Ammonia point-source detection above livestock and fertiliser facilities
Thermal infrared hyperspectral sounders can resolve concentrated NH3 plumes above intensive farms and urea-application events. This page explains what IASI, CrIS, TROPOMI and EMIT actually measure, where the method breaks down, and what a regulator can do with the output.
Sensors
- IASI (MetOp-A/B/C, EUMETSAT): Thermal infrared Fourier-transform sounder covering 645–2760 cm⁻¹ at 0.5 cm⁻¹ apodised resolution. Nadir footprint roughly 12 km diameter. Each MetOp satellite provides two overpasses per day globally; with three satellites the combined revisit is up to six overpasses. The primary satellite dataset for agricultural NH3 climatologies.
- CrIS (Suomi-NPP and NOAA-20, NOAA/NASA): Cross-track Infrared Sounder with 14 km nadir footprint. Full spectral resolution mode covers the NH3 ν2 band near 950 cm⁻¹. Provides independent daily global coverage complementary to IASI, and the two datasets are routinely cross-validated in the literature.
- Sentinel-5P TROPOMI (ESA/Copernicus): UV-VIS-NIR-SWIR push-broom spectrometer at 3.5 × 5.5 km pixel (post-2019 upgrade). NH3 retrieval uses solar backscatter in the 1460–1480 nm window. Better spatial resolution than IASI but a single daily overpass and higher detection floor over cold or cloudy surfaces.
- EMIT (ISS, NASA JPL): Imaging spectrometer covering 380–2500 nm at roughly 60 m ground sampling. Not a dedicated NH3 sensor, but pilot studies have demonstrated plume detection above very large point sources using the 1.5 µm absorption feature. Sub-kilometre resolution is a step change, though ISS orbital precession gives irregular revisit and global coverage is not guaranteed.
What the sounder is actually measuring
IASI and CrIS do not image ammonia directly. They measure upwelling thermal radiation from the Earth's surface and lower atmosphere across thousands of spectral channels. Where NH3 is present, it absorbs and re-emits at characteristic wavenumbers in the ν2 band near 950 cm⁻¹ and the ν1 band near 3300 cm⁻¹. Retrieval algorithms compare the observed radiance spectrum against a forward model of the atmosphere without NH3, then solve for the column amount that closes the residual. The result is a total column in molecules per square centimetre, or equivalently in kg NH3 per km².
Sensitivity depends critically on the thermal contrast between the surface and the overlying air. A warm surface beneath a cooler air mass gives strong contrast and good sensitivity, which is why morning overpasses over sunlit agricultural land in summer tend to produce the clearest retrievals. Cold surfaces, isothermal profiles, and high cloud all degrade or destroy sensitivity. This is not a minor caveat: over cool, moist or heavily clouded regions, detection limits can rise by an order of magnitude or more.
Detection floors and what slips through them
Published detection limits for IASI single-overpass retrievals are typically cited around 0.3–1 ppbv column-mean concentration, depending on surface temperature and atmospheric state. In practice, this corresponds to sources emitting on the order of hundreds to low thousands of tonnes of NH3 per year when integrated over a 12 km footprint. A single large broiler shed complex or a pig unit holding tens of thousands of animals can sit at or above this threshold on peak-emission days. A single dairy farm or a small-scale fertiliser application event generally will not.
The 12 km footprint is the harder constraint for enforcement purposes. Two or three intensive units within that footprint are indistinguishable from one large one. Attributing a retrieved column to a specific operator requires either a much smaller footprint (EMIT's 60 m is genuinely useful here, though coverage is opportunistic) or a dispersion model run backwards from the column to candidate sources using concurrent wind data. Neither approach eliminates ambiguity entirely. TROPOMI's 3.5 × 5.5 km pixel helps somewhat but its NH3 retrieval carries a higher noise floor than IASI over the same scene.
Urea application: a short window that satellites can catch
Urea broadcast onto fields hydrolyses to NH3 within hours to a few days, depending on soil temperature and moisture. The resulting emission pulse is brief but intense, and it is one of the clearest agricultural NH3 signals in the IASI record. Studies using IASI data have resolved regional NH3 enhancements over intensively farmed areas in Europe and South Asia that correspond temporally and spatially to known fertiliser application calendars.
For a regulator or an agricultural ministry, this matters in two ways. First, it provides an independent check on whether application codes of practice are being followed: incorporation of urea into soil within a specified window dramatically reduces volatilisation, and a persistent surface plume on the day after application is evidence that incorporation did not happen. Second, time-series analysis across multiple years can distinguish structural emitters (permanent intensive livestock units) from seasonal fertiliser pulses, which have different regulatory implications.
Wind dispersion ambiguity and how to reduce it
A retrieved NH3 column is a snapshot of the atmosphere at overpass time. The gas has already moved. Depending on wind speed, the centre of mass of a plume may be several kilometres downwind of the source at the moment the satellite passes. At IASI's 12 km footprint, a plume travelling at 5 m/s for 30 minutes has moved 9 km, which is enough to shift the apparent source location by nearly one full footprint.
