Tropospheric nitrogen dioxide column mapping
TROPOMI on Sentinel-5P retrieves tropospheric NO2 columns daily at 3.5 × 5.5 km, exposing industrial point sources, shipping corridors and urban traffic patterns that ground monitors miss entirely.
Sensors
- Sentinel-5P TROPOMI: Primary instrument for this use case. Pushbroom spectrometer covering UV, visible, near-infrared and shortwave infrared (270–2385 nm). Retrieves tropospheric NO2 columns at 3.5 × 5.5 km (improved from original 7 × 3.5 km after August 2019 processor upgrade). Daily global coverage, with offline Level-2 products typically available within 3 hours of sensing.
- OMI (Aura satellite, NASA): Heritage instrument that established the DOAS-based NO2 retrieval methodology now used by TROPOMI. Spatial resolution 13 × 24 km, daily global revisit. Operational since 2004, providing the longest continuous satellite NO2 record; row-anomaly artefacts affect part of the swath from 2009 onward.
- GEMS (GEO-KOMPSAT-2B, KMA): Geostationary Environment Monitoring Spectrometer, covering East Asia at roughly 3.5 × 8 km. Unique capability: hourly daytime observations, enabling diurnal NO2 cycle analysis that polar-orbiters cannot provide. Operational since 2020.
- Sentinel-5 (planned, ESA/Copernicus): Next-generation operational successor to Sentinel-5P, targeting sub-kilometre NO2 retrieval on a Metop-SG platform. Not yet operational; included here because archive planning for long-term NO2 monitoring should account for the transition.
What the column actually measures, and what it does not
A tropospheric NO2 column is the vertically integrated number density of NO2 molecules between the surface and the tropopause, expressed in molecules per square centimetre (mol/cm²) or in micromoles per square metre. TROPOMI does not measure surface concentration directly. It measures the total sunlight backscattered through the atmosphere, then uses differential optical absorption spectroscopy (DOAS) to isolate the NO2 absorption fingerprint in the 405–465 nm window. The stratospheric contribution, which is globally distributed and relatively stable, is subtracted using a reference sector method over the remote Pacific, leaving the tropospheric residual.
The practical consequence is that the retrieval is a column-weighted average, not a surface reading. Emissions at altitude, from aircraft or elevated stacks, are weighted differently from ground-level traffic. For compliance or epidemiological work, column data must be combined with a chemical transport model to estimate surface concentrations. That step introduces its own uncertainties, typically 20–50% depending on model resolution and meteorological input quality.
Cloud and aerosol: the two enemies of sensitivity
TROPOMI applies a cloud radiance fraction filter. Pixels with an effective cloud fraction above 0.5 are flagged and excluded from most standard analyses. In persistently cloudy regions, such as the maritime tropics or monsoon-affected South and Southeast Asia, this can remove more than half of all daily observations. Monthly compositing recovers spatial coverage but smears episodic emission events.
High aerosol loading, common over dust-laden deserts or heavily polluted urban basins, creates a different problem. Aerosols scatter and absorb UV-visible light in ways that alter the photon path length, biasing the retrieved column. The TROPOMI operational processor applies an implicit aerosol correction through its surface albedo and scattering weight functions, but residual errors of 10–30% are documented in the literature for heavily aerosol-loaded scenes. This is not a reason to discard the data; it is a reason to pair column retrievals with coincident aerosol optical depth observations, such as those from MODIS or VIIRS, when drawing firm conclusions about point-source magnitudes.
Separating a power station from its neighbours
At 3.5 × 5.5 km, TROPOMI can isolate large individual point sources, coal-fired power stations, aluminium smelters, large cement plants, provided they are separated by at least one pixel from other strong emitters. The standard technique is a plume-fitting or mass-balance approach: wind speed and direction from reanalysis data (ERA5 is commonly used) are combined with the observed downwind NO2 enhancement to back-calculate an emission rate in kg NO2 per hour.
Several published studies have applied this to individual facilities globally, cross-referencing against reported emissions inventories. Discrepancies of a factor of two or more between satellite-derived and self-reported figures are not unusual. Shipping lanes are visible as persistent linear enhancements, particularly in the Red Sea, the English Channel and the South China Sea, where vessel density is high and background NO2 is otherwise low. Urban attribution is harder: a city is a diffuse source, and separating traffic from industrial districts from domestic heating requires time-of-day and day-of-week analysis that TROPOMI's single daily overpass cannot supply alone. This is where geostationary instruments like GEMS add genuine value.
