Improving Subseasonal Prediction of Summer Extreme Precipitation Over Southern China Based on a Deep Learning Method

The reliable Subseasonal-to-Seasonal (S2S) forecast of precipitation, particularly extreme precipitation, is critical for disaster prevention and mitigation, which however remains a great challenge for mission agencies and research communities. In this study, a deep learning method based on U-Net with additional atmospheric factor forecasts (e.g., wind and specific humidity at multiple levels) included is proposed to correct the S2S forecasts of summer precipitation derived from European Centre for Medium-range Weather Forecasts (ECMWF) over Southern China. The weighted loss function integrated by mean square error and threat score is introduced to capture extreme precipitation more precisely. Generally, the U-Net model improves forecast skills in terms of both general statistics and extreme events, showing prominent superiorities to the ECMWF and quantile mapping (QM) forecasts. Importantly, it shows pronounced calibrations on extreme precipitation forecasts at lead times of 3-4 weeks with the averaged HSS increased by ~5%, which shows higher improvement magnitudes than those at lead times of 1-2 weeks. For all lead times, the greatest forecast skills are mainly distributed over the middle and lower reaches of the Yangtze River basin, presenting HSS of greater than 30% even for the 4-week lead time. Predictor importance analyses show that at the 1-week lead time, the U-Net forecast skills are mainly derived from the synchronous precipitation forecasts. With the increasing lead times, the contributions from the atmospheric variables (especially those associated with moisture flux) rise rapidly. Therefore, the channel combining numerical weather prediction model and deep learning framework is demonstrated promising in S2S precipitation forecasts. Thus, combining numerical models and deep learning is very promising in subseasonal precipitation forecasts and can also be applied to the routine forecast of other atmospheric and ocean phenomena in the future.