Benzene column retrieval above petrochemical and coking facilities

Benzene is a Group 1 carcinogen regulated under most national air-quality frameworks, yet routine satellite monitoring remains constrained by signal-to-noise limits. This page covers what TROPOMI, EMIT and airborne sensors can and cannot currently resolve.

Sensors

Sentinel-5P TROPOMI: The primary spaceborne reference for benzene column work. TROPOMI's SWIR spectrometer covers 2305–2385 nm, which includes benzene absorption features, though benzene retrievals are experimental rather than operational. Nadir pixel size is 5.5 × 3.5 km (post-August 2019 upgrade from 7 × 3.5 km). Daily global coverage. Signal-to-noise constraints mean only very large, concentrated sources are resolvable against background variability.
EMIT (ISS-mounted imaging spectrometer): NASA's Earth Surface Mineral Dust Source Investigation instrument covers 380–2500 nm at roughly 60 m ground sampling distance from the ISS at approximately 400 km altitude. Published work has demonstrated methane and CO2 point-source detection; the SWIR spectral range overlaps benzene absorption bands, making it a candidate for large industrial plume work, though benzene retrievals from EMIT are not yet operationally validated at the time of writing.
AVIRIS-NG (airborne): The Airborne Visible/Infrared Imaging Spectrometer Next Generation, operated by NASA JPL, is the published reference standard for benzene plume mapping. It covers 380–2510 nm with roughly 5 nm spectral sampling and sub-metre to low-metre spatial resolution depending on flight altitude. Studies over Los Angeles Basin refineries and Californian coking facilities have demonstrated benzene column detection at concentrations relevant to regulatory thresholds, but AVIRIS-NG is a campaign instrument, not a persistent monitor.
GHGSat-C series: GHGSat's commercial Fabry-Pérot spectrometers target methane and CO2 in the SWIR at roughly 25 m resolution per scene. The instrument design is not optimised for benzene retrieval, and GHGSat has not published benzene detection capability. Included here because the C-series demonstrates that compact SWIR spectrometers can achieve point-source sensitivity at industrial scales, which informs expectations for planned hyperspectral successors.
Planned hyperspectral missions (Carbon Mapper, CHIME, SBG): Carbon Mapper (Tanager-1 launched 2024) targets methane and CO2 at roughly 30 m resolution with full SWIR coverage. ESA's CHIME and NASA-USGS SBG missions, both targeting launch in the late 2020s, will provide imaging spectroscopy at 20–30 m resolution with revisits of days to weeks. These are the platforms most likely to make routine benzene column retrieval at industrial point sources tractable.

Why benzene is hard to see from space

Benzene (C6H6) has absorption features in the shortwave infrared, principally near 2.4 µm, that are in principle distinguishable from methane, CO2 and water vapour using high-spectral-resolution instruments. In practice, the challenge is concentration. Ambient benzene columns above even heavily industrialised areas are typically in the parts-per-billion range, producing column enhancements of a few hundred nanomoles per square metre above background. TROPOMI's SWIR detector, designed primarily for methane and CO2, has a noise floor that places benzene detection at the edge of feasibility for anything short of a major facility cluster or an accident scenario.

Thermal contrast also matters. SWIR retrievals depend on reflected solar radiation, which means they are daytime-only and perform best when surface temperatures are elevated relative to the overlying air column. Coking plants and petrochemical crackers often run hot, which helps. Cloud cover, as with every passive optical technique, is a hard blocker. High-latitude facilities face additional seasonal penalties as low solar elevation angles reduce signal.

What the published retrieval literature actually shows

The most rigorous published benzene plume retrievals from imaging spectrometers come from AVIRIS-NG campaigns. Work published in peer-reviewed remote sensing journals, including studies accessible through MDPI's Remote Sensing journal, has demonstrated that AVIRIS-NG can map benzene concentration enhancements above refineries and coking facilities in California at spatial resolutions fine enough to attribute emissions to individual process units. The method typically uses a matched-filter approach: a spectral template for benzene is cross-correlated against the measured radiance spectrum after atmospheric correction, with the residual signal converted to a column concentration using radiative transfer modelling.

TROPOMI benzene retrievals are a different proposition. A small number of research groups have attempted to extract benzene columns from TROPOMI's SWIR channel using differential optical absorption spectroscopy adapted for the infrared. Results are promising for large petrochemical clusters, particularly in regions such as the Yangtze River Delta and the US Gulf Coast, where many facilities are co-located and the aggregate column enhancement rises above the instrument noise floor. Single-facility attribution at TROPOMI's pixel size remains unreliable for all but the largest emitters. This is an honest constraint, not a temporary software problem.

Coking plants as a specific detection target

Coking facilities deserve particular attention. Metallurgical coke production involves heating coal to temperatures above 1000°C in the absence of oxygen, a process that releases a mixture of volatile organic compounds including benzene, toluene and naphthalene. Benzene emissions from coking are regulated under the US National Emission Standards for Hazardous Air Pollutants (NESHAP) and equivalent European directives, but ground-level monitoring networks are sparse and self-reported inventories are the norm.

From a remote sensing perspective, coking plants are attractive targets because they tend to be large, thermally active and spatially isolated from other industrial sources, which reduces the spectral mixing problem. Several Chinese steel complexes, which include integrated coking operations, are large enough that TROPOMI column enhancements attributable to co-emitted species such as NO2 and SO2 are well-documented. Whether benzene columns from these same facilities are retrievable at TROPOMI resolution is an active research question. The co-emission signature is useful circumstantial evidence even when direct benzene retrieval is ambiguous.

