Supported Datasets¶

Datasets available for extraction (configured independently in iac-metadata’s datasets.yaml for metadata generation, and in this repo’s path_constructor.py for extraction):

Dataset	`files_query_options.dataset`	Scenarios
ERA5 CDS	`era5_cds`	—
ERA5-Land	`era5-land_cds`	—
CERRA	`cerra`	—
CERRA-Land	`cerra-land`	—
CMIP5	`cmip5`	historical, rcp26, rcp45, rcp60, rcp85
CMIP6	`cmip6`	historical, ssp119, ssp126, ssp245, ssp370, ssp434, ssp460, ssp585
CMIP6-NG	—	historical + SSPs
CORDEX EUR-11	`cordex`	historical, rcp26, rcp45, rcp85
CORDEX-REKLIES EUR-11	`cordex-reklies`	historical, rcp26, rcp85

The exact field values expected for each dataset (variable, gcm, rcm, ensemble_member, …) are documented per-dataset in Request JSON Format.

Grid Type Handling¶

Datasets are stored on three different native grids. Zephyr detects the grid type automatically (src/zephyr/fileload.py) and, for the two non-regular cases, reprojects to regular latitude/longitude coordinates before extraction — so extraction_query_options always uses plain lat/lon regardless of dataset:

Regular lat/lon

Used by ERA5, CMIP5, CMIP6, CMIP6-NG. No transformation needed.

Rotated lat/lon

Used by CORDEX datasets. Zephyr unrotates the grid (src/zephyr/unrotation.py) and assigns regular lat/lon coordinates before extraction.

Lambert Conformal Conic (LCC)

Used by CERRA and CERRA-Land. Zephyr transforms the projection coordinates to regular lat/lon (src/zephyr/transform_lcc.py) before extraction.

Extraction Methods¶

A request specifies one of two extraction methods — see Request JSON Format for the exact JSON:

Nearest grid point — extracts the time series of the grid cell closest to a given latitude/longitude coordinate.
Rectangular clipping — clips a spatial bounding box defined by min/max latitude and longitude.