Fig. 3). The cumulative effects of gravel mining in both Dry Creek and Russian River, and dam construction in Russian River, that commenced about the turn of the century lowered base level for Dry Creek by about 3m. Coincident fires and large storms caused the study reach of Dry Creek to respond to the base level lowering, and as a result between 1900 and 1984 mean channel width and depth increased (Table 1). Dam closure in 1983 did not cause further degradation of Dry Creek because the 3km-long reach below the dam is armored and eroding tributaries are contributing significant quantities of sediment. Also, the emplacement of 3 grade-control sills in 1981 and the presence of bedrock in the channel will prevent further degradation (Fig.4). Therefore, in contrast to the expected situation dam closure had little affect on Dry.Creek because of its prior history. Base level lowering of 3m induced 3.2m of degradation in Dry Creek (Begin et al., 1981) which in turn lowered the base level of the tributaries to Dry Creek. The upstream tributaries are still actively eroding while those in the lower reach are much less active. Degradation of Dry Creek produced 9 identifiable longitudinal profiles whose slopes were not different statistically (Table 1). Therefore, it appears that in Dry Creek the major response to the imposed changes has been an increase in channel width, which suggests that change of slope does not necessarily occur even when conditions allow (Haible, 1980; Knighton, 1984). ACKNOWLEDGEMENTS This study was funded by the U.S. Army Corps of Engineers, Sacramento District, under Contract No. DACW05-85-P-0064 to Water Engineering and Technology, Inc., P.O. Box 1946, Fort Collins,

10 460 Michael D.Harvey & Stanley A.Schumm Colorado We wish to thank Messrs. Edward Sing, Richard Nishio and Harold Huff and Ms. Lori Copeland of the Corps of Engineers for their support and assistance. REFERENCES Begin, Z.B., Meyer, D.F., and Schumm, S.A. (1981). Development of longitudinal profiles of alluvial channels in response to base-level lowering. Earth Surface Processes and Landforms. 6, Cleveland, G.B., and Kelley, F.R. (1977). Erosion along Dry Creek, Sonoma County, California. California Div. Mines, Spec. Rept Hack, J.T. (1957). Studies of longitudinal stream profiles in Virginia and Maryland. U.S. Geological Survey Prof. Paper. 294B. Haible, W.W. (1980). riolocene profile changes along a California coastal stream. Earth Surface Processes. 5, Harvey, M.D., and Schumm, S.A. (1985X Geomorphic analysis of Dry Creek, Sonoma County, California, from Warm Springs dam to Russian River. U.S. Army Corps of Engineers, Sacramento District: Contract DACW05-85-P Harvey, M.D., and Watson, C.C. (1986). Fluvial processes and morphological thresholds in incised channel restoration. Water Resources Bull. 22(3), Knighton, D~. (1984). Fluvial Forms and Processes. Edward Arnold, Baltimore. McBride, J.R., and Strahan, J. (1983). Evaluating rip-rapping and other streambank stabilization techniques. California Agriculture. 7, 7-9. Ritter, J.R., and Brown, W.M. (1971). Turbidity and suspendedsediment transport in the Russian River basin, California. U.S. Geological Survey Open-File Rept. Schumm, S.A., Harvey, M.D., and Watson, C.C. (1984). Incised Channels: Morphology, Evolution and Control. Water Resources Publ. Littleton, Colorado. Simons, D.B., and Li, R.M. (1980). Erosion and sedimentation analyses of Dry Creek, Sonoma County, California. Simons, Li and Associates Fort Collins, Colorado. Thorne, C.R., and Tovey, N.K. (1981). Stability of composite river banks. Earth Surface Processes and Landforms. 6, Williams, G.P., and Wolman, M.G. (1984). Downstream effects of dams on alluvial rivers. U.S. Geological Survey Prof. Paper