Deep Learning Improves GFS Wintertime Precipitation Forecast Over Southeastern China

Wintertime precipitation, especially snowstorms, significantly impacts people's lives. However, the current forecast skill of wintertime precipitation is still low. Based on data augmentation (DA) and deep learning, we propose a DABU-Net which improves the Global Forecast System wintertime precipitation forecast over southeastern China. We build three independent models for the forecast lead times of 24, 48, and 72 hr, respectively. After using DABU-Net, the mean Root Mean Squared Errors (RMSEs) of the wintertime precipitation at the three lead times are reduced by 19.08%, 25.00%, and 22.37%, respectively. The threat scores (TS) are all significantly increased at the thresholds of 1, 5, 10, 15, and 20 mm day(-1) for the three lead times. During heavy precipitation days, the RMSEs are decreased by 14% and TS are increased by 7% at the lead times within 48 hr. Therefore, combining DA and deep learning has great prospects in precipitation forecasting.

Automated summary

Wintertime precipitation, especially snowstorms, has a significant impact on people's lives, but the current forecast skill for wintertime precipitation is still low. By combining data augmentation (DA) and deep learning, a DABU-Net model is proposed to improve the forecast accuracy of wintertime precipitation in southeastern China. Three independent models are built for forecast lead times of 24, 48, and 72 hours, and the DABU-Net model reduces the root mean squared errors (RMSEs) of wintertime precipitation at these lead times by 19.08%, 25.00%, and 22.37% respectively. The threat scores (TS) are also significantly increased at various thresholds for the three lead times. Moreover, during heavy precipitation days, the RMSEs are decreased by 14% and TS is increased by 7% within 48 hours lead time. This highlights the great prospects of combining DA and deep learning in precipitation forecasting.