Efficient spatio-temporal weather forecasting using U-Net
This work addresses the need for efficient weather prediction to reduce computational costs, but it appears incremental as it builds on existing U-Net methods without claiming major breakthroughs.
The paper tackled the problem of spatio-temporal weather forecasting by applying SmaAt-UNet, an efficient U-Net-based autoencoder, to the Weather4cast 2021 Challenge, achieving competent results with low computational resources.
Weather forecast plays an essential role in multiple aspects of the daily life of human beings. Currently, physics based numerical weather prediction is used to predict the weather and requires enormous amount of computational resources. In recent years, deep learning based models have seen wide success in many weather-prediction related tasks. In this paper we describe our experiments for the Weather4cast 2021 Challenge, where 8 hours of spatio-temporal weather data is predicted based on an initial one hour of spatio-temporal data. We focus on SmaAt-UNet, an efficient U-Net based autoencoder. With this model we achieve competent results whilst maintaining low computational resources. Furthermore, several approaches and possible future work is discussed at the end of the paper.