WeatherBench 2 GraphCast v2 Forecasts (ERA5-initialised, 2018, 2020 and 2022)

Provider
Google DeepMind / Google Research (WeatherBench 2)
License
CC-BY-4.0 (source)
Format
Zarr (in sampled files)
Stored at
Google Cloud Storage, gs://weatherbench2/datasets/graphcast_v2 (us)
Access
Free and public. No account with the data provider is needed.
DOI
10.1126/science.adi2336
Category
Climate & Weather
Source
https://doi.org/10.1126/science.adi2336

About WeatherBench 2 GraphCast v2 Forecasts (ERA5-initialised, 2018, 2020 and 2022)

Second release of the deterministic GraphCast medium-range forecasts in the WeatherBench 2 bucket, adding a 2022 test year to the 2018 and 2020 years of the first release. GraphCast (Lam et al., Science 2023) is a graph neural network trained on ERA5; each store holds 10-day forecasts initialised from ERA5 every 6 hours at 40 lead times from 6 h to 240 h. Measured 2026-11-16: 2022-1440x721.zarr spans initialisation times 2021-12-01 to 2022-12-30 (1,580 times) on the native 0.25 degree grid with 37 pressure levels (1 to 1000 hPa), about 70 TB uncompressed; the 2020 store spans 2019-12-01 to 2021-01-21. Variables: geopotential, temperature, specific humidity, u and v wind, wind speed, vertical velocity, 2 m temperature, 10 m wind, mean sea level pressure, and 6 h and 24 h total precipitation. Each year also comes regridded to 240x121 and 64x32 conservative grids. The folder holds nine consolidated Zarr stores (three years by three resolutions); open with xarray.open_zarr(..., storage_options={'token': 'anon'}) and score with the weatherbench2 package against ERA5. Licensed CC BY 4.0 by Google LLC (LICENSE file in the folder); cite Lam et al. 2023 and Rasp et al. 2024.

How to access this dataset in Python

import gcsfs
fs = gcsfs.GCSFileSystem(token='anon')
fs.ls('weatherbench2/datasets/graphcast_v2/', detail=False)[:10]

Lists the files with anonymous access. Runs on any machine, with no cloud account.

Data structure

File types in sampled files: Zarr

Files

Example file: datasets/graphcast_v2/2018-1440x721.zarr/10m_u_component_of_wind/.zarray

Analyze this dataset without downloading it

On Scigantic, WeatherBench 2 GraphCast v2 Forecasts (ERA5-initialised, 2018, 2020 and 2022) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Climate & Weather datasets