EMIT and the path toward 60-metre retrievals

EMIT's published detection work has focused on methane and CO2, where the instrument has demonstrated point-source sensitivity at industrial scales. The spectrometer's coverage of the 2.3–2.5 µm region means benzene absorption features fall within its spectral range. NASA's EMIT team has not published validated benzene retrievals as of the time of writing, but the instrument's 60 m ground sampling distance and full SWIR spectral coverage make it a credible candidate for detecting large benzene plumes from petrochemical clusters, particularly where thermal contrast is favourable.

The ISS orbit limits EMIT's revisit to roughly once every few days at mid-latitudes, and the precessing orbit means coverage is not predictable in the way a sun-synchronous mission would be. For enforcement applications requiring scheduled revisits, this is a significant operational limitation. The instrument is best treated as a proof-of-concept for what a dedicated hyperspectral mission in a sun-synchronous orbit could achieve systematically.

Signal-to-noise, the detection floor and what regulators should expect

A frank summary of current capability: airborne SWIR spectrometers such as AVIRIS-NG can detect and map benzene plumes above large petrochemical and coking facilities at concentrations relevant to regulatory thresholds, but they require campaign-style deployment and are not persistent monitors. TROPOMI can, under favourable conditions and with careful averaging over multiple overpasses, indicate elevated benzene columns above major industrial clusters, but pixel-level attribution to a single facility is not reliable at 5.5 km resolution.

The minimum detectable column enhancement for TROPOMI benzene work is roughly estimated in published research at several hundred nanomoles per square metre above background, which corresponds to a large and sustained emission. Planned missions such as Carbon Mapper's Tanager constellation and ESA's CHIME, operating at 20–30 m resolution with full SWIR spectral coverage, are expected to reduce this floor substantially, though published detection limits for benzene specifically from these systems are not yet available.

Regulators and compliance teams should treat current satellite benzene data as a screening and prioritisation tool rather than a primary enforcement record. A strong TROPOMI signal above a facility cluster justifies directing airborne or ground-based monitoring assets toward that location. It does not, on its own, constitute a defensible emission measurement. Satellize's analytics work for clients such as the Kingdom of Tonga crop-estimation programme demonstrates the same principle in a different domain: satellite data narrows the search space; it rarely closes the case alone.

Practical workflow for an enforcement agency

A tiered approach makes the most of current capability. First, use TROPOMI multi-overpass averages to identify petrochemical and coking clusters where column enhancements in the SWIR are persistently elevated relative to upwind background. Second, cross-reference against co-emitted species, particularly NO2 and SO2, which are better constrained at TROPOMI resolution and can confirm that a facility is actively operating at high intensity. Third, schedule EMIT tasking or an AVIRIS-NG campaign flight over priority facilities to obtain spatially resolved plume maps that can support regulatory action.

Wind field data from ERA5 reanalysis or ECMWF operational forecasts is essential at every tier. Without a credible wind vector, column enhancements cannot be converted to emission rates, and source attribution at the facility level is guesswork. The combination of a spectral detection and a well-constrained wind field is what turns a satellite observation into a number an enforcement agency can act on.

Typical figures

TROPOMI spatial resolution (SWIR)	5.5 × 3.5 km per pixel (post-August 2019)
TROPOMI revisit	Daily global coverage; ~14 orbits per day
EMIT ground sampling distance	~60 m; ISS orbit gives irregular revisit of roughly 1–4 days at mid-latitudes
AVIRIS-NG spatial resolution	Sub-metre to ~5 m depending on flight altitude; campaign deployment only
Spectral range for benzene retrieval	Primary absorption feature near 2.4 µm (SWIR); also UV features used in ground-based DOAS
Estimated TROPOMI benzene detection floor	Several hundred nmol/m² column enhancement above background (research estimate; not operationally validated)
Cloud cover limitation	Passive SWIR retrieval blocked by cloud optical depth >~0.5; no all-weather capability
TROPOMI archive depth	May 2018 to present (Sentinel-5P launch April 2017; SWIR science data from mid-2018)
Planned hyperspectral mission resolution	Carbon Mapper Tanager-1: ~30 m; ESA CHIME and NASA SBG: 20–30 m (late 2020s)
Delivery formats	NetCDF (TROPOMI L2 products); GeoTIFF (EMIT radiance cubes); CSV/GIS layer for derived column maps

Analytics Satellize can run

TROPOMI benzene column anomaly map	Multi-overpass averaging and differential spectral fitting in the SWIR; upwind/downwind background subtraction	Monthly GeoTIFF layer flagging facility clusters with persistent column enhancements above a defined threshold
Co-emission correlation index	Cross-correlation of TROPOMI benzene candidate signal with co-located NO2 and SO2 columns to confirm active industrial operation	Tabular report ranking facilities by co-emission consistency score, updated quarterly
Wind-corrected emission flux estimate	Integrated mass enhancement method using TROPOMI or EMIT column data combined with ERA5 wind fields at overpass time	Point-source emission rate estimate in kg/hour with uncertainty range; PDF report per facility
EMIT plume extent polygon	Matched-filter retrieval applied to EMIT radiance cubes using published benzene spectral template; threshold masking to define plume boundary	GeoJSON polygon of detected plume extent with peak column concentration annotation
Facility prioritisation ranking for airborne follow-up	Composite scoring using TROPOMI column anomaly magnitude, co-emission signal, facility size from public inventory data, and wind-day frequency	Ranked shortlist of facilities recommended for AVIRIS-NG or regulatory ground inspection; delivered as a briefing document
Time-series trend analysis	Monthly TROPOMI column averages over defined facility polygons, tested for monotonic trend using Mann-Kendall statistic	Annual trend chart per facility with statistical significance flags; CSV export for regulatory records

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.