A WRF/WRF-Hydro Coupled Forecasting System with Real-Time Precipitation–Runoff Updating Based on 3Dvar Data Assimilation and Deep Learning

This study established a WRF/WRF-Hydro coupled forecasting system for precipitation–runoff in the Daqing River basin northern China. To fully enhance skill of system, real-time updating was performed both WRF precipitation forecast and WRF-Hydro forecasted runoff. Three-dimensional variational (3Dvar) multi-source data assimilation implemented using model by incorporating hourly weather radar reflectivity conventional meteorological observations to improve accuracy precipitation. A deep learning approach, i.e., long short-term memory (LSTM) networks, adopted flow. The results showed that had positive impact on range trends forecasts. quality outputs significant performance flow at catchment outlet. With runoff driven forecasts being updated 3Dvar assimilation, error flood peak decreased 3.02–57.42%, volume 6.34–39.30%, Nash efficiency coefficient increased 0.15–0.52. implementation LSTM can effectively reduce errors particularly those time-to-peak volumes.