Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting

TLDR

Very few previous studies have examined this crucial and challenging weather forecasting problem from the machine learning perspective. The goal of precipitation nowcasting is to predict the future rainfall intensity in a local region over a relatively short period of time, formulated as a spatiotemporal sequence forecasting problem. By extending the fully connected LSTM to include convolutional structures in both input‑to‑state and state‑to‑state transitions, we propose the convolutional LSTM (ConvLSTM) and build an end‑to‑end trainable model for precipitation nowcasting. Experiments show that ConvLSTM captures spatiotemporal correlations better and consistently outperforms FC‑LSTM and the state‑of‑the‑art ROVER algorithm.

Abstract

The goal of precipitation nowcasting is to predict the future rainfall intensity in a local region over a relatively short period of time. Very few previous studies have examined this crucial and challenging weather forecasting problem from the machine learning perspective. In this paper, we formulate precipitation nowcasting as a spatiotemporal sequence forecasting problem in which both the input and the prediction target are spatiotemporal sequences. By extending the fully connected LSTM (FC-LSTM) to have convolutional structures in both the input-to-state and state-to-state transitions, we propose the convolutional LSTM (ConvLSTM) and use it to build an end-to-end trainable model for the precipitation nowcasting problem. Experiments show that our ConvLSTM network captures spatiotemporal correlations better and consistently outperforms FC-LSTM and the state-of-the-art operational ROVER algorithm for precipitation nowcasting.