Challenge

In a nutshell:

Since 2024, the Weather4cast competition at NeurIPS aims to improve rain 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. Besides keeping data fusion and super-resolution tasks,

This year, on the road to flexible foundation models, we move from basic precipitation prediction to testing generalization performance and emergent capabilities of probabilistic models on a set of downstream tasks, such as cumulative rainfall and unusual event prediction.

We’re also adding a pollution task (see bottom of page).

Competition tasks

To assess how well models develop emerging skills, in this year we introduce downstream tasks on which we will test the generalization performance of the models. In particular, we provide two application scenarios:

• Cumulative rainfall prediction (total rainfall in a given time span and  small area, like a city / urban region)
• Prediction of rare extreme weather event (event location and extent in time and space, event severity)

…where you will be provided with 1 hour of before the prediction timeframe (either satellite only or satellite + radar input will be provided).

For the model pre-training task we share with you a dataset including satellite data and rain rates from OPERA high resolution radar. The satellite data consist of 11-band spectral satellite images. These 11 channels show slightly noisy satellite radiances covering so-called visible (VIS), water vapor (WV), and infrared (IR) bands. Each satellite image covers a 15 minute period and its pixels correspond to a spatial area a few kilometer across (4 km at the equator, 12 km in the North). In addition, this year, for the first time, we also release the 2 km resolution rain rates for the whole context regions , and these high resolution rain data can also be used as an input in the model training phase. You can pre-train your model entirely as you like, including but not limited to

• translation, e.g.,  satellite input to radar output (same time)
• future prediction, e.g.,
  • future satellite from past satellite data
  • future radar from past radar input data
  • future radar from past satellite + radar data
  • future radar from past satellite data only
• spatial gap filling
• super-resolution

You can train at different spatial and temporal resolutions.

You can also incorporate any additional public data or model output, as long as you declare that and these are available to other scientists free of charge for academic / non-commercial use.

Downstream tasks are defined on the high-resolution 2 km grid.

Resources

To start playing with the data, you can use our Starting Kit from the previous competitions. Please note that it predicts OPERA rain rates for small region of interest form the satellite context data. This year, however, we will test the performance of your models on different downstream tasks, such as cumulative rainfall prediction and extreme precipitation event  prediction. We will not assess whole frame rain rate prediction directly as in past years.

To play with weather foundation models pre-trained by others,  you can consider building on Aurora, ClimaX, and TerraMind. Note that these are typically trained on global data and diverse input modalities, and will have to be adapted for regional forecasts and the challenge data provided here. You will find examples for fine-tuning these models for different tasks online and in the original literature.

Complementary Pollution Task

As a complementary task to the above rain based challenges, this year, for the first time, we offer a complementary challenge in forecasting atmospheric pollution. As there is no hidden test data, there is no point in a public leaderboard. Delegates are encouraged to demonstrate improvements over the Aurora Baseline by a metric of their choice and to submit an extended abstract describing their work.

• Aurora CAMS Pollution Baseline (with a worked Jupyter Notebook)
• CAMS Data

Your model can of course be entirely different from Aurora, you do not have to use it at all.

Benefits & Awards

The submissions will be assessed by the Weather4cast Scientific Committee, with the best contributions selected to present at the Weather4cast Workshop at the NeurIPS Competition Track.

Besides publishing your own work in the PMLR NeurIPS Competition Track Proceedings article, you will also be invited to publish a joint summary paper this year. Like last year, the NeurIPS organizers will provide free registrations for the leading submissions.