Hydrofluorocarbon refrigerant leak detection from atmospheric column retrievals
High-spectral-resolution thermal infrared sounders can detect column enhancements of HFC-134a and HFC-23 above individual industrial facilities, turning atmospheric physics into an enforcement tool for the Kigali Amendment.
Sensors
- IASI (MetOp-A/B/C, EUMETSAT): Fourier-transform sounder covering 645–2760 cm⁻¹ at 0.5 cm⁻¹ apodised resolution. Footprint roughly 12 km diameter at nadir. Global twice-daily coverage per satellite; three satellites give up to six overpasses daily at mid-latitudes. The primary published instrument for HFC-134a and HFC-23 column retrievals, with detection sensitivities for HFC-134a reported in the low-ppt range under favourable thermal contrast conditions.
- CrIS (Suomi-NPP / JPSS-1, NOAA/NASA): Cross-track infrared sounder with spectral resolution up to 0.625 cm⁻¹ in the longwave band (650–1095 cm⁻¹). Nadir footprint 14 km. Covers the same 8–12 µm window as IASI and provides an independent retrieval basis. JPSS-1 (NOAA-20) extends the record and cross-validates IASI trends.
- AIRS (Aqua, NASA): Grating spectrometer, 3.74–15.4 µm, spectral resolving power roughly 1200. Footprint 13.5 km at nadir. Lower spectral resolution than IASI or CrIS limits HFC specificity, but the archive from 2002 onward provides a long baseline for trend analysis and seasonal background characterisation.
- GOSAT-2 (JAXA/MOE Japan): Thermal and near-infrared Fourier-transform spectrometer. The thermal infrared band (5.56–14.3 µm at 0.27 cm⁻¹ resolution) is specifically designed to extend HFC retrieval capability beyond IASI. Footprint roughly 9.7 km. Complements IASI by providing retrievals of co-emitted species including CO₂ and CH₄, supporting facility-level mass-balance checks.
Why refrigerants show up in a sounder spectrum
HFC-134a (1,1,1,2-tetrafluoroethane) and HFC-23 (trifluoromethane) absorb strongly in the 8–12 µm atmospheric window, a spectral region where the atmosphere is otherwise relatively transparent and outgoing thermal radiation from the surface reaches the satellite with minimal interference. Both molecules have multiple vibrational modes that produce absorption features spectrally distinct from water vapour, ozone and the other dominant atmospheric absorbers. IASI's apodised spectral resolution of 0.5 cm⁻¹ is sufficient to resolve these features and separate them from interfering species using optimal-estimation retrieval schemes.
The retrieval works by comparing the observed top-of-atmosphere radiance spectrum against a forward model that includes all known absorbers. The HFC signal is the residual after other species are fitted. Thermal contrast, the temperature difference between the surface and the overlying air, drives sensitivity. A warm industrial roof in summer over a cool boundary layer is a favourable geometry. At night or under isothermal conditions, the signal degrades substantially. This is not a minor caveat: many cold-storage facilities operate at ambient temperatures that reduce thermal contrast, and published retrieval studies acknowledge detection limits that vary by a factor of two or more depending on season and surface type.
What published retrievals have actually demonstrated
The most systematic published work on satellite HFC detection uses IASI. Studies have retrieved tropospheric columns of HFC-134a with global mean mixing ratios consistent with in-situ network measurements from AGAGE and NOAA, validating the retrieval approach at the hemispheric scale. Detecting an individual leaking facility is harder. A large refrigerated warehouse complex or a foam-blowing plant with a significant fugitive release can produce a column enhancement of a few ppt above the local background, but the 12 km IASI footprint integrates over a wide area, diluting the signal from a point or area source into the surrounding air mass.
Published work has demonstrated that anomalous column enhancements co-located with known industrial zones are detectable when multi-day composites are used to suppress meteorological variability. Single-overpass detection of a facility-level leak is not reliably achievable with current sounders for anything short of a catastrophic release. The honest framing is that these instruments are better suited to identifying persistent, elevated regional backgrounds above industrial districts than to pinpointing a single malfunctioning chiller unit. GOSAT-2's thermal infrared channel, with its slightly smaller footprint and higher spectral resolution, may improve facility attribution, though the published retrieval record for HFCs from GOSAT-2 is still developing.
The Kigali Amendment enforcement gap these data address
The Kigali Amendment to the Montreal Protocol, in force since 2019, commits signatory nations to phasing down HFC production and consumption on defined schedules. Compliance monitoring currently relies on national self-reporting and occasional atmospheric inversion studies using sparse ground-based networks. The gap between reported and inferred emissions has been documented for HFC-23 in particular, where satellite-derived atmospheric burdens have exceeded what production statistics would predict.
Space-based column retrievals offer a way to cross-check national inventories without requiring access to facilities or cooperation from regulated entities. An environmental regulator or treaty body can commission a multi-year IASI or CrIS time series over a country's known industrial refrigeration and chemical manufacturing zones and compare the trend in column burden against reported phase-down trajectories. The analysis does not prove a specific facility is at fault, but it can flag a jurisdiction whose atmospheric signal is inconsistent with its reported figures, triggering targeted ground inspection. That is a meaningful enforcement tool even if it falls short of facility-level attribution.
Atmospheric transport complicates everything
A column enhancement observed above a grid cell does not mean the source is underneath it. HFCs are long-lived: HFC-134a has an atmospheric lifetime of roughly 14 years, and HFC-23 around 228 years. They accumulate in the background troposphere and are transported across national boundaries on timescales of days to weeks. Attributing a local column anomaly to a specific source requires coupling the satellite retrieval to an atmospheric transport model, typically a Lagrangian particle dispersion model or a regional chemical transport model, to trace air mass back-trajectories to likely source regions.
