The goal of the competition is a short-term prediction of selected weather products based on meteorological satellites data obtained in collaboration with AEMET/ NWC SAF. Following recent success of our Traffic4cast competitions at NeurIPS in 2019 and 2020, this challenge presents weather forecast as a video frame prediction task. The weather movies consist of multi-channel images encoding the cloud properties, temperature, turbulence, and rainfall. The images are recorded at 15 minute intervals through the entire year. Each pixel in the images represents the area of about 4 km x 4 km, and each region contains 256 x 256 pixels.  The regions span varying landscapes including mountains, deserts,  islands and seas, and others. The training and validation data contain observations in various earth regions and the test data include additional new regions. The challenge is to predict the encoded weather products in all regions. The competition outcomes are expected to go beyond methodological advances in weather forecasting and video frame prediction. The challenge offers real-world benchmark for few shot and transfer learning and allows testing multi-sensor data fusion. Learn more…