r.in.emit downloads NASA EMIT (Earth Surface Mineral Dust Source Investigation) L2A Estimated Surface Reflectance granules directly into the current GRASS GIS mapset, or a brand new GRASS project/location, by querying NASA’s Common Metadata Repository (CMR) and fetching matching granules straight from LP DAAC (no STAC cube service is involved — EMIT granules are downloaded whole, as delivered, not cropped to the region like r.in.sentinel’s cubo-based Sentinel-2 cubes).
Instead of reflectance, it can also fetch EMIT L1B at-sensor radiance (-r) and the EMIT L2B greenhouse gas products for methane (-m) and carbon dioxide (-c).
Requires a free NASA Earthdata account and a Bearer token (Earthdata login → User Profile → Generate Token).
g.region n=25.34 s=25.30 e=55.72 w=55.65
r.in.emit start=2024-01-01 end=2024-12-31 clouds=20 \
output=dubai -i strds=dubai_tsBy default, downloaded granules (and, with -i, imported raster_3d cubes) go straight into the GRASS project/mapset r.in.emit was launched from — exactly like every other r.in.*/i.in.* importer, and consistent with r.in.sentinel. No extra options are needed for this, the module’s default operation.
Give project= to create a brand new GRASS location (under dbase=, default the current GISDBASE) and import there instead, leaving the calling session’s own project untouched. EMIT L2A reflectance is delivered already orthorectified onto a WGS84 geographic grid via its bundled Geolocation Lookup Table (GLT), so the new project defaults to epsg=4326 — override only if you have a specific reason to reproject immediately. If the named project already exists, it is reused (not recreated) so repeated runs accumulate into the same project. After the first successful import, the new project’s default region is set from that cube so it starts in a sane state.
Each granule becomes one file (and, with -i, one raster_3d map) named {output}_{acquisition timestamp}, e.g. emit_20240619T081558 — using the full acquisition timestamp (not just the date) since a region can have more than one EMIT overpass on the same calendar day, unlike Sentinel-2’s one-scene-per-tile-per-day cadence that r.in.sentinel mosaics.
The module always reads the current computational region (g.region) of the session it was launched from to determine the CMR search bounding box — reprojected to WGS84 automatically via g.region -b, so this works identically whether the calling project is itself geographic or projected. This is true regardless of project=: the search area always comes from the calling session’s own region, only the output location differs.
-m (methane, CH4) and -c (carbon dioxide, CO2) switch the module from reflectance to the EMIT L2B greenhouse gas collections. Both flags can be given together. gas_products= selects which products are fetched for each gas (both by default):
EMITL2BCH4ENH, EMITL2BCO2ENH), holding three single-band GeoTIFFs: the enhancement itself (ENH, in ppm m), its sensitivity (SENS) and its uncertainty (UNCERT).EMITL2BCH4PLM, EMITL2BCO2PLM), holding a single-band GeoTIFF of the plume and a GeoJSON outline (PLMMETA) carrying the plume attributes (plume ID, maximum concentration and its location, and the emission rate estimate with its uncertainty when one was made).short_name= is ignored with -m/-c. Gas files are small (a few tens of MB per enhancement scene, tens of kB per plume) compared with reflectance granules.
clouds= filters enhancement scenes only. A plume complex exists only where the retrieval already found enough clear pixels to detect it, and its CMR cloud cover describes the whole scene rather than the plume.
With -i, every GeoTIFF is imported with r.import as an ordinary 2D raster map, not a raster_3d cube, since each holds a single band. Enhancement layers span the whole ~75 km swath and are cropped to the current region (extent=region); plume rasters are kept whole so that no plume is cut at the region edge. Each plume outline is imported with v.import as a vector map, with the plume attributes in its attribute table. Rasters get their acquisition time with r.timestamp and plume vectors with v.timestamp. Map names are {output}_{layer}_{acquisition timestamp}, where the layer is ch4enh, ch4sens, ch4uncert or ch4plm (co2... for carbon dioxide). Plume maps also end with the plume ID, since one overpass can detect several plumes, e.g. uae_ch4plm_20240825T064423_003501; a plume’s raster and vector maps share the same name.
With strds=, one space-time raster dataset (strds) is created per imported layer, named {strds}_{layer}, e.g. uae_ch4enh and uae_ch4plm.
-r downloads EMIT L1B at-sensor radiance (EMITL1BRAD) instead of reflectance, for processing that starts from radiance, such as methane matched-filter retrievals with i.hyper.methane. Each granule holds two NetCDF files, both downloaded: the radiance (RAD) and the observation geometry (OBS: solar and view angles and path length). -r can be combined with -m/-c in the same run.
As for reflectance, only the newest collection version of each acquisition is downloaded (see Collection versions below).
With -i, each radiance file is imported as a raster_3d cube by i.hyper.import with its -r flag, which reads only the part of the swath covering the current region (a full L1B scene needs more than 11 GB of memory to import). Map names are {output}_rad_{acquisition timestamp}, and strds= creates a str3ds named {strds}_rad. The OBS file is downloaded but not imported. This needs an i.hyper.import whose EMIT reader imports L1B radiance and has the -r flag; with an older one, -r together with -i stops with an error before any download starts.
Unlike r.in.sentinel (which requests an exact bounding-box cube via cubo), EMIT has no equivalent per-request cropping service at LP DAAC — each matching granule’s full reflectance NetCDF file (its actual flight-line swath extent, several hundred MB to a few GB) is downloaded as-is. Only the reflectance (_RFL_) file is fetched; the companion uncertainty (_RFLUNCERT_) and quality-mask (_MASK_) files are skipped automatically, since i.hyper.import’s EMIT reader only needs the reflectance NetCDF (it carries its own GLT for georeferencing, and its own per-band good_wavelengths validity flags).
