ECMWF AIFS Single forecast, virtual

Name: ECMWF AIFS Single forecast, virtual
Creator: dynamical.org
License: https://creativecommons.org/licenses/by/4.0/

map-optimized · low-latency

Spatial domain	Global
Spatial resolution	0.25 degrees (~20km)
Time domain	Forecasts initialized 2024-04-01 00:00:00 UTC to Present
Time resolution	Forecasts initialized every 6 hours
Forecast domain	Forecast lead time 0-360 hours (0-15 days) ahead
Forecast resolution	6 hourly

STAC (browse) · validation report

The Artificial Intelligence Forecasting System (AIFS) is a data driven forecast model developed by the European Centre for Medium-Range Weather Forecasts (ECMWF). This is the non-ensemble configuration of AIFS that produces a single forecast trace. AIFS is trained on ECMWF's ERA5 re-analysis and ECMWF's operational numerical weather prediction (NWP) analyses.

This dataset is an archive of past and present ECMWF AIFS Single forecasts, optimized for spatial (map) access patterns. Forecasts are identified by an initialization time (init_time) denoting the start time of the model run, and step forward along the lead_time dimension from 0 to 360 hours (15 days) at a 6 hourly step.

Chunks reference the bytes of ECMWF's original GRIB files and are decoded on read, so this archive carries every variable ECMWF publishes for AIFS Single. Surface and single-level variables are at the dataset root; variables carried on pressure levels are in the pressure_level group.

Note: dynamical-catalog>=0.8.0 (or zarr>=3.2 icechunk>=2.0 gribberish>=1.5) is required.

Related Datasets

Examples

	Quickstart (Github)
	Quickstart (Colab)

import dynamical_catalog  # dynamical-catalog>=0.8.0

ds = dynamical_catalog.open("ecmwf-aifs-single-forecast-virtual", chunks=None)
ds["temperature_2m"].sel(init_time="2026-03-01T00", lead_time="24h")

# Variables with a vertical dimension live in the pressure_level group
ds_pressure = dynamical_catalog.open("ecmwf-aifs-single-forecast-virtual", group="pressure_level", chunks=None)
ds_pressure["geopotential_height"].sel(pressure_level=500)

import icechunk
import pystac
import xarray as xr

catalog = pystac.Catalog.from_file("https://stac.dynamical.org/catalog.json")
collection = catalog.get_child("ecmwf-aifs-single-forecast-virtual")
asset = collection.assets["icechunk-https"]

authorize = icechunk.containers_credentials({"s3://ecmwf-forecasts/": icechunk.s3_anonymous_credentials()})
repo = icechunk.Repository.open(
    icechunk.http_storage(asset.href),
    authorize_virtual_chunk_access=authorize,
)
session = repo.readonly_session("main")

ds = xr.open_zarr(session.store, chunks=None)
ds["temperature_2m"].sel(init_time="2026-03-01T00", lead_time="24h")

# Variables with a vertical dimension live in the pressure_level group
ds_pressure = xr.open_zarr(session.store, group="pressure_level", chunks=None)
ds_pressure["geopotential_height"].sel(pressure_level=500)

ECMWF AIFS Single forecast, virtual · Temperature map

Dimensions

	min	max	units
init_time	2024-04-01T00:00:00Z	Present	seconds since 1970-01-01

latitude	-90	90	degree_north

lead_time	0	1296000	seconds

longitude	-180	179.75	degree_east

pressure_level	10	1000	hPa

Variables

Access a variable by its name (the bold identifier, e.g. ds["convective_precipitation_run_total_surface"]).

