CONVCAST: AN EMBEDDED CONVOLUTIONAL LSTM BASED ARCHITECTURE FOR PRECIPITATION NOWCASTING USING SATELLITE DATA

Nowcasting of precipitation is a difficult spatiotemporal task because of the non-uniform characterizationof meteorological structures over time. Recently, convolutional LSTM has been shown to besuccessful in solving various complex spatiotemporal based problems. In this research, we proposea novel precipitation nowcasting architecture ‘Convcast’ to predict various short-term precipitationevents using satellite data. We train Convcast with ten consecutive NASA’s IMERG precipitation datasets each at intervals of 30 minutes. We use the trained neural network model to predict the eleventhprecipitation data of the corresponding ten precipitation sequence. Subsequently, the predicted precipitationdata are used iteratively for precipitation nowcasting of up to 150 minutes lead time. Convcastachieves an overall accuracy of 0.93 with an RMSE of 0.805 mm/h for 30 minutes lead time, andan overall accuracy of 0.87 with an RMSE of 1.389 mm/h for 150 minutes lead time. Experimentson the test dataset demonstrate that Convcast consistently outperforms other state-of-the-art opticalflow based nowcasting algorithms. Results from this research can be used for nowcasting of weatherevents from satellite data as well as for future on-board processing of precipitation data.