SF6 emission detection at high-voltage electrical substations

Sulphur hexafluoride leaks from high-voltage switchgear carry a global warming potential ~23,500× CO2. Thermal infrared sounders can detect SF6 column enhancements at the district scale, though single-facility attribution remains beyond current thresholds.

Sensors

IASI (MetOp-A, -B, -C): Infrared Atmospheric Sounding Interferometer; 12 km nadir footprint, twice-daily global coverage per satellite (three satellites in operation give up to six overpasses daily at mid-latitudes). Spectral range 645–2760 cm⁻¹ at 0.5 cm⁻¹ apodised resolution covers the SF6 ν3 absorption band near 948 cm⁻¹ (10.5 µm). Published retrieval sensitivity is in the low-ppt range for total-column SF6 under clear-sky conditions.
CrIS (Suomi-NPP, NOAA-20/JPSS-1): Cross-track Infrared Sounder; 14 km nadir footprint, once-daily polar overpass. Covers the same SF6 absorption window with spectral resolution sufficient for column retrievals, though slightly coarser spectral sampling than IASI limits sensitivity at the low end of the detection range.
AIRS (Aqua): Atmospheric Infrared Sounder; 13.5 km footprint, once-daily equatorial crossing. Operational since 2002, giving a 20-year archive useful for long-term trend analysis of regional SF6 backgrounds. Spectral resolution is lower than IASI, making it better suited to background monitoring than anomaly detection.
Sentinel-5P / TROPOMI: TROPOMI does not have a dedicated SF6 retrieval channel; its shortwave-infrared and ultraviolet bands target CH4, NO2, SO2 and CO. Included here because co-located NO2 columns from TROPOMI (3.5 × 5.5 km pixel) can confirm substation cluster activity and help distinguish industrial from background air masses when interpreting IASI SF6 anomalies.

Why a gas with a 3,200-year atmospheric lifetime deserves orbital attention

SF6 does not react with anything in the troposphere or stratosphere. Once released, it accumulates. The IPCC AR6 assessment places its 100-year global warming potential at approximately 23,500 times that of CO2, making even small annual leak rates from electrical infrastructure climatically significant at national scale. Global atmospheric SF6 concentrations have risen from near zero in the 1950s to roughly 11 parts per trillion by volume as of the early 2020s, driven almost entirely by the electrical power sector.

Regulatory pressure is increasing. The EU F-Gas Regulation requires leak detection and reporting for SF6-containing equipment above defined thresholds, and several jurisdictions are pushing toward SF6-free switchgear mandates for new installations by the early 2030s. The compliance gap between what utilities self-report and what atmospheric measurements suggest is the enforcement problem that satellite observation is positioned to help close, at least at the district or national aggregate level.

What the 10.5 µm window actually reveals

SF6 has an exceptionally strong absorption feature centred near 948 cm⁻¹, in the thermal infrared. IASI, which operates as a Fourier-transform spectrometer, resolves this feature clearly against the atmospheric background. Published studies using IASI data have demonstrated the ability to retrieve SF6 total columns with a precision of roughly 0.1–0.3 ppt under clear-sky conditions, which corresponds to detecting enhancements of a few percent above a background of ~10 ppt. That sounds impressive until you do the arithmetic on what a single leaking substation contributes to a 12 km footprint.

A large gas-insulated switchgear installation might contain several hundred kilograms of SF6 and leak at rates of 0.5–2% per year. Spread across a 12 km IASI footprint, that annual flux is diluted to a column enhancement well below the retrieval noise floor. The practical detection unit is therefore not a single facility but a cluster: a dense urban or industrial zone containing dozens of substations, gas-insulated lines and associated switchgear, all leaking simultaneously into a relatively confined air mass. Under stable atmospheric conditions and favourable wind geometry, such clusters have been shown in published literature to produce detectable column anomalies above the regional background.

The aggregation strategy: districts, not addresses

The honest framing for satellite SF6 work is regional inventory verification, not facility-level attribution. A national grid operator or regulator can use IASI time-series data to ask whether a particular metropolitan region's SF6 column is growing faster than the global background trend, and whether that growth is consistent with declared equipment inventories. This is analogous to how atmospheric inversion methods are used for CO2 and CH4: the satellite constrains the top-down estimate, and the discrepancy with bottom-up reporting flags where ground investigation is warranted.

Seasonal and meteorological stratification matters considerably. SF6 retrievals are only valid under clear-sky conditions; cloud cover introduces gaps that can be substantial in maritime climates. Boundary-layer mixing height affects how much a surface flux elevates the column. Reliable analysis therefore requires multi-year data stacking to separate signal from meteorological noise, and co-registration with reanalysis wind fields to attribute column anomalies to likely source regions rather than upwind transport.

