r.in.ecostress downloads NASA ECOSTRESS (ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station) thermal imagery for the current region and a date range, and optionally imports it into the current GRASS project or a new one. ECOSTRESS flies on the International Space Station, like EMIT, and its products are distributed the same way: the module searches NASA's Common Metadata Repository (CMR) and downloads from LP DAAC, as r.in.emit does.
It reads the tiled ECOSTRESS products: 70 m Cloud-Optimized GeoTIFFs on the Sentinel-2 MGRS tiling grid (110 km tiles in their UTM zone), one file per layer. Only the requested layers are downloaded. The swath (L1B, L2) and gridded (L1CG, L2G) HDF5 products are not supported.
A free NASA Earthdata
account and Bearer token are needed (Earthdata login, User
Profile, Generate Token). The token is read from
token=, from the EARTHDATA_TOKEN
environment variable, or from a file named TOKEN in
output_dir.
g.region n=5640000 s=5630000 e=580000 w=570000 res=70 -a
r.in.ecostress start=2025-07-01 end=2025-07-31 daynight=day \
output=be clouds=50 coverage=10 strds=be -iproduct= selects the collection; layers= the layers to fetch, named as in the granule files. The default layers are:
| product | collection | default layers |
|---|---|---|
| lste | ECO_L2T_LSTE | LST, LST_err, QC, cloud |
| rad | ECO_L1CT_RAD | radiance_1 ... radiance_5 |
| jet | ECO_L3T_JET | ETdaily |
| etaux | ECO_L3T_ETAUX (V003) | all |
| esi | ECO_L4T_ESI | ESI, PET |
| wue | ECO_L4T_WUE | WUE, GPP |
| stars | ECO_L2T_STARS | NDVI, albedo |
| sm | ECO_L3T_SM (V002) | SM |
| met | ECO_L3T_MET (V002) | Ta, RH |
| seb | ECO_L3T_SEB (V002) | Rn, Rg |
| et_alexi | ECO_L3T_ET_ALEXI (V002, CONUS) | ETdaily, ETdailyUncertainty |
| esi_alexi | ECO_L4T_ESI_ALEXI (V002, CONUS) | ESIdaily, ESIdailyUncertainty |
layers=all fetches every layer of the granules. Use -l to see the layers available. A requested layer that no granule holds stops the module with an error, listing the available layers; a layer missing from some acquisitions only (for example because V002 and V003 differ) gives a warning.
The tiled layers hold physical values, with no scale factor: LST in kelvin, with NaN (null) under clouds and outside the swath; the cloud and water masks are 0/1. QC is the bit-packed quality word described in the ECOSTRESS L2 user guide.
The ISS orbit is not sun-synchronous: ECOSTRESS sees a place at varying times of day and night, often several times a day. daynight= keeps day or night overpasses only.
One overpass of the region is one ISS orbit. Its swath is cut into scenes of about 52 seconds, so a region can be covered by several tiles, and by two consecutive scenes on the same tile. All the granules of one orbit are therefore grouped: each layer is imported once per overpass, with the tiles and scenes mosaicked with r.patch.
ECOSTRESS Collection 2 (V002) covers the whole mission; V003 is being produced and covers part of it. By default the newest version of each tile and acquisition is kept, so a time series can mix versions (the module warns when it does). version= restricts the search to one version. Duplicate granules of the same tile and time (other scene numbers or builds of the same data) are dropped the same way.
The search box is the current region, reprojected to WGS84 with
g.region -b. CMR matches each tile on its bounding box;
tiles that straddle the antimeridian get a box covering every longitude
and would match any search, so their longitudes are checked against
their UTM zone instead.
With -i, each layer is imported with r.import using extent=region and resolution=region: the maps are reprojected onto the current region grid, so every map of a time series is aligned. resample= sets the method for continuous layers; QC, masks and flags always use nearest neighbour.
Map names are {output}_{layer}_{acquisition time}, with
the layer in lower case, e.g. be_lst_20250725T164442 and
be_lst_err_20250725T164442. Maps that already exist are
kept unless --overwrite is given.
Each map gets its acquisition time (UTC, the start of the first scene
of the overpass) with r.timestamp, its units (K
for LST), and the granule titles and day/night flag in its metadata
(r.info).
With strds=, a space-time raster dataset is created
per layer, named {strds}_{layer}, e.g. be_lst,
and the maps are registered with their timestamps.
CMR has no cloud cover for ECOSTRESS. With -i, clouds= and coverage= measure each overpass over the region instead: the cloud mask and the first requested data layer are imported first. A cell is observed if it is flagged cloudy or the data layer has a value there (outside the swath, the cloud mask reads 0 and the data is null).
An overpass that fails is removed and its other layers are not downloaded. The cloud mask is removed afterwards unless it was among the requested layers.
Layers are saved under output_dir (default
$HOME/RSDATA/ECOSTRESS_{output}), in one directory per
year, with the LP DAAC file names. Files already present are not
downloaded again unless --overwrite is given. A layer
tile is a few MB.
project= creates (or reuses) a project under dbase= and imports there. Its CRS is the UTM zone of the first tile unless epsg= is given, and its region is set to the search area at 70 m. The search area always comes from the calling session's region.
List the night overpasses of a month, with their tiles and layers:
r.in.ecostress start=2025-07-01 end=2025-07-31 daynight=night -lImport day-time LST with its error and quality layers, keeping overpasses that see at least half of the region with at most 20% cloud, into a time series:
r.in.ecostress start=2025-06-01 end=2025-08-31 daynight=day \
layers=LST,LST_err,QC,cloud coverage=50 clouds=20 output=be strds=be -i
t.rast.list be_lstDaily evapotranspiration of the same overpasses:
r.in.ecostress product=jet start=2025-06-01 end=2025-08-31 \
coverage=50 clouds=20 output=be strds=be_et -iTop-of-atmosphere radiance of the five thermal bands, V002 only:
r.in.ecostress product=rad version=002 start=2025-07-25 end=2025-07-25 -ig.region, r.import, r.in.emit, r.in.landsat, r.in.sentinel, r.patch, r.timestamp, t.create, t.register
Yann Chemin
Available at: r.in.ecostress source code (history)
Accessed: Sunday Nov 15 11:07:38 2026
Main index | Raster index | Topics index | Keywords index | Graphical index | Full index
© 2003-2026 GRASS Development Team, GRASS 8.5.1dev Reference Manual