Tropospheric NO₂ column mapping for urban and industrial emission inventories
TROPOMI and OMI retrieve tropospheric NO₂ columns daily at city-to-complex scale, giving regulators and industry an independent check on ground-based emission inventories. Retrieval physics, honest resolution limits, and cloud-screening trade-offs explained.
Sensors
- TROPOMI / Sentinel-5P: UV-Vis-NIR-SWIR push-broom spectrometer; 3.5 × 5.5 km nadir pixel (upgraded from 3.5 × 7 km in August 2019); daily global coverage; NO₂ slant columns retrieved by DOAS in the 405–465 nm window; operational Level-2 product latency approximately 3 hours after observation.
- OMI / Aura: Hyperspectral UV-Vis CCD; 13 × 24 km pixel at nadir, degraded by row-anomaly masking since 2007; daily near-global swath; the longest continuous satellite NO₂ record from 2004, enabling trend analysis over nearly two decades.
- GEMS / GEO-KOMPSAT-2B: Geostationary UV-Vis spectrometer covering East Asia (5°S–45°N, 75°–145°E); sub-hourly scans (roughly every 30–60 minutes during daylight); 3.5 × 8 km pixel; first geostationary air-quality sensor in routine operation, allowing diurnal NO₂ cycle resolution.
- TEMPO / GOES-R host: Geostationary UV-Vis spectrometer for North America; sub-hourly daytime scans planned; sub-2 km pixel resolution anticipated; launched April 2023 and undergoing commissioning as of the knowledge cut-off. Will provide the first hourly NO₂ mapping over North American urban corridors.
What the spectrometer actually measures, and what it does not
Every TROPOMI pixel records the intensity of sunlight that has been scattered back through the atmosphere in the 405–465 nm window. NO₂ molecules absorb at characteristic wavelengths within that window. Differential Optical Absorption Spectroscopy (DOAS) isolates those fingerprint absorptions from the broadband background, yielding a slant column density: the total NO₂ integrated along the effective photon path through the atmosphere, not along a vertical line.
Converting a slant column to a vertical column density requires an air-mass factor (AMF). The AMF depends on solar zenith angle, viewing geometry, surface albedo, aerosol loading, and the assumed vertical profile shape of NO₂ itself. That profile is typically taken from a chemical transport model such as TM5 or GEOS-Chem. When the model profile disagrees with reality, for instance over a city where near-surface concentrations are far higher than the model predicts, the AMF can be wrong by 30–50 percent. This is not a minor footnote. It is the dominant uncertainty in comparing satellite columns to ground-based measurements.
Resolution: what TROPOMI can resolve, and what it cannot
At 3.5 × 5.5 km, a TROPOMI pixel is small enough to separate a large refinery from a neighbouring residential district, or to distinguish port shipping lanes from an adjacent industrial estate. A steel mill or a large power station occupying several square kilometres will produce a detectable signal above background if its emissions are strong enough.
Street-level gradients are out of reach. The difference in NO₂ between a busy arterial road and a parallel side street, relevant to hyperlocal health exposure studies, sits below the pixel floor. Attributing a column anomaly to a single stack rather than to a cluster of sources within the pixel requires additional dispersion modelling or higher-resolution airborne data. OMI, with its 13 × 24 km nadir pixel, is coarser still; it remains useful for city-aggregate trends over years but cannot resolve individual industrial complexes.
Cloud cover: the tax that tropical and monsoon regions pay
DOAS retrieval requires photons that have passed through the troposphere. Cloud decks block that path. Standard practice is to screen pixels with a scene cloud radiance fraction above 0.5, though some applications use a stricter 0.3 threshold. In tropical regions and during monsoon seasons, this removes roughly 30–50 percent of daily observations. Over equatorial industrial zones such as parts of Indonesia or Central Africa, cloud-free data may be available on fewer than half the days in a given month.
Temporal averaging is the practical remedy. Monthly composites built from cloud-screened daily passes recover spatial patterns reliably, because emission sources are persistent while cloud occurrence is not perfectly correlated with emission events. Annual means are used for trend detection and inventory validation. The trade-off is that episodic events, a plant running at unusually high load for a week, or a sudden shutdown, may be diluted into the monthly signal unless the cloud-free window happens to coincide.
From columns to emission rates: the mass-balance and inverse approaches
A tropospheric NO₂ column map is not the same as an emission inventory. To get from one to the other, analysts use two main approaches. The mass-balance method integrates the NO₂ column enhancement downwind of a source, applies an assumed or modelled chemical lifetime for NO₂ (typically 4–8 hours in daytime summer conditions, shorter in winter when photolysis is slower), and back-calculates a surface flux. The method works reasonably well for large isolated point sources but becomes ambiguous where multiple sources overlap within a few tens of kilometres.
