Achievement: The precipitation and flood forecasting methods based on radar observation system

Achievement Owner: Tsinghua University

Introduction: The precipitation and flood forecasting methods which can generate accurate forecasts are developed based on the monitoring system in this project. The methods are applied in the Suyukou Valley of Helan Mountain to support the flood control operation and perform well. The developed short-term precipitation forecasting method, which couples the physical mechanism and deep learning method and is based on the scanning data of the Doppler radar, is used to forecast precipitation in the future 30 minutes and the Nash-Sutcliffe efficiency coefficients (NSEs) are between 0.69 and 0.89. Based on the predicted and observed precipitation, the NSE value of the flood forecasts with 2-hour forecast lead time is 0.65.