EUMETSAT Data Store — introduction#
earthlens.eumetsat is one unified backend over the EUMETSAT Data
Store, the access point for EUMETSAT's satellite archive. A single
OAuth2 consumer key / secret unlocks every collection the Data Store
serves. The catalog curates the entire Data Store — all ~180
collections — grouped into ten mission families:
| Group | Coverage |
|---|---|
| MTG | Meteosat Third Generation — FCI imager, Lightning Imager, FCI L2 suite |
| MSG | Meteosat Second Generation — SEVIRI L1.5, cloud products, CDRs |
| MFG | Meteosat First Generation — MVIRI / GSA climate data records |
| Metop | EPS — ASCAT scatterometer, IASI sounder, AVHRR / GOME-2 / AMSU / MHS, CDR/FDR |
| Metop-SG | Metop Second Generation (GRAS-2, auxiliary, BUFR) |
| Sentinel-3 | OLCI / SLSTR / SRAL marine mirror (NRT + reprocessed baselines) |
| Sentinel-5P | TROPOMI atmospheric composition mirror |
| Sentinel-6 | Poseidon-4 altimetry + microwave radiometer mirror |
| OSI-SAF | Ocean & Sea Ice — sea-ice concentration, SST |
| Other | Multimission / legacy climate & fundamental data records (ATMS, MWHS, SSM/T, HIRS, CM SAF, SARAH, AWS, RO) |
The backend wraps the official
eumdac client: one
consumer-key/secret pair mints an auto-refreshing bearer token, and every
collection is searched by bounding box + time window and fetched as a
whole native product to disk.
Output kind#
Like the NASA Earthdata backend, EUMETSAT's OUTPUT_KIND is
per-collection, not fixed. Each catalog row carries an
output_kind (raster / vector / tabular) that the backend copies
onto the instance at construction. Most collections are gridded
(raster); the point/vector products — Atmospheric Motion Vectors,
ASCAT winds, and the Lightning Imager event/flash/group products — are
tagged vector. A request that mixes output kinds is rejected up front.
How it maps onto the EarthLens facade#
from earthlens.earthlens import EarthLens
el = EarthLens(
data_source="eumetsat",
start="2024-06-01",
end="2024-06-01",
variables={"msg-hrseviri": ["HRSEVIRI"]},
lat_lim=[0.0, 10.0],
lon_lim=[0.0, 10.0],
path="eumetsat_output",
)
paths = el.download() # native products on disk
variables is a {collection_key: [selector, ...]} mapping (see
Usage); the bbox is a search filter (which products
intersect it), not a pixel crop.
Installation#
The backend's SDK is an optional extra. The package imports without it;
the eumdac import is lazy and only happens when you actually
authenticate or download.
Capabilities#
The backend fetches native products by default, and supports EUMETSAT Data Tailor for server-side customisation:
- Data Tailor (server-side subset / reproject / reformat) — the
tailor=TailorConfig(...)knob. Returns the customised GeoTIFF / NetCDF. See Data Tailor. (aggregate=is the separate, unimplemented temporal reducer and raisesNotImplementedError.) - Native SEVIRI / FCI client-side reading — needs a satpy reader
bridge in
pyramids(a deferred follow-on); tailoring to GeoTIFF sidesteps it. NetCDF mirror products (Sentinel-3 / -5P / -6, OSI SAF) are readable withpyramidstoday.
See Authentication, Usage, and the Catalog & tooling reference for the full picture.