Perfluorocarbon emission detection at aluminium smelters
Anode effects in aluminium electrolysis cells emit CF4 and C2F6, gases with atmospheric lifetimes exceeding 10,000 years. Thermal infrared sounders aboard MetOp and Aqua can detect column enhancements above the largest smelter clusters, though only at the upper end of the emission scale.
Sensors
- IASI (MetOp-A, -B, -C): The Infrared Atmospheric Sounding Interferometer covers 645–2760 cm⁻¹ at 0.5 cm⁻¹ apodised resolution. Its sensitivity near 7.8 µm (the CF4 ν3 absorption band) allows column retrievals over large industrial complexes. Nadir footprint is approximately 12 km diameter; three MetOp satellites together provide roughly twice-daily global coverage. Published studies place the detection threshold for CF4 enhancements at roughly 0.5–1 ppb above background for a single overpass, meaning only clusters of multiple large smelters are currently resolvable.
- CrIS (Suomi-NPP, JPSS-1/NOAA-20): The Cross-track Infrared Sounder shares spectral coverage with IASI and operates at 14 km nadir footprint. Its apodised spectral resolution of 0.625 cm⁻¹ in the shortwave infrared window gives comparable CF4 sensitivity. CrIS adds a second independent instrument family for cross-validation and extends the record beyond the MetOp series.
- AIRS (Aqua): The Atmospheric Infrared Sounder has operated since 2002, providing the longest continuous thermal infrared sounding record available. Spectral resolution is coarser than IASI (roughly 1200 channels versus 8461), but AIRS data underpin published global CF4 trend analyses and offer archive depth no other sounder can match. Footprint is approximately 13.5 km at nadir.
- TROPOMI (Sentinel-5P): TROPOMI does not retrieve CF4 directly; its shortwave infrared channels target CH4, CO and aerosol. It is included here because co-emitted species, particularly CO from anode baking furnaces, can corroborate smelter activity and help distinguish anode-effect events from background variability. Spatial resolution is 5.5 × 3.5 km since the 2019 detector upgrade.
Why aluminium smelters are a special case in greenhouse gas accounting
Most industrial greenhouse gas monitoring focuses on CO2 and CH4 because they dominate the mass budget. Perfluorocarbons from aluminium smelting are different: the quantities are small in tonnes but the radiative forcing is disproportionate. CF4 has a global warming potential of approximately 6,630 over 100 years and an atmospheric lifetime estimated at 50,000 years. C2F6 is worse still on a per-molecule basis. A smelter that underreports its anode-effect frequency by even a modest margin accumulates a climate liability that persists for geological timescales.
Anode effects occur when the alumina feed in an electrolysis cell drops too low, causing the bath chemistry to shift and fluorocarbon gases to form at the carbon anode. Modern smelters suppress them with automated feed control, but the events still happen, and their frequency varies widely between facilities and between shifts. Self-reported anode-effect data feed into national greenhouse gas inventories, but independent verification from outside the plant boundary has historically been difficult. That is the gap thermal infrared sounders are beginning to close, partially.
What IASI actually measures, and what it cannot
IASI measures the upwelling thermal emission spectrum of the Earth's surface and atmosphere at roughly 8,461 spectral channels. CF4 absorbs strongly near 1283 cm⁻¹ (7.8 µm), a window where the atmosphere is otherwise relatively transparent and surface emission is detectable. Retrieval algorithms fit the observed radiance spectrum against a forward model to estimate the total column abundance of CF4. The background tropospheric concentration of CF4 is currently around 85–90 parts per trillion by volume, rising at roughly 1 ppt per year from all sources combined. A smelter complex producing a detectable signal needs to elevate the column by a fraction of a part per billion above that background, within a footprint of roughly 100–150 km².
The honest constraint is the detection floor. Published work by Duhamel and colleagues, and by the broader IASI science community, shows that individual large smelters are generally below the single-overpass detection threshold. What has been demonstrated in the peer-reviewed literature is the detection of regional enhancements attributable to smelter clusters, for example in Xinjiang, in the Arabian Gulf industrial corridor, and in parts of Siberia where multiple large facilities operate within a few hundred kilometres of each other. A single 300,000-tonne-per-year smelter is unlikely to be unambiguously resolved. A cluster totalling 1–2 million tonnes of annual capacity in a compact geography may be. Thermal sensitivity, surface emissivity uncertainty, and cloud cover all add noise. Averaging multiple overpasses improves the signal-to-noise ratio but reduces temporal resolution.
From spectral retrieval to emission estimate: the methodological chain
Turning a column enhancement into a facility-level emission rate requires several steps, each of which introduces uncertainty. First, the retrieved CF4 column (in molecules per cm²) must be converted to a surface concentration enhancement using atmospheric transport modelling. Trajectory analysis with tools such as HYSPLIT, or adjoint runs with a chemistry-transport model, links the observed plume to candidate source regions. Second, the enhancement must be separated from the background trend, which itself has a seasonal component driven by atmospheric dynamics rather than emission variability.