Downloaded files are kept in output_dir (default $HOME/RSDATA/EMIT_<output>, one subdirectory per acquisition year) and are not re-downloaded on a later run unless --overwrite is given — EMIT granules are large enough that a persistent, skip-if-present local cache matters far more here than for Sentinel-2 tiles.
EMIT L1B radiance and L2A reflectance are being reprocessed from collection version V001 to V002, and CMR returns both versions of an acquisition when both exist (V002 granule names also drop the orbit and scene suffix, e.g. EMIT_L2A_RFL_002_20230403T143443 next to EMIT_L2A_RFL_001_20230403T143443_2309310_003). Only the newest version of each acquisition is listed and downloaded, for reflectance, radiance and gas enhancement scenes alike. Without this, the same multi-GB scene was downloaded twice and both copies got the same map name, so only one of them was imported. Plume complexes are not filtered by version, since several plumes share one acquisition time.
The version filter runs before clouds=: the versions of one acquisition report different cloud cover (0 % for V002 and 15 % for V001 on 2023-04-03 over Ninove, Belgium), and the newest is the reference.
clouds= filters by each granule’s CMR-reported scene cloud-cover percentage (client-side — the CMR granule search API has no server-side cloud-cover query parameter for this collection, unlike a spatial/temporal filter).
A Bearer token is required for every request to LP DAAC’s protected data URLs. Resolved in this order: token=, then the EARTHDATA_TOKEN environment variable, then a TOKEN file inside output_dir. Generate one at urs.earthdata.nasa.gov (User Profile → Generate Token) — tokens expire, so expect to refresh this periodically.
For reflectance, -i runs i.hyper.import (product=emit) on every downloaded granule (gas products are imported differently, see above), producing one raster_3d hyperspectral cube per granule in the target project. composites=, composites_custom=, strength=, and -n are simply forwarded to i.hyper.import and only take effect together with -i. A granule already imported (a raster_3d map of the same name already exists) is skipped unless --overwrite is given, mirroring the download cache’s own skip behaviour.
Every imported cube gets its real acquisition date/time set via r3.timestamp, read directly from the granule’s own CMR time_start metadata — not just the calendar date, since EMIT’s flight-line acquisitions can occur multiple times per day over the same area.
strds= (requires -i) creates one str3ds — GRASS’s temporal type for a time series of 3D rasters, the raster_3d equivalent of a STRDS — via t.create type=str3ds + t.register type=raster_3d, collecting every cube imported in that run in one time-aware dataset, ready for t.rast3d.list, t.rast3d.algebra, or similar temporal-3D tools.
pip install netCDF4(netCDF4 is required by i.hyper.import’s EMIT reader when -i is used; the download step itself has no extra Python dependency beyond the standard library.)
g.region n=25.34 s=25.30 e=55.72 w=55.65
r.in.emit start=2024-01-01 end=2024-12-31 output=dubaig.region n=25.34 s=25.30 e=55.72 w=55.65
r.in.emit start=2022-01-01 end=2024-12-31 clouds=20 -lAvailable granules:
2023-07-27T10:02:31+00:00 cloud_cover=10% EMIT_L2A_RFL_001_20230727T100231_2320807_026.nc
2023-07-31T08:26:09+00:00 cloud_cover=13% EMIT_L2A_RFL_001_20230731T082609_2321206_020.nc
2024-06-21T08:27:46+00:00 cloud_cover=5% EMIT_L2A_RFL_001_20240621T082746_2417306_048.nc
2024-07-23T10:40:33+00:00 cloud_cover=11% EMIT_L2A_RFL_001_20240723T104033_2420507_022.nc
g.region n=25.34 s=25.30 e=55.72 w=55.65
r.in.emit start=2024-01-01 end=2024-12-31 clouds=30 \
output=dubai -i strds=dubai_ts composites=rgb,swir_geology
t.rast3d.list input=dubai_tsr.in.emit start=2024-01-01 end=2024-12-31 output=dubai \
-i project=emit_dubai_new dbase=$HOME/grassdatag.region n=25.34 s=25.30 e=55.72 w=55.65
r.in.emit start=2024-08-20 end=2024-08-31 -m -r -lAvailable granules:
2024-08-25T06:44:23+00:00 EMITL1BRAD cloud_cover=46% EMIT_L1B_RAD_001_20240825T064423_2423804_027 (2 files)
2024-08-25T06:44:23+00:00 EMITL2BCH4ENH cloud_cover=46% EMIT_L2B_CH4ENH_002_20240825T064423_2423804_027 (3 files)
2024-08-25T06:44:23+00:00 EMITL2BCH4PLM cloud_cover=0% EMIT_L2B_CH4PLM_002_20240825T064423_003501 (2 files)
g.region n=25.34 s=25.30 e=55.72 w=55.65
r.in.emit start=2024-08-25 end=2024-08-25 output=uae -m -i strds=uae
g.list type=raster,vector pattern="uae_*"
v.db.select map=uae_ch4plm_20240825T064423_003501r.in.emit start=2023-01-01 end=2024-12-31 output=uae -m -c \
gas_products=plumes -ig.region n=25.36 s=25.26 e=55.75 w=55.60
r.in.emit start=2024-08-25 end=2024-08-25 output=uae -r -ig.region n=25.34 s=25.30 e=55.72 w=55.65
r.in.emit start=2024-01-01 end=2024-12-31 -pg.proj, g.region, i.hyper.import, i.hyper.methane, r.import, r.in.sentinel, r.timestamp, r3.timestamp, t.create, t.register, v.import, v.timestamp
Yann Chemin
Available at: r.in.emit 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