Rain Radar and Meteorological Parameters Fusion for Precipitation Nowcasting: Effect Exploration and Lightweight Model

Multi-source data can reflect the development process of a single meteorological element from different perspectives, and the combination of these data can provide a more comprehensive understanding of the current weather conditions and better predict future meteorological changes. In order to explore the influence of different elements on rainfall radar in precipitation nowcasting, we collected six different meteorological elements and fused them with rainfall radar to perform precipitation nowcasting. At the same time, considering the increase in model calculation caused by multiple data fusion and the gradient problem caused by model depth deepening, we developed a lightweight data fusion precipitation nowcasting model ReU-CA-Net model. Experimental results show that humidity, wind and temperature have more effective effects on rainfall radar, while total precipitation and dew point have weaker effects.