The published approach for IASI CF4 retrievals typically uses optimal estimation, producing a retrieved profile with averaging kernels that describe the vertical sensitivity. Because CF4 is well-mixed and has no significant tropospheric sink, the column retrieval is relatively clean compared with reactive species. The main ambiguity is spatial: attributing an enhancement to a specific facility within a cluster requires ancillary information, such as wind direction at overpass time, facility production records, or co-emission signatures from species with better spatial resolution. TROPOMI CO or NO2 can narrow the attribution when smelter activity coincides with anode-baking or power-generation emissions.
Building a monitoring programme around an imperfect signal
The practical value of IASI-based PFC monitoring is not in catching a single anode-effect event in real time. The revisit cadence and footprint make that impossible. The value is in multi-year trend analysis: a facility or cluster whose reported anode-effect rate is declining should show a corresponding decline in the seasonal-mean CF4 enhancement above its region. A facility claiming compliance improvements that are absent from the IASI record has a discrepancy worth investigating.
Archive depth matters here. IASI-A data begin in 2007, IASI-B in 2012, IASI-C in 2018. AIRS extends the sounder record back to 2002. That is long enough to bracket major smelter expansions in China and the Middle East and to test whether reported emission reductions correspond to observable atmospheric changes. For a regulator or a sovereign government assessing whether its aluminium sector's inventory submissions are internally consistent, this multi-year perspective is more actionable than any single overpass.
Satellize can structure this kind of longitudinal retrieval programme, combining IASI and CrIS archive processing with transport modelling to produce annual facility-cluster assessments. The workflow draws on the same open-data infrastructure used in the Tonga crop-estimation programme, adapted for atmospheric rather than surface retrievals.
Honest limits and what would improve them
No currently operational sounder can attribute a CF4 enhancement to a single smelter with high confidence unless that smelter is geographically isolated and very large. The 12–14 km footprint, combined with the fraction-of-a-ppb detection floor, means the technique works at the cluster or regional scale. Buyers should not expect facility-level emission rates with the kind of precision that direct-measurement stack monitoring provides.
What would improve matters: a dedicated PFC-sensing instrument with finer spatial resolution and higher spectral resolving power in the 7.8 µm window, combined with a low-orbit constellation for more frequent revisit. No such instrument is currently in routine operation. The IASI-NG instrument planned for the MetOp-SG series will offer improved spectral resolution (0.25 cm⁻¹ apodised) and a smaller footprint, which should push the detection threshold meaningfully lower. Until then, the honest position is that IASI-based CF4 monitoring provides credible regional-scale evidence for use in inventory cross-checking, not facility-level enforcement with legal precision.
Typical figures
| Primary sensor footprint (nadir) | ~12 km diameter (IASI); ~13.5 km (AIRS); ~14 km (CrIS) |
| Spectral band for CF4 retrieval | ~1283 cm⁻¹ (7.8 µm), ν3 absorption feature |
| Spectral resolution | 0.5 cm⁻¹ apodised (IASI); 0.625 cm⁻¹ (CrIS); ~4 cm⁻¹ equivalent (AIRS) |
| Revisit (combined MetOp-A/B/C) | ~twice daily global coverage; single-satellite ~12-hour repeat |
| Minimum detectable CF4 enhancement (single overpass) | ~0.5–1 ppb above background; cluster-scale only |
| Tropospheric CF4 background (current) | ~85–90 ppt, rising ~1 ppt/year |
| Archive depth | AIRS from 2002; IASI-A from 2007; IASI-B from 2012; IASI-C from 2018 |
| Cloud sensitivity | Optically thick cloud blocks surface/lower-troposphere signal; clear-sky screening required |
| Latency (operational products) | Near-real-time IASI L2 products available within ~3 hours of overpass via EUMETSAT |
| Delivery formats | NetCDF column retrieval grids; GeoTIFF anomaly maps; tabular cluster time series |
Analytics Satellize can run
| Regional CF4 column anomaly map | Optimal-estimation retrieval on IASI L1C spectra; multi-overpass averaging to suppress noise | Seasonal GeoTIFF showing CF4 column departure from background, gridded at 0.25° for named industrial regions |
| Smelter-cluster emission trend report | Multi-year time series of CF4 column means over defined cluster polygons, tested against reported anode-effect frequency data | Annual PDF report with trend plots, confidence intervals and inventory-consistency assessment |
| Source attribution via transport back-trajectory | HYSPLIT or equivalent trajectory ensemble run at overpass time; intersection with smelter facility database | Per-overpass attribution table linking column enhancements to candidate facility clusters with probability scores |
| Cross-sensor consistency check | Parallel retrieval on IASI and CrIS for overlapping periods; bias characterisation and merged record construction | Merged CF4 time series with instrument-agreement statistics, suitable for inventory submission support |
| Co-emission corroboration layer | TROPOMI CO and NO2 columns co-located with IASI CF4 enhancements to confirm smelter-complex activity | GIS layer overlaying CF4 anomaly polygons with TROPOMI co-emission signals, delivered as GeoPackage |
| Regulatory discrepancy flag | Comparison of IASI-derived regional CF4 trend against national inventory submissions; statistical significance testing | Structured alert report identifying facilities or regions where atmospheric evidence is inconsistent with declared emission reductions |
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.