dataset · Global
ECMWF IFS Analysis N320 - Beyond Weather Icechunk Zarr
The ECMWF Integrated Forecasting System (IFS) operational analysis, every 6 hours from 2011 to 2024, on the native N320 reduced Gaussian grid (542,080 points, about 31 km). It contains a selection of variables: 6 upper-air variables on 18 pressure levels, 18 surface variables, and the land-sea mask and orography of each IFS cycle. This is the input set of data-driven weather models such as ECMWF's AIFS, for training, fine-tuning and verification. Stored as versioned Icechunk Zarr, chunked by tim
Visit source ↗Source details
Licence: [Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/), following the [ECMWF terms of use](https://apps.ecmwf.int/datasets/licences/general/). Copyr
Evidence quality: 41
Listed in aws-odr public catalogue API on 2026-10-16; relevance/value screened (haiku).
Metadata checked: 2026-10-14T13:20:00.000Z
What's in it
- Data offered docs-verified
- 6-hourly ECMWF IFS operational analysis · 6 upper-air variables on 18 pressure levels · 18 surface variables · land-sea mask · orography
- Coverage docs-verified
- Global · 2011-01-01 to 2024-12-31
- Resolution docs-verified
- Spatial: N320 reduced Gaussian grid (542,080 points, about 31 km). Temporal: 6-hourly.
- Update frequency docs-verified
- Not currently updated
- Formats docs-verified
- Icechunk Zarr v3
- Access docs-verified
- Free and open under CC BY 4.0, following the ECMWF terms of use. No AWS account required; anonymous S3 access via aws s3 --no-sign-request.
Useful for
- Use cases llm-inferred
- Training or verifying a local data-driven weather model (AIFS-style) for site-level meteorological forecasting · Reconstructing atmospheric conditions for back-analysis of historical storm or flood events · Building ML emulators of weather inputs for scenario testing in drainage or hydrological studies
- Audience llm-inferred
- hydrologists · flood risk consultants · drainage engineers · ML and data scientists · GIS analysts
Limitations
- Constraints llm-inferred
- Static archive: not currently updated and ends 2024-12-31 · Reduced Gaussian grid at about 31 km and a selected subset of variables and levels, not the full IFS output · Modified by Beyond Weather (variable and level selection, GRIB-to-Zarr conversion, lossy bit rounding), not the original ECMWF data · Analysis fields are not a rainfall observation product · ECMWF accepts no liability for errors, omissions or availability · Needs Zarr/Icechunk tooling (e.g. xarray, icechunk); N320 data usually needs regridding for standard GIS tools
Evidence
- arn:aws:s3:::beyond-weather-ecmwf-ifs-analysis, arn:aws:sns:us-east-2:239830941214:beyond-weather-ecmwf-ifs-analysis-object_created
- https://registry.opendata.aws/beyond-weather-ecmwf-ifs-analysis/
Check upstream coverage, uncertainty and licence before use. Station observations are not forecasts or design levels.