This report presents an analysis and evaluation of the trend of the sediment yield for the Brandywine Creek at Wilmington, Del., for the period from December 1946 to September 1955. The interest in such an analysis and evaluation stems from the efforts of the Brandywine Valley Association and others to reduce erosion and improve land use in the watershed.
The data used for the analysis were taken from the continuous suspended-sediment and water-discharge records of the stream at Wilmington and the precipitation records at 8 standard and 1 recording rain gages. The analysis was made on the basis of 123 storm events for this period of record using only the water and sediment discharge attributed to direct runoff. These data represent 89 percent of the total sediment discharge and 19 percent of the total water discharge.
The sediment load for each of the storm runoff events was correlated with storm runoff, rainfall intensity, and season to remove the effect, if ant, of the variation caused by these factors. The evaluation of the relative trend of sediment yield was made by two methods; first, the accumulative graph pr double mass curve as a graphical method, and second, the rank correlation method which results in a numerical coefficient and its significant. The graphical method of this evaluation shows an approximate 38 percent decrease in sediment yield for the period 1952 to 1955 from that for the period 1947 to 1951. The rank correlation coefficient was 0.152 for the same analysis showing a very high level (almost 99 percent) of confidence in the significance of a decreasing trend.
A parallel analysis to that above using "peakedness" instead of rainfall intensity as a measure of storm intensity was made because "peakedness" is easier to evaluate than rainfall intensity. The results of this analysis again indicates the probably decreasing trend of sediment yield as shown by the change in slope of the accumulative graph from 0.77 fir the 1947 to 1951 period to 1.05 for the 1952 to 1955 period or 28%, and by the rank correlation coefficient of 0.114 with a 94 percent level of confidence of a decreasing trend.
Appendix A of this report shows a similar analysis by usede of monthly sediment discharge and "direct" runoff as determined for an analyses of runoff patterns. The results indicate only a small decreasing trend, if any, in sediment yield. This is probably due to the fact that the correlations with rainfall and season for adjusting these data were rather poor--indicating bias by unknown parameters that cannot be evaluated.
Appendix B presents the methodology of using rank correlation to evaluate the trend of variate-values with respect to time with special reference to the trend of sediment yield from a watershed.