Tropospheric ozone burden mapping for surface-level health risk
Tropospheric ozone forms in sunlight from NOx and VOC precursors and peaks in afternoon hours, when respiratory risk is highest. Geostationary sensors TEMPO and GEMS now resolve this diurnal cycle hourly, enabling city-scale exposure mapping that was impossible with daily LEO overpasses alone.
Sensors
- TEMPO (Tropospheric Emissions: Monitoring of Pollution): NASA geostationary instrument on Intelsat 40E, operational since 2023. Scans North America roughly hourly at approximately 10 km × 10 km nadir resolution. Retrieves tropospheric ozone columns in the UV (305–490 nm range), enabling diurnal build-up curves across urban basins for the first time from orbit.
- GEMS (Geostationary Environment Monitoring Spectrometer): Korean Meteorological Administration instrument on GEO-KOMPSAT-2B, operational since 2020. Covers East and Southeast Asia at approximately 7 km × 8 km resolution, with hourly scans. Provides the Asian counterpart to TEMPO's North American coverage, capturing afternoon ozone peaks over Seoul, Beijing and the Indo-Gangetic Plain.
- Sentinel-5P TROPOMI: ESA/Copernicus LEO instrument, launched 2017. Daily global coverage at 5.5 km × 3.5 km (post-2019 update). Tropospheric ozone retrieved via the CCD algorithm. One overpass per day limits diurnal tracking, but the archive depth (2018 to present) and global reach make it the backbone for trend analysis and inter-calibration with GEO sensors. Known positive bias over high-albedo surfaces (desert, snow) of order 5–15 DU in published validation studies.
- OMI (Ozone Monitoring Instrument, Aura): NASA LEO instrument, operational since 2004. Spatial resolution approximately 13 km × 24 km, daily global coverage. Now affected by row-anomaly data gaps, but its 20-year archive is irreplaceable for long-term tropospheric ozone trend attribution and seasonal climatology.
Why the afternoon hour matters more than the daily average
Tropospheric ozone is not emitted directly. It assembles photochemically from nitrogen oxides and volatile organic compounds in the presence of sunlight, which means concentrations climb through the morning and typically peak between roughly 14:00 and 17:00 local time. A single daily satellite overpass, taken at 13:30 local time for Sentinel-5P, happens to catch the system near its peak on many days, but it cannot reveal the rate of build-up, the timing of the peak, or the speed of evening decay. Those details matter clinically: the WHO air quality guideline for ozone is expressed as a peak 8-hour mean, not a daily average, and asthma and COPD exacerbation rates track the afternoon window closely.
TEMPO changed this. By sitting in geostationary orbit and scanning North America approximately once per hour, it produces a time-lapse of ozone column density across the continent. Cities in valley basins, where topography traps precursors, show particularly sharp diurnal signals. The same logic applies to GEMS over Asia. Together, the two instruments cover the two largest regional source areas for anthropogenic ozone precursors on the planet.
The column-to-surface problem: what the photochemical box model does
Satellites measure ozone columns, integrated through the depth of the atmosphere. Surface mixing ratios, the quantity that matters for a person breathing at street level, are not the same thing. Converting one to the other requires a photochemical box model: a constrained set of chemical equations that describes how ozone, NOx, VOCs, and OH radicals interact under a given temperature, humidity, and photolysis rate. The model uses the satellite column as a top-level constraint and ingests ancillary inputs, typically reanalysis meteorology from ERA5 or MERRA-2, to partition the column between the boundary layer and the free troposphere.
The approach has real uncertainty. Boundary layer height is the largest single source of error: on days when the mixing layer is shallow, a given column maps to a much higher surface concentration than on a well-mixed day. Published studies using TROPOMI columns with WRF-Chem or similar models report surface ozone estimates with root-mean-square errors of roughly 8–15 ppb against ground monitors in urban areas. That is adequate for identifying high-risk days and ranking cities by exposure burden, but it is not a substitute for a dense ground-monitoring network. Honest use of the method means communicating that uncertainty to decision-makers, not hiding it in a colour scale.
The high-albedo bias and where it bites
TROPOMI's tropospheric ozone retrieval relies on the contrast between UV radiance at wavelengths inside and outside the Hartley-Huggins absorption bands. Over dark surfaces, the geometry works well. Over bright surfaces, particularly desert sand, salt flats, and snow, the increased backscattered UV reduces the sensitivity of the retrieval to the lower troposphere and introduces a positive bias. Published validation work has quantified this at roughly 5–15 Dobson Units over the Sahara and Arabian Peninsula, and a smaller but non-negligible bias over snow-covered boreal regions.
The practical consequence is that ozone burden estimates over arid cities, think Phoenix in summer or Riyadh year-round, should be treated with extra caution when derived from TROPOMI alone. TEMPO's retrieval algorithm incorporates surface reflectance databases that partially mitigate this, but the physics of the problem does not disappear. Cross-validation against ground-based Dobson spectrophotometers or Pandora spectrometers, where they exist, is the correct check.