This adds uncertainty. Transport models have their own errors, and the source inversion problem is underdetermined when only a handful of satellite overpasses constrain it. Retrieval precision for HFC-134a from IASI is on the order of 5–10% for monthly regional means, which is adequate for trend detection but tight for single-event attribution. Analysts should treat facility-level attribution as a hypothesis to be tested with additional evidence, not a conclusion the satellite data alone can sustain.
Combining sounders with ground networks and inspection records
The most defensible enforcement workflow pairs satellite column time series with complementary data. AGAGE and NOAA surface monitoring networks provide in-situ mixing ratio measurements that anchor the retrieval calibration and give an independent check on regional trends. Facility inspection records, equipment age databases and import/export records for refrigerant cylinders provide the ground truth needed to move from a regional atmospheric anomaly to a specific regulated entity.
Satellize's analytics work on this topic follows the same logic: satellite retrievals establish where the atmospheric signal is anomalous relative to background and trend, and that flags the geography for closer scrutiny. The approach is analogous in structure to the crop-estimation methodology Satellize runs for the Kingdom of Tonga, where satellite signals inform but do not replace ground-level verification. For HFC enforcement, the satellite layer is particularly valuable in jurisdictions where ground monitoring infrastructure is thin and self-reported inventories are the only alternative. Regulators seeking to build an independent atmospheric evidence base for treaty compliance reviews are the primary audience for this type of analysis.
Limits to be honest about before commissioning an analysis
No current satellite sounder can detect a single leaking refrigeration unit. The minimum detectable column enhancement for HFC-134a from IASI, under good thermal contrast conditions, is roughly in the range of a few ppt above background, which corresponds to a substantial aggregated source over the instrument footprint. Persistent cloud cover degrades retrievals; the thermal infrared window is blocked by thick cloud, and retrievals are typically flagged as invalid when cloud fraction exceeds around 20–30% of the footprint. High-latitude winter scenes with low thermal contrast are similarly problematic.
Spectral interference from other absorbers, particularly ozone and water vapour in the 8–12 µm window, requires careful forward-model treatment. Retrieval errors are correlated across the vertical profile, so column totals are more reliable than profile shape. The archive depth is meaningful: IASI data from MetOp-A runs from 2007, giving a 17-year record against which current anomalies can be assessed. CrIS extends from 2012. AIRS from 2002. For a treaty compliance investigation, that archive is genuinely useful. For real-time operational leak detection at individual facilities, the technology is not yet there.
Typical figures
| Spatial footprint at nadir | IASI: ~12 km diameter; CrIS: ~14 km; GOSAT-2: ~9.7 km; AIRS: ~13.5 km |
| Spectral resolution (longwave IR) | IASI: 0.5 cm⁻¹ apodised; CrIS: 0.625 cm⁻¹; GOSAT-2: ~0.27 cm⁻¹; AIRS: resolving power ~1200 |
| Spectral window for HFC retrieval | 8–12 µm (830–1250 cm⁻¹); HFC-134a features near 920 and 1100 cm⁻¹; HFC-23 near 1150 cm⁻¹ |
| Revisit frequency | IASI (3 satellites): up to 6 overpasses/day at mid-latitudes; CrIS: ~2/day; GOSAT-2: ~3-day repeat at equator |
| Archive depth | AIRS from 2002; IASI-A from 2007; CrIS from 2012; GOSAT-2 from 2019 |
| Minimum detectable column enhancement (HFC-134a) | Low ppt range above background under favourable thermal contrast; degrades significantly under isothermal or cloudy conditions |
| Cloud tolerance | Retrievals typically invalid above ~20–30% cloud fraction in footprint; thermal IR blocked by thick cloud |
| Retrieval precision (regional monthly mean) | ~5–10% for HFC-134a column from IASI; single-overpass precision considerably lower |
| Data latency (operational products) | IASI near-real-time products available within ~3 hours of overpass via EUMETSAT; science-quality retrievals typically days to weeks |
| Delivery formats | NetCDF column retrieval files; time-series CSV by facility region; anomaly maps as GeoTIFF or WMS layer |
Analytics Satellize can run
| Regional HFC column anomaly map | Optimal-estimation retrieval applied to IASI L1C spectra; multi-day composite to suppress meteorological noise; anomaly scored against 5-year seasonal baseline | Monthly GeoTIFF anomaly layer with z-score per grid cell, delivered as GIS layer or PDF report |
| Facility-zone time series | Extraction of column retrievals within defined radius of registered industrial facilities; trend decomposition separating seasonal cycle from interannual signal | CSV time series per facility zone with trend line and confidence interval; quarterly PDF summary |
| Treaty compliance divergence assessment | Comparison of satellite-derived regional column burden trend against national inventory submissions; atmospheric inversion using HYSPLIT or similar Lagrangian transport model to attribute anomalies to source regions | Written assessment report with supporting figures, suitable for submission to a regulatory or treaty body |
| Thermal contrast quality flag layer | Surface minus air temperature difference computed from ERA5 reanalysis and IASI skin temperature retrievals; flags overpasses where retrieval sensitivity is degraded | Per-overpass quality flag appended to retrieval CSV; monthly summary of valid-retrieval fraction by region |
| Multi-species co-emission check | Cross-referencing HFC column anomalies with GOSAT-2 CO₂ and CH₄ columns to identify facility-level mass-balance inconsistencies indicative of active industrial processes | Correlation table and scatter plots by facility type; flagged sites ranked by anomaly consistency |
| Archive baseline characterisation | Statistical analysis of AIRS (2002 onward) and IASI (2007 onward) records to establish pre-Kigali background and phase-down trajectory for a defined geographic domain | Baseline report with decadal trend figures and uncertainty bounds; GIS polygon defining the analysis domain |
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.