WeatherBench 2 Operational GraphCast Forecasts Initialised from ECMWF HRES (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_hres_init_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 this dataset

Forecasts from the operational GraphCast configuration, the 13-pressure-level model fine-tuned on ECMWF HRES analyses and initialised from the HRES operational analysis (HRES t=0) rather than ERA5, as published for benchmarking in WeatherBench 2. This is the setup that matches real-time use, so these stores are the fair comparison against IFS HRES forecasts. Measured 2026-11-16: 2022-1440x721.zarr has 1,544 initialisation times from 2021-12-01 to 2022-12-21 at 6-hourly spacing, 40 lead times from 6 h to 240 h, a 0.25 degree 1440x721 grid and 13 pressure levels from 50 to 1000 hPa, about 25 TB uncompressed. Variables: geopotential, temperature, specific humidity, u and v wind, wind speed, vertical velocity, 2 m temperature, 10 m u and v wind, mean sea level pressure, and 6 h and 24 h accumulated precipitation. Six consolidated Zarr stores: 2020 and 2022, each at native resolution and regridded to 240x121 and 64x32. Open with xarray and the weatherbench2 evaluation code, scoring against HRES t=0. Licensed CC BY 4.0 by Google LLC (folder LICENSE); 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_hres_init_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_hres_init_v2/2020-1440x721.zarr/10m_u_component_of_wind/.zarray

Analyze this dataset without downloading it

On Scigantic, WeatherBench 2 Operational GraphCast Forecasts Initialised from ECMWF HRES (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