Dimensions: init_time × lead_time × latitude × longitude

variable	units
convective_precipitation_run_total_surface Convective precipitation (`cp`)	kg m-2
dew_point_temperature_2m 2 metre dewpoint temperature (`2d`)	degree_Celsius
downward_long_wave_radiation_run_total_surface Surface long-wave (thermal) radiation downwards (`strd`)	W s m-2
downward_short_wave_radiation_run_total_surface Surface short-wave (solar) radiation downwards (`ssrd`)	W s m-2
geopotential_height_surface Geopotential height (`gh`)Surface (orography) geopotential height. Time-invariant field published at lead time 0 only.	m
high_cloud_cover High cloud cover (`hcc`)	percent
land_sea_mask_surface Land-sea mask (`lsm`)Fraction (0-1) of the grid box that is land. Time-invariant field published at lead time 0 only.	1
low_cloud_cover Low cloud cover (`lcc`)	percent
medium_cloud_cover Medium cloud cover (`mcc`)	percent
pressure_reduced_to_mean_sea_level Pressure reduced to MSL (`prmsl`)	Pa
pressure_surface Surface pressure (`sp`)	Pa
runoff_water_equivalent_run_total_surface Runoff water equivalent (surface plus subsurface) (`rowe`)Applies over land only. Water points are NaN from init time 2026-05-12T06:00 and unmasked before it.	kg m-2
skin_temperature_surface Skin temperature (`skt`)	degree_Celsius
slope_of_sub_gridscale_orography_surface Slope of sub-gridscale orography (`slor`)Time-invariant field published at lead time 0 only.	1
snow_area_fraction_surface Snow cover (`snowc`)Fraction (0-1) of the grid box covered by snow. Applies over land only; NaN over water.	1
snowfall_water_equivalent_run_total_surface Snowfall water equivalent (`sf`)	kg m-2
soil_temperature_layer_1 Soil temperature (`sot`)ECMWF soil level 1, the uppermost soil layer. Over water this is the sea surface temperature, not a soil temperature.	degree_Celsius
soil_temperature_layer_2 Soil temperature (`sot`)ECMWF soil level 2, the second soil layer from the surface. Over water this is the sea surface temperature, not a soil temperature.	degree_Celsius
standard_deviation_of_sub_gridscale_orography_surface Standard deviation of sub-gridscale orography (`sdor`)Time-invariant field published at lead time 0 only.	m
temperature_2m 2 metre temperature (`2t`)	degree_Celsius
total_cloud_cover_atmosphere Total cloud cover (`tcc`)	percent
total_column_water_atmosphere Total column water (`tcw`)	kg m-2
total_precipitation_run_total_surface Total precipitation (`tp`)	kg m-2
volumetric_soil_moisture_layer_1 Volumetric soil moisture (`vsw`)ECMWF soil level 1, the uppermost soil layer. Applies over land only. Water points are NaN from init time 2026-05-12T06:00 and unmasked before it.	1
volumetric_soil_moisture_layer_2 Volumetric soil moisture (`vsw`)ECMWF soil level 2, the second soil layer from the surface. Applies over land only. Water points are NaN from init time 2026-05-12T06:00 and unmasked before it.	1
wind_u_100m 100 metre U wind component (`100u`)	m s-1
wind_u_10m 10 metre U wind component (`10u`)	m s-1
wind_v_100m 100 metre V wind component (`100v`)	m s-1
wind_v_10m 10 metre V wind component (`10v`)	m s-1

Pressure Level (6 variables)

These variables live in the pressure_level Zarr group (e.g. pass group="pressure_level" to dynamical_catalog.open() or xr.open_zarr()).

Dimensions: init_time × lead_time × latitude × longitude × pressure_level

variable	units
geopotential_height Geopotential height (`gh`)	m
specific_humidity Specific humidity (`q`)The source provides no 10 hPa level for this variable; that level is always NaN.	1
temperature Temperature (`t`)	degree_Celsius
vertical_velocity Vertical velocity (`w`)	Pa s-1
wind_u U component of wind (`u`)	m s-1
wind_v V component of wind (`v`)	m s-1

Don't see what you're looking for? Let us know at feedback@dynamical.org.

Details

License

Dataset licensed under CC BY 4.0 and ECMWF Terms of Use.

Attribution and citation

ECMWF AIFS Single forecast data processed by dynamical.org from ECMWF Open Data.

Or ECMWF AIFS Single from dynamical.org.

Source

The source grib files this archive is constructed from are provided by ECMWF Open Data and accessed from the AWS Open Data Registry.

ECMWF does not provide user support for the free & open datasets. Users should refer to the public User Forum for any questions related to the source material.

Model updates

AIFS is updated regularly. Find details of recent and upcoming changes to the forecasting system on the ECMWF website.

Storage

Storage for this dataset is generously provided by AWS Open Data.

Chunks

This dataset is stored in Zarr format, which splits each variable into a grid of chunks — the smallest unit read from storage. When possible, aligning your reads with this dataset's chunk grid can significantly improve data access speed.

The element count and coordinate span of this dataset:

dimension	chunk
init_time	1 (6 hours)
lead_time	1 (6 hours)
latitude	721 (180.25°)
longitude	1440 (360°)
uncompressed	7.9 MiB

Validation report

Review the validation report to understand variable availability, missing data, known quirks, fill values, and approximate spatial, temporal, and value distributions.

Compression

The data values in this dataset have been rounded in their binary floating point representation to improve compression. See Klöwer et al. 2021 for more information on this approach. The exact number of rounded bits can be found in our reformatting code.