The standard mitigation is to run a Gaussian dispersion model or a Lagrangian particle model backwards from the retrieved column centroid, using reanalysis wind fields (ERA5 is the common choice) to trace probable source locations. This narrows the candidate set but rarely identifies a single facility without corroborating data. Cross-referencing with facility registers, land-use maps, and historical emission inventories is part of any credible attribution workflow. Where EMIT data are available for the same area, the higher spatial resolution can confirm or rule out individual buildings.
Building a monitoring programme from open data
IASI Level 2 NH3 products are distributed by EUMETSAT and have been processed into multi-year climatologies by research groups including those associated with the ULB IASI NH3 dataset, which covers more than a decade. CrIS NH3 retrievals are available through NASA's NOAA-20 and Suomi-NPP data streams. TROPOMI NH3 data are available via the Copernicus Data Space. EMIT data are accessible through NASA Earthdata. None of these require a commercial licence.
The practical challenge is not data access but interpretation. A single overpass column is almost meaningless in isolation. Useful enforcement intelligence comes from anomaly detection relative to a multi-year baseline for the same location and season, trend analysis to identify facilities whose emissions are growing, and event detection for acute application episodes. Satellize runs exactly this kind of time-series analytics on open constellations; the workflow is similar in structure to the crop-estimation programme it operates for the Kingdom of Tonga, adapted to atmospheric rather than surface retrievals. Outputs can be delivered as GIS layers, facility-level time-series reports, or alert feeds triggered when a column exceeds a defined percentile threshold.
What the next generation of sensors will change
EMIT has demonstrated that imaging spectrometers at sub-100 m resolution can detect large NH3 point sources from orbit. The logical successor is a dedicated hyperspectral thermal infrared imager with both the spectral range to retrieve NH3 and the spatial resolution to resolve individual buildings. Several proposals exist in the literature and in agency roadmaps, but no operational system at that specification is yet in orbit.
In the near term, the most practical improvement is data fusion: combining IASI or CrIS columns for regional context with TROPOMI for intermediate spatial detail and EMIT for high-resolution confirmation on priority sites. This is not a complete solution. Cloud remains an irreducible problem, thermal contrast limits persist in cool climates, and the detection floor means that small farms will continue to fall below the threshold regardless of sensor combination. Honest use of this technology means being clear about what is below the noise floor, not just what is above it.
Typical figures
| Spatial footprint (IASI/CrIS) | ~12–14 km diameter nadir; degrades to ~20 km at swath edge |
| Spatial resolution (TROPOMI NH3) | 3.5 × 5.5 km (post-August 2019 upgrade) |
| Spatial resolution (EMIT) | ~60 m ground sampling distance; irregular coverage from ISS |
| Revisit (IASI, three MetOp satellites combined) | Up to 6 overpasses per day globally; ~2 per satellite |
| Revisit (TROPOMI) | ~1 overpass per day; near-global daily coverage |
| Spectral bands used for NH3 | ν2 band ~950 cm⁻¹ (10.5 µm) for IASI/CrIS; ~1460–1480 nm for TROPOMI; ~1.5 µm for EMIT |
| Minimum detectable column (IASI, favourable conditions) | ~0.3–1 ppbv column-mean; source strength roughly ≥ low hundreds of t NH3/yr at 12 km footprint |
| Key sensitivity constraint | Thermal contrast between surface and lower troposphere; cloud cover causes data loss; cold surfaces raise detection floor |
| Archive depth | IASI: MetOp-A from 2007; CrIS: Suomi-NPP from 2012; TROPOMI: from May 2018 |
| Data access | IASI via EUMETSAT; CrIS via NASA Earthdata; TROPOMI via Copernicus Data Space; EMIT via NASA Earthdata. All open, no commercial licence required. |
Analytics Satellize can run
| Facility-level NH3 anomaly score | Percentile ranking of retrieved column against multi-year seasonal baseline for each facility footprint; based on published IASI climatology methods | Monthly facility report with anomaly score, overpass count, and cloud-screening statistics |
| Urea application event detection | Time-series change detection on daily TROPOMI and IASI columns; temporal correlation with known application windows from agronomic calendars | Alert feed triggered when column exceeds 90th-percentile threshold for that location and calendar week |
| Source attribution probability map | Backward Lagrangian dispersion modelling from retrieved column centroid using ERA5 wind reanalysis; intersection with facility register | GIS layer showing probability-weighted source footprint polygons per overpass event |
| Multi-year emission trend analysis | Linear regression on annual mean NH3 column per facility cluster; seasonal decomposition to separate livestock and fertiliser signals | Annual trend report per facility or administrative area, with confidence intervals and data-gap disclosure |
| High-resolution plume confirmation (EMIT) | Matched-filter retrieval on EMIT radiance cubes for the NH3 1.5 µm absorption feature, applied to priority sites identified by IASI/TROPOMI screening | Sub-100 m plume extent map for confirmed events, delivered as GeoTIFF with detection confidence layer |
| Regulatory compliance screening report | Cross-referencing satellite-derived NH3 columns with national emission inventory declarations; statistical outlier identification at facility level | Ranked list of facilities where satellite-observed columns are inconsistent with declared emission factors, for regulator review |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.