Volcanic SO2 emission and plume dispersion tracking
Sentinel-5P TROPOMI and Aura OMI detect sulphur dioxide by its ultraviolet absorption signature, tracking plume mass and trajectory from eruption onset through dispersal. This page covers detection thresholds, ash interference, dispersion modelling, and the latency that separates a satellite overpass from an operational alert.
Sensors
- Sentinel-5P TROPOMI: UV-VIS spectrometer covering 312–405 nm; SO2 column retrieved at 3.5 × 5.5 km pixel (upgraded to 3.5 × 3.5 km after August 2019 processor update); daily global coverage; detection sensitivity below 0.5 DU for large eruption columns, though passive degassing at 0.1–0.3 DU is near the noise floor in the mid-troposphere retrieval.
- Aura OMI: UV nadir sounder at 13 × 24 km pixel; daily near-global coverage; operational since 2004, providing the longest continuous SO2 satellite record. Row anomaly since 2008 reduces effective swath by roughly 25 %. Useful for trend analysis and cross-calibration with TROPOMI.
- GOES-16/17 ABI: Geostationary imager; no dedicated SO2 channel, but the 8.5 µm and 11 µm thermal infrared bands detect SO2 absorption indirectly and track plume geometry at 2 km resolution and 5–10 minute cadence. Primary value is high-temporal tracking of plume movement after TROPOMI overpass.
- Meteosat SEVIRI: European geostationary imager with 15-minute full-disc repeat; 8.7 µm band sensitive to SO2 in high-altitude plumes. Covers Atlantic, Europe, Africa and Indian Ocean arc. Complements TROPOMI for inter-overpass plume evolution in the Eastern Hemisphere.
What the ultraviolet actually measures
Sulphur dioxide absorbs ultraviolet radiation in the 300–330 nm range with a distinctive cross-section that satellite spectrometers can separate from ozone, aerosol and surface reflectance. TROPOMI's principal SO2 product reports the total vertical column in Dobson Units, with separate retrievals optimised for SO2 centred at the boundary layer (roughly 0–1 km), the lower troposphere (around 2.5 km) and the upper troposphere and stratosphere (around 7 km). The choice of retrieval height matters enormously: a plume injected to 15 km by an explosive eruption will be badly underestimated if the boundary-layer retrieval is applied, and vice versa for a passively degassing vent.
The retrieval also assumes that SO2 and volcanic ash occupy the same altitude. In practice they often do not. Ash can absorb and scatter UV, biasing the SO2 column high or low depending on geometry. ESA's TROPOMI SO2 product documentation flags pixels with high aerosol index as potentially contaminated. Analysts should treat SO2 retrievals in dense ash columns as uncertain by a factor of two or more, and cross-check against the ash cloud product (covered in the sibling page on aviation hazard).
Detection limits: what TROPOMI can and cannot see
For large explosive eruptions injecting SO2 into the upper troposphere or stratosphere, TROPOMI is highly capable. The 2022 Hunga Tonga–Hunga Ha'apai eruption produced an SO2 mass estimated at roughly 0.4–0.5 Tg, clearly resolved in a single overpass. The 2021 La Palma eruption was tracked daily throughout its 85-day duration. These are the easy cases.
Passive degassing is harder. Many persistently active volcanoes emit 100–1,000 tonnes of SO2 per day. At the 3.5 km pixel scale, a 500 t/day source spread across a boundary-layer plume can sit below 0.3 DU per pixel, close to TROPOMI's single-pixel noise level. Ground-based DOAS networks and miniaturised SO2 cameras resolve this better than any current satellite. TROPOMI adds value for passive emitters primarily through temporal averaging across many overpasses, which can extract a statistically significant signal from the noise over weeks. Do not expect a satellite-only system to catch a volcano that is quietly ramping up emissions by 50 t/day.
From column amount to plume trajectory: dispersion modelling
A satellite SO2 column map tells you where the gas was at overpass time. To forecast where it will be in six or twenty-four hours, you need a transport model initialised with the satellite-derived source term: emission rate, injection height and start time. The two models most commonly used in operational volcanic aviation and civil protection contexts are NOAA's HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) and the UK Met Office NAME (Numerical Atmospheric-dispersion Modelling Environment). Both ingest meteorological reanalysis or forecast fields and propagate a particle ensemble forward or backward in time.
Injection height is the dominant uncertainty. It is typically estimated from satellite-derived plume top temperature (using thermal IR from GOES or SEVIRI) or from ground-based radar where available. An error of 2 km in injection height can shift a forecast plume position by hundreds of kilometres after 24 hours, given wind shear. Ensemble runs that sample a range of injection heights are standard practice in well-resourced volcanic observatories, though they are computationally heavier. The satellite source term constrains total SO2 mass; the meteorological model does the rest.
Latency: the gap between overpass and alert
TROPOMI's standard SO2 product is available within three hours of sensing under the Copernicus near-real-time (NRT) processing stream. The offline product, with better calibration, arrives within one to two days. Three hours is fast enough for strategic civil protection decisions, but not for immediate aviation rerouting, where SIGMET issuance typically needs information within 30–60 minutes of a significant eruption.
