Why WeatherBench?

WeatherBench 2 is a framework for evaluating and comparing data-driven and traditional numerical weather forecasting models. WeatherBench consists of:

  • Publicly available, cloud-optimized ground truth and baseline datasets. For a complete list, see this page.

  • Open-source evaluation code. See this quick-start to explore the basic functionality or the API docs for more detail. Since high-resolution forecast files can be large, the WeatherBench 2 code was written with scalability in mind. See the command-line scripts based on Xarray-Beam and this guide for running the scripts on GCP using DataFlow.

  • A website displaying up-to-date scores of many of the state-of-the-art data-driven and physical approaches.

  • A paper describing the rationale behind the evaluation setup.

WeatherBench 2 has been built as an evolving tool for the entire community. For this reason, we welcome any feedback (ideally, submitted as GitHub issues) or contributions. If you would like your model to be part of WeatherBench, check out this guide.


To get started using the WeatherBench 2 code, first clone the GitHub repository:

git clone

Then, install using pip

cd weatherbench2
pip install .

If you would like to actively develop the code, install using

pip install -e .

If you plan to launch jobs on GCP, also install the GCP requirements:

pip install .[gcp]