Impact of assimilating spaceborne microwave signals for improving hydrological prediction in ungauged basins

Abstract

The availability of in-situ data has been a constraining issue in hydrological prediction, especially in those regions that are only sparsely monitored or completely ungauged. The application of remote-sensing data, without conventional in-situ hydrological measurements, to force, calibrate and update a hydrologic model is a major contribution of this study. First, a rainfall-runoff hydrological model called CREST, coupled with an Ensemble Square Root Filter (EnSRF), is used for exceedance probability-based flood prediction. Then, this advanced flood prediction framework, with different experimental designs, is forced by TRMM precipitation while Aqua AMSR-E microwave brightness temperature signals are used for model calibration and data assimilation for progressively improved river discharge prediction. Results indicate that solely relying on remote-sensing data for model forcing, parameter calibration, and state updating with EnSRF, the designed framework can adequately predict flooding events. A high flow threshold was applied and has further improved modeling performance, particularly in the flooding seasons, with a flood warning lead-time of one day. Given the anticipated global availability of satellite-based precipitation (i.e. GPM) and AMSR-E like passive microwave signal information (i.e. SMAP) in near real-time, this proposed research framework could potentially contribute to the exceedance probability-based flood prediction in the vast sparsely gauged or ungauged basins around the world.