Improving cold-region streamflow estimation by winter precipitation adjustment using passive microwave snow remote sensing datasets

Abstract

Winter precipitation estimations and spatially sparse snow observations are key challenges when predicting snowmelt-driven floods. An improvement in streamflow prediction is achieved in a snowmelt-dominant basin, i.e. the Red River Basin (RRB), by adjusting the amounts of snowfall through satellite-borne passive microwave observations of snow water equivalent (SWE). A snowfall forcing dataset is scaled to minimize the difference between simulated and observed SWE over the RRB. Advanced microwave scanning radiometer-E (AMSR-E) SWE products serve as the observed SWE to obtain the solution to the linear equation between the AMSR-E and the baseline (no snowfall-forcing adjustment) SWE to yield a multiplication factor (M factor). In the headwaters of the RRB in the United States, a Nash–Sutcliffe efficiency (NSE) of 0.74 is obtained against observed streamflow, with M factor-adjusted streamflow during the snowmelt seasons (January to April). The baseline streamflow simulation without M factor exhibits an NSE of 0.38 owing to an underestimated SWE.