# WeatherBench 2 WeatherNext 2 (FGN) Ensemble Forecasts (2022)

## Facts

- Provider: Google DeepMind (WeatherBench 2)
- License: CC-BY-4.0 (https://storage.googleapis.com/weatherbench2/datasets/fgn/LICENSE)
- Format: Zarr (in sampled files)
- Stored at: Google Cloud Storage, gs://weatherbench2/datasets/fgn (us)
- Access: public, readable without an account
- Category: Climate & Weather
- Source: https://arxiv.org/abs/2506.10772
- Page: https://scigantic.com/archive/dd97f5cb-8d3b-5a6b-af70-77fadf035650
- Keywords: Climate & Weather, weather forecasting, Google Cloud Storage

## About

Ensemble global weather forecasts from WeatherNext 2, Google DeepMind's Functional Generative Network (FGN) model described in 'Skillful joint probabilistic weather forecasting from marginals' (Alet et al., arXiv 2506.10772), published in the WeatherBench 2 bucket. Measured 2026-11-14: the 0.25 degree store (2022-1440x721.zarr) covers initialisation times every 6 hours through 2022 (1,460 times), 64 ensemble members and 60 lead times from 6 h to 360 h, at 500, 700 and 850 hPa plus surface fields. Variables include geopotential, temperature, specific humidity, u and v wind, wind speed, vertical velocity, 2 m temperature, 10 m wind and wind speed, mean sea level pressure, sea surface temperature and 6-hour and 24-hour total precipitation. The full ensemble is about 675 TB uncompressed; a 240x121 (1.5 degree) conservative regrid is about 19 TB and a 64x32 regrid is smaller, each with an ensemble-mean companion. Consolidated Zarr; open with xarray.open_zarr and compare with GenCast, IFS ENS and ERA5 using the weatherbench2 package. Licensed CC BY 4.0 by Google LLC; cite Alet et al. 2025 and WeatherBench 2.

## Load the data

```python
import gcsfs
fs = gcsfs.GCSFileSystem(token='anon')
fs.ls('weatherbench2/datasets/fgn/', 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

Example file: `datasets/fgn/2022-1440x721.zarr/10m_u_component_of_wind/.zarray`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/dd97f5cb-8d3b-5a6b-af70-77fadf035650
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/dd97f5cb-8d3b-5a6b-af70-77fadf035650
- Over MCP: `get_schema_card` with id `dd97f5cb-8d3b-5a6b-af70-77fadf035650` at https://api.scigantic.com/mcp

## Related topics

- [Climate & Weather datasets](https://scigantic.com/topics/climate-and-weather)
- [Weather forecasting datasets](https://scigantic.com/topics/weather-forecasting)