Inverse modelling feeds satellite columns into a chemical transport model and adjusts prior emission estimates until the model reproduces the observed columns. Studies using OMI and TROPOMI have revised national NOₓ inventories for China, India and parts of Europe, in some cases finding that official bottom-up inventories underestimated emissions from specific sectors by factors of two or more, and in other cases finding the reverse. The quality of the inversion depends heavily on the model's meteorological inputs and on the AMF uncertainty discussed above. Neither method produces emission rates with the precision of a continuous stack monitor; the honest figure for large-source annual totals is an uncertainty range of roughly ±20–40 percent.
Geostationary sensors change the diurnal picture
GEMS over East Asia has demonstrated something that polar-orbiters cannot: the NO₂ column over a city rises through the morning rush, peaks around midday as photochemistry and traffic interact, then evolves through the afternoon in ways that depend on boundary-layer mixing and local meteorology. A single daily overpass from TROPOMI samples this cycle at one moment, typically around 13:30 local solar time. That snapshot can be unrepresentative on days with unusual meteorology.
TEMPO, once its calibration is complete, will offer the same diurnal resolution for North American cities. For regulators trying to attribute column anomalies to morning commuter traffic versus daytime industrial activity, the hourly cadence is genuinely informative in ways that daily revisit is not. The caveat is that geostationary sensors have fixed fields of view: GEMS covers East Asia, TEMPO covers North America. Other regions still depend on polar-orbiting assets.
Building an inventory: what satellite data can and cannot replace
Governments and large emitters use satellite NO₂ columns for three distinct tasks: independent verification of bottom-up emission inventories, detection of unreported or under-reported sources, and long-term trend monitoring to assess whether abatement measures are working. TROPOMI's archive from 2018 onward, combined with OMI's record from 2004, gives a nearly two-decade time series that predates many current regulatory regimes.
Satellite columns cannot replace stack monitors, dispersion measurements, or facility-level reporting. They add an independent, spatially continuous layer that is difficult to manipulate and covers sources that ground networks do not reach. The combination is more informative than either alone. Satellize runs operational NO₂ column analytics on open TROPOMI data, including monthly and annual composite products and source-attribution workflows, as part of its broader atmospheric monitoring offering alongside the crop-estimation work it conducts for the Kingdom of Tonga. For any client building a national or regional emission inventory, the practical starting point is a baseline audit: which facilities show persistent column enhancements, and how do those enhancements compare to declared emission figures?
Typical figures
| Nadir pixel size (TROPOMI) | 3.5 × 5.5 km (since August 2019 upgrade) |
| Nadir pixel size (OMI) | 13 × 24 km (unaffected rows; row anomaly reduces usable swath) |
| Revisit (TROPOMI / OMI) | Daily global coverage (polar-orbiting, ~13:30 local solar time overpass) |
| Revisit (GEMS) | Sub-hourly daytime scans over East Asia (approx. every 30–60 minutes) |
| Spectral retrieval window | 405–465 nm (DOAS NO₂ fit window, UV-Vis) |
| Cloud screening threshold (standard) | Scene cloud radiance fraction < 0.5; strict applications use < 0.3 |
| Effective data loss to cloud (tropics) | Roughly 30–50% of daily pixels screened in tropical and monsoon regions |
| Level-2 product latency (TROPOMI) | Approximately 3 hours after observation (near-real-time stream) |
| Archive depth | TROPOMI from May 2018; OMI from October 2004 |
| Emission-rate retrieval uncertainty (large point sources) | Approximately ±20–40% for annual totals using mass-balance or inversion methods |
Analytics Satellize can run
| Monthly tropospheric NO₂ column composite | Cloud-screened TROPOMI Level-2 aggregation with quality flag filtering (qa_value > 0.75) | GeoTIFF and GIS layer at 0.01° grid; monthly cadence |
| Annual trend analysis per facility or administrative zone | Linear regression over TROPOMI and OMI time series; Mann-Kendall significance test | Time-series report with trend magnitude, confidence interval, and change-point flags |
| Persistent anomaly detection (unreported or under-reported sources) | Residual mapping against background field; cluster analysis of column enhancements exceeding 2σ above regional mean | Ranked facility watchlist with column enhancement values and spatial footprints |
| Emission-rate estimate for large point sources | Downwind mass-balance integration using TROPOMI columns and ERA5 wind fields; NO₂ lifetime parameterised by season and latitude | Tabular emission estimates (mol s⁻¹ and kt NO₂ yr⁻¹) with stated uncertainty range |
| Inventory cross-check report | Comparison of satellite-derived column enhancements to facility-declared NOₓ figures from national registries; discrepancy flagging | PDF audit report with facility-level comparison table and map of priority discrepancies |
| Diurnal NO₂ cycle characterisation (East Asia / North America) | GEMS or TEMPO sub-hourly column time series; boundary-layer mixing height correction using ERA5 or MERRA-2 | Hourly column animation and peak-hour attribution summary for named urban or industrial zones |
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.