Predicting Peak Flow of Small Watersheds by use of Channel Characteristics

Abstract

In previous studies, a method was developed for predicting the effects of channel characteristics, including watershed size and shape on peak flow from small watersheds. The method was incomplete, however, since it lacked a working method of estimating the time parameter for ungaged watersheds. Therefore, the first objective of this study was to satisfy this need. The second objective was to test the overall method as a means of predicting peak flow for small ungaged watersheds, given the runoff volume. The overall method begins with a hydrologic analysis of numerous rainfall-runoff events observed at selected experimental watersheds. This yields certain hydrologic parameters which can be evaluated only for gaged watersheds. Then, the physical characteristics of these watersheds, primarily the channel characteristics, are utilized to evaluate the same parameters by use of an hydraulic or flow approach. If this can be accomplished successfully, the same procedure can be applied to ungaged watersheds.

The following conclusions can be made based on the results of the study: A new time parameter, time to 50% of equilibrium, T50, was proposed. It can be evaluated hydrologically, i.e. from observed hydrographs in many but not all cases this is essential if it is to be used in peak flow predictions for other, ungaged watersheds. T50 increases with watershed size, approximately as watershed area to the 1/3 power. It decreases as the mean rate of rainfall excess (qs), increases, varying as qs to the minus ½ power (roughly). The residual variability is substantial, indicating that other factors also affect T50 significantly. The channel characteristics, cross-sectional area and wetted perimeter can be estimated with reasonable accuracy from measurements of bankfull topwidth and depth. However, a number of complete channel cross-sections must be taken or be available in the region in order to evaluate the two coefficients needed, one for area and one for wetted perimeter. It appears likely that these coefficients can be generalized through further study, and also that relationships will have other applications in watershed engineering. Three methods used to divide the watershed into an upper and lower half hyrdologically gave only slightly different results, and therefore, the simplest method (arc) appears preferable. The travel time approach to evaluating the time parameter yields values (designated Tch) that are consistently and significantly lower than the true values (T50). Thus a coefficient applicable to the region is necessary to related T50 to TCH. The peak flow predictions by the methods of this study were quite variable, as compared to the observed values, but on the average were about the right magnitude., i.e. neither consistently high or consistently low. The combination of peak flow equation, the time parameter, T50, and the relationship of Cp, the peak flow coefficient, to the ration D/T50, where D is the duration of rainfall excess, appears to provide a satisfactory but not highly accurate procedure for estimating peak runoff, given the volume of rainfall excess and its approximate time distribution.