From column maps to public health outputs
The epidemiological literature links short-term ozone exposure to emergency department visits for asthma, hospital admissions for COPD, and all-cause respiratory mortality. The concentration-response functions from large multi-city studies allow satellite-derived surface estimates to be converted into attributable health burden, expressed as excess cases per 100,000 population per season or per year. This is the step that turns a geophysical data product into something a health ministry can act on.
Useful outputs include daily peak-8-hour ozone maps at city or district scale, seasonal exposure climatologies for chronic-disease burden modelling, and near-real-time alerts when modelled surface concentrations are forecast to exceed national standards. The alert application depends on latency: TEMPO Level-2 data are typically available within a few hours of observation, which is fast enough for same-day public advisories. GEMS operates on a similar schedule.
Satellize runs this kind of column-to-surface analytics pipeline on open constellations, combining TROPOMI archive depth with TEMPO's diurnal resolution for clients who need both trend context and operational alerting.
What satellite ozone mapping cannot do
No current satellite instrument resolves ozone at the block or street level. The finest operational retrievals are at 5–10 km, which means intraurban gradients driven by local traffic patterns or park greenery are invisible. Ozone is also chemically titrated by fresh NO emissions near busy roads, producing a counterintuitive dip in ozone concentration directly at the kerbside even as the regional burden is high. Satellite data captures the regional signal, not this local suppression.
Cloud cover is a persistent limit. TROPOMI and the GEO instruments require clear-sky or near-clear-sky conditions for valid retrievals. In persistently cloudy regions, such as the maritime tropics, data gaps can span weeks. Compositing strategies help for climatological products but are not useful for acute-episode alerting. And the photochemical box model step introduces its own assumptions: the results are only as good as the meteorological reanalysis driving them, which degrades in regions with sparse radiosonde coverage.
Typical figures
| Finest operational spatial resolution (GEMS) | Approximately 7 km × 8 km at nadir |
| Finest operational spatial resolution (TROPOMI, post-2019) | 5.5 km × 3.5 km |
| Diurnal revisit (TEMPO / GEMS) | Approximately hourly over North America and Asia respectively |
| LEO revisit (TROPOMI / OMI) | Once per day, global |
| Spectral bands used for ozone retrieval | UV 305–340 nm (Hartley-Huggins bands); some algorithms extend to 490 nm |
| Typical column retrieval uncertainty (TROPOMI tropospheric ozone) | 6–8 DU random; +5 to +15 DU systematic bias over high-albedo surfaces |
| Surface mixing ratio RMSE (box-model step, urban areas) | Approximately 8–15 ppb against ground monitors in published studies |
| Data latency for near-real-time products | TEMPO and GEMS Level-2: typically 2–4 hours after observation; TROPOMI offline: ~3 hours |
| Archive depth | OMI: 2004 to present; TROPOMI: 2018 to present; GEMS: 2020 to present; TEMPO: 2023 to present |
| Cloud sensitivity limit | Retrievals flagged at cloud radiance fraction above approximately 0.5; dense cloud produces data voids |
Analytics Satellize can run
| Diurnal ozone build-up curves by urban basin | Hourly TEMPO or GEMS Level-2 column aggregation over defined city polygons, with time-of-peak extraction | Daily time-series chart and GeoTIFF showing peak-hour column density, delivered by 18:00 local time |
| Modelled surface ozone mixing ratio maps | Photochemical box model constrained by satellite columns and ERA5 boundary-layer height; published HCHO/NO2 ratio approach for VOC-sensitivity regime classification | Gridded NetCDF and clipped GeoTIFF at sensor native resolution, with uncertainty band layer |
| Peak 8-hour ozone exceedance alert | Rolling 8-hour mean of modelled surface ozone compared against national or WHO threshold; triggered when exceedance probability exceeds configurable confidence level | JSON alert feed and optional SMS/email bulletin for health authority distribution |
| Seasonal ozone exposure climatology | TROPOMI multi-year archive composited by month and season; box-model applied to monthly median columns; population-weighted exposure calculated against gridded population data | Annual PDF report with district-level exposure rankings and concentration-response burden estimates |
| High-albedo bias flagging and corrected column product | Surface reflectance database (OMI/MODIS climatology) used to identify retrievals over bright surfaces; bias-correction factor applied from published validation literature | Quality-flagged raster with bias-corrected column values and per-pixel confidence score |
| Long-term tropospheric ozone trend attribution | OMI and TROPOMI archive merged using inter-calibration offsets; Mann-Kendall trend test applied per grid cell; precursor emission inventory comparison for source attribution | Decadal trend map (ppb per year) with statistical significance layer, formatted for policy briefing |
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.