GET Gridded Forecast

This endpoint returns gridded forecast output as NetCDF (default) or Zarr. Request a native or derived variable by name, or set variable to all to retrieve all native variables. Each file covers the model's full grid unless you crop it to a region with bbox.

The variable and level parameters are used to specify the variable and level to retrieve.

Use the model-specific URL below to choose which weather model's forecast to retrieve. You can find a list of available models here.

WeatherMesh-6

WeatherMesh-6 processes its 128 member-ensemble to output a number of summary statistics, including mean, standard deviation, percentiles, threshold probabilities, and precipitation-type probabilities. Use the variables endpoint to query which fields are available for each parameter. Use the include_distribution parameter to request these fields, include_members to get all 128 raw ensemble members, or include_deterministic to add the deterministic forecast alongside the mean. Use skip_mean to omit the ensemble mean from the output.

Archived Forecasts

For archived WeatherMesh-6 forecasts, native variables can be requested individually or together by setting variable to all. Derived variables are available only for non-archived forecasts.

Available Levels

For upper-level atmospheric variables, set the pressure level in hectopascals (hPa) using the level query parameter alongside the variable. To get the same list as JSON via the API, see the Variables endpoint

Available Variables

Native variables for each model and derived variables for WeatherMesh-6 are listed below. To get the same list as JSON via the API, use the Variables endpoint.

Query Parameters

Response

Returns a netCDF (.nc) or Zarr (.zarr.zip) file containing gridded forecast data for the specified variable. This will generally redirect to a presigned download URL, so you will need to follow the redirect to get the file. Use the -L flag in curl. Only in the short time it takes for the file to be uploaded will it be served directly.

JSON Response

With as_url=true, returns download metadata instead of the forecast file.

Edge Cases

  1. On wm-6-3km and hrrr, whose grid rows don't follow lines of latitude, a bbox file is the smallest block of the native grid that covers the box, and points in that block outside the box are NaN. For example, bbox=-125,32,-114,42 on hrrr returns a 437 × 404 grid in which about 31% of values are NaN.
  2. On ecmwf-det, aifs, and gfs, longitudes run from 0° to 360°, so a bbox that crosses the prime meridian returns longitudes that wrap from 360° to 0°. For example, bbox=-10,35,30,60 on ecmwf-det returns longitudes 350, 350.25, …, 359.75, 0, …, 30.

Use the Downloaded File

Run the first download for your model above, then run its Python plot in the same directory. For NetCDF, install xarray, netCDF4, and numpy. For zipped Zarr, install zarr>=3 and numpy. All plots use matplotlib. The coastline and state overlays also use cartopy, which downloads its Natural Earth map data on first use.

For a NetCDF file, you can also inspect the dimensions and variables with ncdump -h temperature_2m.nc from the NetCDF command-line tools.

Plot the Downloaded Temperature Field
Download a Gridded Forecast