airmosaic-ai

ERA5 acquisition

English 中文

Use the CDS form or official Python client. This skill has no executable tool ID. Select variables and averaging with the ERA5 method.

Configure and acquire

  1. Choose single levels, pressure levels, ERA5-Land or daily statistics explicitly; fix variables, levels, dates, UTC times and north/west/south/east bounds.
  2. The user signs into CDS, accepts applicable dataset terms and prepares official-client credentials. The client may internally use an explicitly authorized existing .cdsapirc; do not inspect or copy it. Hidden token input remains an option in the existing first-use guide.
  3. For the website route, submit the reviewed form and save its output to a fresh task folder. For API code and non-secret request configuration, use official-client operations. Confirm the request against the current product form before submitting.
  4. Start with a bounded request; check its existing server state after interruption before resubmitting. Expand date/variable/level batches after the first content check.

Check and deliver

Decode GRIB with ecCodes or NetCDF with a compatible reader. Check parameter identity, level, units, UTC/valid times, grid, finite values and missing time steps against the request. File size and download completion alone are insufficient.

The 2026-10-26 single-level example checked one 2m_temperature field for 2024-01-01 00:00 UTC: K, 0.25°, 21×21 points, no missing values. It is a historical content check, not a daily statistic or independent accuracy evaluation.

Keep non-secret requests in AIRMOSAIC_CONFIG_DIR/copernicus/ and new downloads in AIRMOSAIC_ASSETS_DIR/results/<project_id>/<run_id>/01_socioeconomic_natural_change/era5/. Reusable versions use data/01_socioeconomic_natural_change/ERA5/<version>/ after review. Keep configuration, assets and authentication roots separate outside Git; exclude tokens and signed URLs from records. ERA5 files require separate adaptation to GEOS-Chem input contracts.