7930 - Laserfiche WebLink

Home Browse Search

No trends were established, and it appeared that it was a matter of chance that the sediment yield estimates were close to the sediment dam yields. The most severe underestimation of sediment yield by both methods was for the first 2 years after road construction in watershed SC-4 and for years of high annual water yield in some watersheds. These are the data points below the dashed line in figure 8. This was probably the result of too few samples to accurately define the sediment-discharge rela- tionship following disturbance and at times of active channel bed movement. During higher than normal streamflow, channel sediment storage features may break up and re-form. This may release pulses of sedi- ment that were not measured by weekly or every-other- day sampling. This again points out the importance of storage in these high-energy third-order streams. Because the time-integration method appears to esti- mate sediment yield as well as the rating equations but does not require continuous or mean daily flow records, this method is probably more desirable for use on ungauged streams and on gauged streams where good sediment rating equations cannot be developed. This conclusion is based on the results from the Silver Creek research watersheds just discussed. The same conclusion cannot be reached for the data from the Forests because there is no known value to check against. To make the time-integration method work, samples must be well dis- tributed over the entire hydrograph so that highs as well as lows are sampled. Streams in Silver Creek were gener- ally sampled every other day during snowmelt with a morning sample one time and an afternoon sample the next time. In this way not only the general hydrograph was sampled, but daily hydrograph variations were accounted for such that, by averaging the sediment rates of two consecutive samples, near average condi- tions were simulated. If all samples were taken near hydrograph peaks, sediment yield would be overesti- mated. On the other hand, if peak flows were not sam- pled, sediment yield would be underestimated. Poor sam- pling will provide erroneous estimates of sediment yield. The number of samples necessary to characterize a hydrograph depends on flow conditions. In snow- dominated systems typical of the Northern Rocky Moun- tains, the majority of sediment transport occurs during spring snowmelt. This may be completed in 3 or 4 weeks one year and 6 or 8 weeks the next year. Generally, sam- pling during this melt period is sufficient to predict annual sediment, particularly in bedload-dominated sys- tems, because this is the major time that streamflow is sufficient to transport bedload. Where suspended sedi- ment is more important, samples may be required throughout the year because suspended sediment may frequently be mobilized during summer storms. Manage- ment impacts may produce readily available sediment that may be transported during the summer as well. This should be considered in sample collection planning. The sediment yield estimates, based on the pumped samples and DH-48 hand samples, gave differing esti- mates, and these estimates are quite different from the sediment dam estimates (table 2). Pumping samplers are designed for sampling suspended load, but the intakes for these samplers in Silver Creek were in a hydraulic jump 1 to 2 cm from the bottom, in an attempt to sam- ple bedload as well. Because of the large percentage of suspended sediment rating equations that were not sig- nificant, a comparison of suspended sediment yield from DH-48 samples and pumped samples was done using the time-integration method estimates (table 4). The cor- respondence of the paired sediment yield estimates was variable, but a paired t test indicated that the pumping sampler estimates were significantly higher than the hand samples at the 90 percent level. This suggested that because of the intake position, the pumping sam- plers were indeed sampling more than just the sus- pended sediment load of the streams. However, the pumping samplers were not fully sampling total sedi- ment when compared with the sediment dams (table 2). The detailed analysis of rising and falling limbs of the hydrograph for the pumped samples did not provide enough significant rating equations to model each storm event during the year. Also, many of the significant equations that resulted from five to 10 samples on a ris- ing or falling limb covered such a narrow range of dis- charge that extreme caution was necessary when predict- ing sediment. So narrow were many of these discharge ranges that the equation was not useful for prediction. This was probably because the relationship between the five to 10 sediment concentrations and discharges had a high variance, and therefore the true line may have been much different. Many of these relationships had extremely low coefficients of determination. Less than half the significant equations had coefficients of determi- nation of 0.60 or more. The best results were achieved by dividing the data set at the annual snowmelt hydro- graph peak or by season. In some years the most useful relationship resulted from using the entire data set. Table 4-Comparison of suspended sediment yield estimates using DH-48 hand samples and automatic pumping sampler samples Sediment dam Sediment yield by time integration Stream Year yield DH-48 Pumping sampler Megagrams per year SC-1 79 5.3 1.9 2.1 81 15.4 4.9 3.4 SC-2 79 3.4 .5 .6 80 42.8 6.1 1.2 81 9.3 2.1 2.5 SC-3 79 2.9 .6 .6 80 14.1 2.7 4.1 81 7.3 2.6 2.6 SC-4 79 4.5 .9 .7 80 50.6 7.2 14.4 81 43.5 6.1 16.6 SC-5 79 2.7 .4 1.9 80 10.7 2.1 13.7 81 8.5 1.8 4.8 SC-6 79 3.5 1.1 1.2 80 20.8 12.7 12.0 81 9.1 2.4 3.1 10