Ensemble streamflow prediction adjustment for upstream water use and regulation

Abstract

• Hydrologic forecasts are commonly biased in watersheds with water use/regulation. • Approach to (a) characterize flow biases/uncertainties and (b) adjust forecasts. • Demonstrations/validation in three watersheds of the American River, California. • Readily applicable to other regions. • Developed for operational use by the NWS River Forecast Centers. Hydrologic model forecasts are commonly biased in watersheds where water use and regulation activities cause flow alterations. Furthermore, direct accounting of such biases in forecast preparation is impractical as the information required is extensive and usually unavailable. This article introduces a new method to characterize the aggregate flow alteration biases and associated uncertainty in watersheds with important but largely undocumented water use and regulation activities. It also uses these assessments to adjust the ensemble streamflow predictions at downstream locations. The method includes procedures to (a) detect the presence of significant upstream regulation and water use influences; (b) correct the ensemble streamflow predictions and associated uncertainty for any biases in periods when such influences are detectable; and (c) assess the adjusted forecast reliability improvements. Applications in three watersheds of the American River in California demonstrate that the new method leads to significant forecast skill improvements and is also readily applicable to other regions.