Estimating Uncertainty of Stormwater Runoff Computations

Abstract

The stormwater runoff quantity component of the U.S. Army Corps of Engineers storage, treatment, overflow, and runoff model (STORM) program has been successfully calibrated and verified for a residential community located in Dallas, Texas. Limited water quality data preclude application of typical statistical testing of model runoff quality predictive capabilities. A Monte Carlo simulation technique is therefore employed to ascertain probable ranges of STORM water quality predictions in light of both water quantity and quality input parameter uncertainties. The resulting 95% occurrence intervals of probable model runs are compared with a limited water quality data set to test model adequacy. An original modeling scenario, utilizing suggested areal accumulation rates derived from a study conducted in Seattle, Washington, is rejected because the measured total suspended solids concentrations are far above the upper bound of the computed 95% occurrence interval. A second modeling scenario using areal accumulation rates obtained from Tulsa, Oklahoma, cannot be rejected based upon the comparison of measured data and the computed 95% occurrence intervals.