Weather4cast at NeurIPS 2025 – Multi-task Challenges for Weather and Pollution Pattern Prediction on the Road to Hi-Res Foundation Models

Aleksandra Gruca, Xavier Calbet, Pilar Rípodas, Llorenç Lliso, Federico Serva, Levi Harris, Tianlong Chen, Luciano Araujo Dourado Filho, Almir Moreira da Silva Neto, Anthony Miyaguchi, Rodrigo Pereira David, Rodrigo Tripodi Calumby, Lukas Picek, Kaushik Gopalan, Atharva Deshpande, Jeet Shah, Hrishikesh Simu, Ansh Kushwaha, Anushree Bhuskute, David P. Kreil
Proceedings of the NeurIPS 2023-2025 Competition Track, PMLR 276:106-122, 2026.

Abstract

The Weather4cast competition series aims to advance AI and machine learning algorithms through a highly topical interdisciplinary challenge (www.weather4cast.net). The core task involves predicting high-resolution rain radar movies from multi-band satellite sensors, which requires the application of data fusion across complementary signal sources, multi-channel video frame prediction, and super-resolution techniques. To evaluate the models’ ability to extract relevant mechanistic patterns, the competition incorporates spatio-temporal shifts, evaluating predictions both within and across diverse Earth regions with varying climates and precipitation distributions. Now, in its fifth year, Weather4cast moves to improve forecasts world-wide on an expansive data set with over a magnitude more hi-res rain radar data, allowing a move towards Foundation Models through multi-modality, multi-scale, multi-task challenges. Accurate rain predictions are becoming ever more critical for everyone, with climate change increasing the frequency of extreme precipitation events. Notably, the new models and insights will have a particular impact for the many regions on Earth where costly weather radar data are not available. As a complementary application-specific forecasting endpoint, in 2025, for the first time, we added a pollution forecasting challenge.

Cite this Paper


BibTeX
@InProceedings{pmlr-v276-gruca26b, title = {Weather4cast at {NeurIPS} 2025 – Multi-task Challenges for Weather and Pollution Pattern Prediction on the Road to Hi-Res Foundation Models}, author = {Gruca, Aleksandra and Calbet, Xavier and R{\'i}podas, Pilar and Lliso, Lloren{\c c} and Serva, Federico and Harris, Levi and Chen, Tianlong and Dourado Filho, Luciano Araujo and da Silva Neto, Almir Moreira and Miyaguchi, Anthony and David, Rodrigo Pereira and Calumby, Rodrigo Tripodi and Picek, Lukas and Gopalan, Kaushik and Deshpande, Atharva and Shah, Jeet and Simu, Hrishikesh and Kushwaha, Ansh and Bhuskute, Anushree and Kreil, David P.}, booktitle = {Proceedings of the NeurIPS 2023-2025 Competition Track}, pages = {106--122}, year = {2026}, editor = {Qin, Tao and Zhang, Kun and Frellsen, Jes}, volume = {276}, series = {Proceedings of Machine Learning Research}, month = {02--07 Dec}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v276/main/assets/gruca26b/gruca26b.pdf}, url = {https://proceedings.mlr.press/v276/gruca26b.html}, abstract = {The Weather4cast competition series aims to advance AI and machine learning algorithms through a highly topical interdisciplinary challenge (www.weather4cast.net). The core task involves predicting high-resolution rain radar movies from multi-band satellite sensors, which requires the application of data fusion across complementary signal sources, multi-channel video frame prediction, and super-resolution techniques. To evaluate the models’ ability to extract relevant mechanistic patterns, the competition incorporates spatio-temporal shifts, evaluating predictions both within and across diverse Earth regions with varying climates and precipitation distributions. Now, in its fifth year, Weather4cast moves to improve forecasts world-wide on an expansive data set with over a magnitude more hi-res rain radar data, allowing a move towards Foundation Models through multi-modality, multi-scale, multi-task challenges. Accurate rain predictions are becoming ever more critical for everyone, with climate change increasing the frequency of extreme precipitation events. Notably, the new models and insights will have a particular impact for the many regions on Earth where costly weather radar data are not available. As a complementary application-specific forecasting endpoint, in 2025, for the first time, we added a pollution forecasting challenge.} }
Endnote
%0 Conference Paper %T Weather4cast at NeurIPS 2025 – Multi-task Challenges for Weather and Pollution Pattern Prediction on the Road to Hi-Res Foundation Models %A Aleksandra Gruca %A Xavier Calbet %A Pilar Rípodas %A Llorenç Lliso %A Federico Serva %A Levi Harris %A Tianlong Chen %A Luciano Araujo Dourado Filho %A Almir Moreira da Silva Neto %A Anthony Miyaguchi %A Rodrigo Pereira David %A Rodrigo Tripodi Calumby %A Lukas Picek %A Kaushik Gopalan %A Atharva Deshpande %A Jeet Shah %A Hrishikesh Simu %A Ansh Kushwaha %A Anushree Bhuskute %A David P. Kreil %B Proceedings of the NeurIPS 2023-2025 Competition Track %C Proceedings of Machine Learning Research %D 2026 %E Tao Qin %E Kun Zhang %E Jes Frellsen %F pmlr-v276-gruca26b %I PMLR %P 106--122 %U https://proceedings.mlr.press/v276/gruca26b.html %V 276 %X The Weather4cast competition series aims to advance AI and machine learning algorithms through a highly topical interdisciplinary challenge (www.weather4cast.net). The core task involves predicting high-resolution rain radar movies from multi-band satellite sensors, which requires the application of data fusion across complementary signal sources, multi-channel video frame prediction, and super-resolution techniques. To evaluate the models’ ability to extract relevant mechanistic patterns, the competition incorporates spatio-temporal shifts, evaluating predictions both within and across diverse Earth regions with varying climates and precipitation distributions. Now, in its fifth year, Weather4cast moves to improve forecasts world-wide on an expansive data set with over a magnitude more hi-res rain radar data, allowing a move towards Foundation Models through multi-modality, multi-scale, multi-task challenges. Accurate rain predictions are becoming ever more critical for everyone, with climate change increasing the frequency of extreme precipitation events. Notably, the new models and insights will have a particular impact for the many regions on Earth where costly weather radar data are not available. As a complementary application-specific forecasting endpoint, in 2025, for the first time, we added a pollution forecasting challenge.
APA
Gruca, A., Calbet, X., Rípodas, P., Lliso, L., Serva, F., Harris, L., Chen, T., Dourado Filho, L.A., da Silva Neto, A.M., Miyaguchi, A., David, R.P., Calumby, R.T., Picek, L., Gopalan, K., Deshpande, A., Shah, J., Simu, H., Kushwaha, A., Bhuskute, A. & Kreil, D.P.. (2026). Weather4cast at NeurIPS 2025 – Multi-task Challenges for Weather and Pollution Pattern Prediction on the Road to Hi-Res Foundation Models. Proceedings of the NeurIPS 2023-2025 Competition Track, in Proceedings of Machine Learning Research 276:106-122 Available from https://proceedings.mlr.press/v276/gruca26b.html.

Related Material