Geostationary sensors close that gap. GOES-16 ABI and Meteosat SEVIRI both produce SO2-sensitive imagery at 10–15 minute cadence with latency under 15 minutes. The trade-off is that geostationary SO2 retrievals are noisier and less quantitatively reliable than TROPOMI, particularly for low-altitude plumes where the signal-to-noise ratio is poor. The operational workflow for a serious eruption therefore layers the two: geostationary for rapid detection and trajectory tracking, TROPOMI for accurate mass quantification once the overpass arrives.
Practical limits that buyers should price in
Cloud cover does not block UV retrievals the way it blocks optical imagery, but thick cloud tops above the SO2 layer can still shield the column from the satellite, reducing retrieved amounts. Stratospheric plumes above most cloud are essentially unaffected. Boundary-layer plumes under a deep convective system may be largely invisible.
Archive depth is a genuine asset. OMI data runs from 2004; TROPOMI from late 2017. Together they provide roughly two decades of SO2 column records for any volcano on Earth, enough to characterise seasonal degassing patterns, detect anomalous pre-eruptive SO2 increases and build baseline emission inventories for national hazard assessments. Satellize runs analytics on both archives under the open Copernicus and NASA data licences. Our work tends to be most useful when a government agency needs those records processed into decision-ready products rather than raw NetCDF files, which is broadly the same logic behind our crop-estimation programme for the Kingdom of Tonga.
What a monitoring programme actually delivers
A well-structured volcanic SO2 monitoring service has three layers. The first is automated detection: an alert triggered when SO2 column amounts in a defined watch-zone exceed a threshold, delivered within the latency of the NRT product. The second is quantification: a daily or per-overpass SO2 mass estimate with uncertainty bounds, logged to a time series that feeds into eruption intensity assessments. The third is trajectory forecasting: a HYSPLIT or NAME ensemble run, initialised from the satellite source term, producing 24- and 48-hour plume position probability maps for civil aviation and civil protection authorities.
None of these layers requires bespoke satellite hardware. They run on open data from TROPOMI and OMI, supplemented by geostationary imagery for inter-overpass continuity. The cost is in the processing pipeline, the meteorological data feeds, the model infrastructure and the people who interpret ambiguous retrievals sensibly. That is where the value sits, and where generic automated products most often fall short.
Typical figures
| TROPOMI SO2 pixel size | 3.5 × 3.5 km (post-August 2019 processor); 3.5 × 5.5 km earlier data |
| OMI SO2 pixel size | 13 × 24 km (nadir); larger off-nadir; row anomaly reduces usable swath |
| Revisit (TROPOMI) | Daily global coverage; single overpass per day at any given latitude |
| Geostationary cadence (GOES/SEVIRI) | 5–15 minutes; latency under 15 minutes for SO2-sensitive thermal IR imagery |
| NRT product latency (TROPOMI) | Typically within 3 hours of sensing; offline product within 1–2 days |
| Minimum detectable SO2 column (large eruption, upper troposphere) | Below 0.5 DU per pixel for TROPOMI; passive degassing near noise floor at 0.1–0.3 DU |
| Spectral bands used | UV 312–405 nm (TROPOMI/OMI SO2 retrieval); 8.5–11 µm thermal IR (GOES/SEVIRI plume tracking) |
| Archive depth | OMI from October 2004; TROPOMI from November 2017; combined ~20-year record |
| Dispersion model outputs | HYSPLIT or NAME ensemble; 24- and 48-hour plume position probability fields |
| Delivery formats | NetCDF (column maps), GeoTIFF (plume masks), GeoJSON (trajectory polygons), PDF briefing |
Analytics Satellize can run
| SO2 mass and emission rate time series | TROPOMI NRT column retrieval integrated over plume extent; OMI cross-calibration for trend continuity | Daily CSV and chart showing SO2 mass (Tg or kt) per overpass, with uncertainty range, delivered to client dashboard |
| Automated eruption SO2 alert | Threshold detection on TROPOMI NRT SO2 column within a user-defined volcano watch-zone polygon | Push alert (email or webhook) within 3 hours of TROPOMI overpass when column exceeds agreed threshold |
| Plume trajectory forecast | HYSPLIT ensemble dispersion model initialised with satellite-derived emission rate and injection height from geostationary plume-top temperature | 24- and 48-hour probability plume polygons as GeoJSON and PDF map, issued per overpass during active eruption |
| Inter-overpass plume position tracking | GOES-16 ABI or Meteosat SEVIRI 8.5/11 µm brightness temperature difference; plume boundary digitised at 10–15 minute cadence | Hourly GeoTIFF plume mask and animated GIF for civil protection situation reports |
| Baseline degassing emission inventory | Multi-year TROPOMI and OMI SO2 column averaging over defined volcanic source regions; statistical extraction of persistent low-level signal from noise | Annual SO2 emission inventory report per volcano, with seasonal breakdown and anomaly flags, in PDF and spreadsheet |
| Ash-SO2 co-location uncertainty assessment | Cross-comparison of TROPOMI SO2 column with aerosol index and volcanic ash retrieval products; flagging of pixels where ash contamination likely biases SO2 by more than 50 % | Annotated GeoTIFF with confidence classification layer; included in eruption situation briefings |
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.