Weather4cast 2023
Super-Resolution Rain Movie Prediction under Spatio-temporal Shifts

  • Predict Super-Resolution Rain Movies
  • Prove Transfer Learning across Space and Time under Strong Shifts
  • Exploit Data Fusion to model Ground Radar and multi-band Satellite Images

The Weather4cast competition presented at NeurIPS 2023 focuses on topics of high impact and practical value for our society – predicting future weather and changes of our environment. Unusual weather is increasing all over the world, reflecting ongoing climate change, and affecting communities in agriculture, transport, public health and safety, etc.

Can you predict future rain patterns with modern machine learning algorithms? Apply spatio-temporal modelling to complex dynamic systems! Get access to unique large-scale data and demonstrate temporal and spatial transfer learning under strong distributional shifts! We provide a super-resolution challenge of high relevance to forecasting unusual local events, where you need to predict future weather as measured by ground-based hi-res rain radar weather stations. Exploit data fusion of high-resolution rain radar maps movies combined, large-scale multi-band satellite images, and static information like topology!

Winning models will advance key areas of methods research in machine learning, of relevance beyond the application domain. The authors of competitive submissions will be invited to present at the NeurIPS Competition Track (virtual) and contribute to a joint article in the NeurIPS Datasets and Benchmark Proceedings.

This year, besides data fusion, super-resolution, and generalization performance, we move from classification to zero-inflated regression. The 2023 challenge is to predict quantitative hi-res rain radar movies from lo-res multi-channel spectral satellite images in a highly unbalanced data scenario.

Sign up to the forums to stay in the loop!

Competition Winners

We are happy to announce the Winners of the Weather4cast 2023 Competition !

CORE LEADERBOARD

1st place:   team ALI_BDIL:  Xinzhe Li, Rui Sun, Yiming Niu, Yao Liu, Alibaba Cloud, China
2nd place: team FIT-CTU: Jiˇrí Pihrt, Petr Šimánek, Czech Technical University in Prague, Czech Republic
3rd place: team SandD: Lu Han, Xu-Yang Chen, Han-Jia Ye, De-Chuan Zhan, Nanjing University, China

 

NOWCASTING LEADERBOARD

1st place:  team ALI_BDIL: Xinzhe Li, Rui Sun, Yiming Niu, Yao Liu, Alibaba Cloud, China
2nd place: team SandD: Lu Han, Xu-Yang Chen, Han-Jia Ye, De-Chuan Zhan, Nanjing University, China
3rd place: team enrflo: Ajitabh Kumar, India

TRANSFER LEARNING LEADERBOARD

1st place:  team SandD: Lu Han, Xu-Yang Chen, Han-Jia Ye, De-Chuan Zhan, Nanjing University, China
2nd place: team ALI_BDIL: Xinzhe Li, Rui Sun, Yiming Niu, Yao Liu, Alibaba Cloud, China
3rd place: team enrflo: Ajitabh Kumar, India

Weather4cast session at NeurIPS 2023

The Weather4cast Competition session will take place virtually on December 15th at the NeurIPS 2023 Conference. Please note that the time below is in the New Orleans time zone (CST – Central Standard Timezone).

Link to the Weatehr4cast virtual Competition at NeurIPS 2023

9:00 – Introduction – David Kreil

9:05 – Current Challenges and ML Approaches for Earth Observations and Earth Sciences – Bertrand Le Saux, European Space Agency, Italy

9:35 – An Overview of Metrics Used for Weather Nowcasting  – Federico Serva, Institute of Marine Sciences (CNR-ISMAR), Italy

9:55 – Dataset Description – Aleksandra Gruca, Silesan University of Technology, Poland

10:05 –  Precipitation Prediction Using an Ensemble of Lightweight Learners, arXiv:2401.09424  – Team ALI_BIDL – Winners of the Core and Nowcasting Leaderboards

10:25 – Learning Robust Precipitation Forecaster by Temporal Frame Interpolation and Multi-Level Dice Loss, arXiv:2311.18341 – Team SandD – Winners of the Transfer Learning Leaderboard

10:45 –  RainAI – Precipitation Nowcasting from Satellite Data, arXiv:2311.18398 – Team rainai

11:05 – PAUNet: Precipitation Attention-based U-Net for rain prediction from satellite radiance data, arXiv:2311.18306  – Team CFG

11:25 –Skilful Precipitation Nowcasting Using NowcastNet, arXiv:2311.17961 – Team enrflo

11:30 – Precipitation Nowcasting With Spatial and Temporal Transfer Learning Using Swin-UNETR,  arXiv:2312.00258 – Team enrflo_t2

11:35 – Efficient Baseline for Quantitative Precipitation Forecasting in Weather4cast 2023, arXiv:2311.18806 – team Akpun

11:40 – Stochastic Atmosphere Posing New Challenges to AI/ML Nowcasting – Xavier Calbet, Spanish Meteorological Agency (AEMET), Spain

12:10 – Award Ceremony

12:15 – Closing Remarks

Competition timeline:

8 Oct.  – Dataset release  &  Start of competition announcement

8 Oct. – 26 Nov. – main competition

26 Nov. (midnight AoE) – Competition submission deadlineShort scientific papers (4-8 pages + references) need to be on arXiv.org and code & parameters on GitHub.com by then!

29 Nov. (midnight AoE) – Short scientific papers (4-8 pages + references) need to be on arXiv.org and code & parameters on GitHub.com by then!