Building a usable time series: archive depth and consistency
Sentinel-5P launched in October 2017. The operational Level-2 offline NO2 product (OFFL) is available from May 2018 onward, after the instrument commissioning phase. That gives roughly six years of consistent global data at the time of writing, processed through a versioned algorithm. Processor version changes, most significantly the move to TROPOMI v2.x, altered retrieved columns by a few percent; long trend analyses must account for these steps.
OMI extends the record back to 2004, but the spatial resolution difference (13 × 24 km versus 3.5 × 5.5 km) means the two series are not directly interchangeable for point-source work. They can be merged for regional or national trend analysis with appropriate bias correction. For a government client building a sovereign air-quality monitoring capability, the practical archive question is not just depth but reproducibility: can the same retrieval algorithm be rerun on raw spectra if the science evolves? Copernicus makes the full Level-1 radiance archive available through the Copernicus Data Space Ecosystem, which matters for long-term programme design.
From column to policy: what the data can and cannot support
Tropospheric NO2 columns are well-suited to three regulatory applications: identifying undisclosed or underreported emission sources, tracking the effect of policy interventions over time, and providing independent verification of national emission inventories submitted under international agreements. The COVID-19 lockdowns of 2020 provided an inadvertent global experiment, with TROPOMI clearly showing NO2 reductions over cities within days of mobility restrictions, then rebounds as activity resumed.
What the data cannot support, without additional modelling, is a direct legal emission limit comparison. Regulatory limits are set in surface concentration units (micrograms per cubic metre), not column amounts. The translation requires assumptions about boundary layer height, mixing, and chemistry that introduce uncertainty. A satellite-derived column anomaly is strong evidence that warrants investigation; it is rarely sufficient, on its own, to constitute proof of a specific violation. Satellize's analytics approach treats the column data as a detection and prioritisation layer, flagging facilities or corridors for closer scrutiny rather than as a standalone enforcement instrument.
GEMS and future geostationary instruments will progressively close the diurnal gap. Until then, the single daily TROPOMI overpass is the practical global standard, and its honest limitations should be part of any briefing to a government client.
Typical figures
| Spatial resolution (TROPOMI, post-Aug 2019) | 3.5 × 5.5 km at nadir |
| Swath width | 2,600 km (TROPOMI); daily global coverage in one orbit cycle |
| Revisit (polar orbit) | Daily global; ~14 orbits per day |
| Revisit (GEMS, geostationary) | Hourly daytime over East Asia |
| Spectral window for NO2 retrieval | 405–465 nm (UV-visible DOAS) |
| Cloud fraction filter threshold | Pixels with effective cloud fraction > 0.5 excluded from standard products |
| Product latency (TROPOMI OFFL) | Typically < 3 hours after sensing |
| Archive depth (Sentinel-5P operational) | May 2018 to present; OMI heritage extends to 2004 |
| Minimum detectable column enhancement | ~0.1–0.5 × 10¹⁵ mol/cm² over low-background regions (instrument noise floor; scene-dependent) |
| Delivery formats (Copernicus) | NetCDF-4 (Level-2 OFFL and NRTI); accessible via Copernicus Data Space Ecosystem |
Analytics Satellize can run
| Industrial point-source emission rate | Plume mass-balance: TROPOMI column enhancement integrated downwind, combined with ERA5 wind reanalysis | Per-facility kg NO2/hour estimate with uncertainty range, delivered as a structured report or GIS point layer |
| Shipping-lane NO2 flux map | Persistent enhancement compositing over 30-day windows; linear feature extraction aligned to vessel routing corridors | Monthly raster and vector overlay of lane-level NO2 burden, suitable for port authority or IMO reporting workflows |
| Urban NO2 trend series | Monthly median compositing to overcome cloud gaps; Mann-Kendall trend test on annual aggregates | City-level time-series chart and tabular data, 2018 to present, with processor-version change points annotated |
| Policy intervention signal detection | Before/after column comparison with meteorological normalisation using wind and boundary-layer height reanalysis | Briefing-ready summary showing statistically significant column change (or absence of one) following a regulatory event |
| Emission inventory cross-check | Satellite-derived facility emission rates compared against national inventory submissions; discrepancy flagging | Ranked list of facilities where satellite-derived figures diverge from reported figures by more than a specified threshold |
| High-aerosol scene quality flag | Co-registration of MODIS or VIIRS aerosol optical depth with TROPOMI pixel quality flags | Quality-screened NO2 layer with aerosol-contaminated pixels masked or uncertainty-weighted, as GeoTIFF or NetCDF |
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.