Global Precipitation Nowcasting of Integrated Multi-satellitE Retrievals for GPM: A U-Net Convolutional LSTM Architecture

This paper presents a deep supervised learning architecture for 30-min global precipitation nowcasts with a 4-h lead time. The architecture follows a U-Net structure with convolutional long short-term memory (ConvLSTM) cells empowered by ConvLSTM-based skip connections to reduce information loss due to the pooling operation. The training uses data from the Integrated Multi-satellitE Retrievals for GPM (IMERG) and a few key drivers of precipitation from the Global Forecast System (GFS). The impacts of different training loss functions, including the mean-squared error (regression) and the focal loss (classification), on the quality of precipitation nowcasts are studied. The results indicate that the regression network performs well in capturing light precipitation (,1.6 mm h²¹), while the classification network can outperform the regression counterpart for nowcasting of high-intensity precipitation (.8 mm h²¹), in terms of the critical success index (CSI). It is uncovered that including the forecast variables can improve precipitation nowcasting, especially at longer lead times in both networks. Taking IMERG as a relative reference, a multiscale analysis, in terms of fractions skill score (FSS), shows that the nowcasting machine remains skillful for precipitation rate above 1 mm h²¹ at the resolution of 10 km compared to 50 km for GFS. For precipitation rates greater than 4 mm h²¹, only the classification network remains FSS skillful on scales greater than 50 km within a 2-h lead time.