Abstract:
Simulation of storm hydrographs in the Oregon Coast Range was explored using the Soil Conservation Service (SCS) curve number methodology, and by developing and testing an antecedent precipitation index (API) method.
Standard SCS procedures over-estimated peak discharge by about a factor of two (i.e., average over-prediction of 118 percent). When an average curve number was derived for Deer Creek (an Oregon Coast Range stream), errors in predicted peak flows averaged 26.8 percent. Even with adjustment of SCS parameters (watershed lag, shape of the unit hydrograph, and curve number), the simulated
hydrograph shape and timing of predicted peak flows did not match with observed hydrographs. The assumed rainfall-runoff relationships of the SCS method are unable to account for changing runoff responses related to the time distribution of precipitation, and therefore provides an unrealistic approach to storm runoff simulation. The SCS runoff curve number method is not recommended for estimation of peak discharge nor simulation of storm hydrographs in Oregon's Coast Range. A simple rainfall-runoff model, was developed, which requires only precipitation and watershed area as inputs.
An antecedent precipitation index (API) was developed by decaying the residual effects of precipitation observations through time. Coefficients used to decay API values were derived from recession analyses of storm
hydrographs during periods of no rainfall. Linear regression was used to correlate API and discharge values for five Coast Range watersheds. Model coefficients for the five watersheds were used to predict the API-discharge relation for a sixth coastal watershed. Errors in peak flow estimates for Deer Creek and the independent test
watershed averaged 10.7 and 17.8 percent, respectively. Storm runoff volume errors for all watersheds averaged 15.9 percent, and storm hydrograph shape was accurately
simulated. Errors in peak discharge and volume estimates may be attributed to differences in timing between observed and simulated hydrographs, seasonal variation in
antecedent moisture, and effects of snowmelt during rainfall. Temporal and spatial variability in precipitation observations were also evaluated. API methods may be useful in frequency analyses (in areas where rainfall records are longer than runoff records), estimation of missing data, slope stability research, and suspended sediment modeling.