A practical workflow combines IASI SF6 columns with TROPOMI NO2 maps (which identify active industrial zones at higher spatial resolution) and with publicly available substation location data from national grid operators or OpenStreetMap. The spatial overlap between NO2 hotspots and SF6 column anomalies narrows the candidate source area considerably, even if it cannot resolve individual bays.

Where the method breaks down

Thermal infrared sounders measure the total atmospheric column. They are not sensitive to where in the column the SF6 sits, which matters when distinguishing surface emissions from transported background air. Retrieval sensitivity is also strongly surface-temperature dependent: IASI performs best when there is a good thermal contrast between the surface and the overlying atmosphere, which favours daytime retrievals over warm land surfaces and disfavours winter high-latitude scenes.

The 12 km footprint means that in dense urban environments, multiple source types overlap. Industrial SF6, any residual natural background (negligible for SF6, but relevant for interfering species), and instrument noise all sit in the same retrieval. False positives from spectral interference with other atmospheric constituents are a documented concern in published validation work, requiring careful forward-model treatment of water vapour and ozone in the same spectral window.

Perhaps most importantly: current satellite retrievals cannot distinguish a leaking substation from one that has been deliberately vented during maintenance. Both produce the same atmospheric signal. Ground-truth correlation with maintenance logs is essential before any enforcement action is contemplated.

Building a practical monitoring programme

A credible SF6 monitoring programme at national scale starts with a baseline: several years of IASI and AIRS data processed through a consistent retrieval algorithm to establish the pre-programme column climatology for each major grid zone. New data are then assessed against that baseline rather than against a global background, which reduces the influence of long-range transport and improves the signal-to-noise for regional anomalies.

Satellize runs this class of atmospheric column analysis on open-access sounder archives, combining retrieval outputs with spatial priors derived from infrastructure databases. The approach is directly analogous to the aggregated district-level analysis described in the published literature, applied to client-defined grid zones rather than research domains. For regulators or grid operators considering an SF6 monitoring obligation, the sensible first step is a retrospective analysis of existing IASI data for the relevant territory, to understand what the current signal-to-noise situation actually looks like before committing to a reporting framework built on satellite evidence.

The technology is not yet at the point where a satellite can serve a penalty notice on a specific substation. It is, however, at the point where a national regulator can credibly tell a grid operator that a particular region's atmospheric SF6 trend is inconsistent with declared leak rates, and that an audit is warranted. That is a meaningful enforcement tool, and it will only improve as next-generation sounders with finer spectral sampling enter service.

Typical figures

Spatial resolution (IASI nadir footprint)	12 km diameter circle
Revisit (three MetOp satellites combined)	Up to 6 overpasses per day at mid-latitudes; 2 per day per satellite
SF6 absorption band exploited	ν3 band, ~948 cm⁻¹ (10.5 µm), thermal infrared
Retrieval precision (clear-sky, IASI)	~0.1–0.3 ppt total column; background ~10–11 pptv (early 2020s)
Minimum detectable source (practical)	District-scale cluster of substations; single-facility attribution not currently feasible
Cloud limitation	Retrievals invalid under cloud cover; gap-filling requires multi-year averaging
Archive depth	IASI from 2007 (MetOp-A); AIRS from 2002; CrIS from 2012
Data latency (near-real-time products)	IASI NRT dissemination via EUMETSAT: typically within 3 hours of overpass
Complementary NO2 layer (TROPOMI)	3.5 × 5.5 km pixel; daily global; used for industrial zone co-registration

Analytics Satellize can run

Regional SF6 column baseline climatology	Multi-year IASI L2 column retrieval stacking; anomaly detection against seasonal climatological mean	GIS layer of mean SF6 columns per grid zone, with interannual trend coefficients; delivered as GeoTIFF and PDF summary report
Column anomaly flags above declared grid zones	Z-score threshold detection on monthly IASI composites, stratified by boundary-layer mixing height from ERA5 reanalysis	Monthly alert report identifying zones where column exceeds baseline by a statistically defined margin, with meteorological context
Top-down vs. bottom-up inventory comparison	Atmospheric inversion using IASI columns as observational constraint; comparison against declared SF6 equipment inventories	Quantitative discrepancy estimate per grid zone (in kg SF6 yr⁻¹ equivalent), with uncertainty range; formatted for regulatory submission
Industrial zone co-location analysis	Spatial overlay of IASI SF6 anomaly pixels with TROPOMI NO2 hotspots and substation location priors from infrastructure databases	Ranked list of candidate source districts with probability scores; GIS polygon layer
Long-term trend monitoring subscription	Automated ingestion of new IASI NRT products; rolling 12-month trend update against historical baseline	Quarterly trend bulletin with time-series charts per monitored zone; optional API feed of column values

Sources

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.