Storm runoff simulation using an antecedent precipitation index (API) model

Abstract

An antecedent precipitation index (API) model for predicting storm runoff was developed for use in the Oregon Coast Range, U.S.A. A coefficient $\text{K}$, derived from recession analysis of storm hydrographs, was used to decay over time the effects of antecedent precipitation upon runoff. Linear regression was then used to establish a relationship between calculated API values and instantaneous streamflow. The model was initially evaluated with data from five “calibration” watersheds. Absolute errors in peak flow and storm-runoff volume estimates, compared to observed values, for the calibration watersheds averaged 15 and 14%, respectively; average errors were only −1 and 4% respectively. Coefficients for the five calibration watersheds were also used to determine API coefficients for a sixth “test” watershed. Absolute errors in peak flow and storm-runoff volume estimates, compared to observed values, for the test watershed averaged 18 and 21%, respectively; average errorswere −6 and −4%, respectively. The API model has minimal data requirements and appears to provide a relatively simple, objective method for stimulating storm hydrographs from a wide variety of storm hyetographs.