Long lead-time radar rainfall nowcasting method incorporating atmospheric conditions using long short-term memory networks

High-resolution radar rainfall data have great potential for predictions up to 6 h ahead (nowcasting); however, conventional extrapolation approaches based on in-built physical assumptions yield poor performance at longer lead times (3–6 h), which limits their operational utility. Moreover, atmospheric factors in estimate errors are often ignored. This study proposed a nowcasting method that attempts achieve accurate of using long short-term memory (LSTM) networks. Atmospheric conditions were considered reduce errors. To build models LSTM networks (LSTM-RN), approximately 11 years radar, gauge rainfall, and from the UK obtained. Compared with built optical flow (OF-RN) random forest (RF-RN), LSTM-RN had lowest root-mean-square (RMSE), highest correlation coefficients (COR), mean bias closest 0. Furthermore, showed growing advantage times, RMSE decreasing by 17.99% 7.17% compared OF-RN RF-RN, respectively. The results also revealed strong relationship between weather conditions. provides an effective solution enhances forecast value supports practical