Thermal infrared volcanic unrest and lava flow mapping
Volcanic heat overwhelms broad-band thermal sensors quickly, so effective monitoring stacks SWIR, mid-wave and thermal infrared data from ASTER, VIIRS, Sentinel-2 and ECOSTRESS to track effusion rates, dome growth and crater-lake unrest without saturation artefacts.
Sensors
- ASTER (Terra): Five thermal infrared bands at 90 m resolution, two SWIR bands at 30 m (now degraded on-orbit), and a VNIR panchromatic at 15 m. Revisit is roughly 16 days at the equator but can be tasked more frequently at higher latitudes. TIR bands 10–14 cover 8–12 µm; SWIR bands 4–9 cover 1.6–2.5 µm and were the primary unsaturated lava tool before the SWIR detector cooler failed in 2008.
- VIIRS (NOAA-20 and Suomi-NPP): The I4 band at 3.74 µm and M13 band at 4.05 µm provide mid-wave infrared at 375 m and 750 m respectively, with near-daily global coverage. Saturation occurs at roughly 343 K in I4 under standard gain, but high-temperature events trigger an on-board dual-gain switch that extends dynamic range. Useful for detecting active lava and tracking radiative power at effusive eruptions when spatial detail is secondary.
- Sentinel-2 MSI (ESA): Bands 11 (1.61 µm) and 12 (2.19 µm) at 20 m resolution provide SWIR coverage that remains unsaturated for most lava flow surfaces cooler than roughly 800–1000 K. Revisit is 5 days at the equator with both satellites. Not a thermal imager in the strict sense, but SWIR radiance from active lava is detectable well below saturation thresholds, making it the workhorse for lava flow front delineation at moderate spatial scale.
- ECOSTRESS (ISS): A five-band thermal infrared spectrometer covering 8–12.5 µm at roughly 70 m resolution, mounted on the International Space Station. Revisit is irregular, between 1 and 5 days depending on ISS orbital precession, and coverage is limited to latitudes below about 52°. Designed for evapotranspiration, but its finer thermal resolution compared to ASTER makes it useful for mapping crater-lake temperature anomalies and cooling lava fields.
- Landsat 8 and 9 TIRS: Two thermal bands (Band 10 at 10.9 µm, Band 11 at 12.0 µm) at 100 m resolution resampled to 30 m. Sixteen-day revisit per satellite, eight days combined. Saturation occurs at high volcanic temperatures, but the OLI SWIR bands (1.6 µm and 2.2 µm) at 30 m remain useful for lava detection in the same manner as Sentinel-2, and the archive extends to 1972 for historical context.
Why your thermal sensor lies at high temperatures
Broad-band thermal infrared detectors are calibrated for Earth surface temperatures, typically 200–340 K. Active lava at a flow front sits between 1000 and 1400 K. That is not a measurement challenge; it is a saturation event. A pixel containing even a small fraction of incandescent material will read at the sensor ceiling, giving you a number that tells you nothing about how hot or how large the source actually is.
The standard fix is to move to shorter wavelengths. At 2.2 µm, the Planck function for a 1200 K surface still produces a radiance that falls within the dynamic range of a SWIR detector. Sentinel-2 Band 12 and Landsat OLI Band 7 exploit this. They were not designed for volcanology, but they are now central to it precisely because they do not saturate where TIR sensors do. The trade-off is that cooler volcanic features, such as a warming crater lake at 320 K or a fumarole field at 380 K, produce negligible SWIR signal and require proper TIR observation.
Stacking sensors to cover the full temperature range
No single sensor covers the full thermal range of volcanic activity, from a 295 K baseline lake warming by two degrees to a 1300 K lava fountain. Effective monitoring requires deliberate stacking. VIIRS I4 and M13 mid-wave bands catch the hottest, largest events at coarse resolution and near-daily cadence. Sentinel-2 SWIR at 20 m resolves flow front geometry and detects moderate-temperature lava. ASTER TIR at 90 m, when tasked, provides calibrated multi-band thermal data for temperature retrieval using the temperature-emissivity separation algorithm. ECOSTRESS adds finer thermal detail at 70 m for crater lakes and cooling fields.
The combination is not elegant; it requires co-registration across sensors with different viewing geometries, overpass times and projection conventions. But the physics demands it. The dual-band method, fitting observed SWIR radiances to a two-component mixture of hot lava and cooler crust, can estimate both sub-pixel lava temperature and fractional area. Studies of Etna and Kilauea eruptions have used this approach to derive effusion rates from Landsat and ASTER data, with uncertainties typically in the range of 20–40% depending on assumed lava rheology and crust emissivity.
What a floating roof gives away: crater lake thermometry
Crater lakes are among the most sensitive indicators of shallow magmatic intrusion. A rise of two or three degrees across a lake surface that spans several hectares represents a substantial increase in total heat flux, often preceding phreatic explosions by days to weeks. ASTER and ECOSTRESS can both resolve lake surfaces at volcanoes where the water body exceeds roughly 100 m in diameter. For smaller lakes, pixel mixing with surrounding cold rock suppresses the apparent temperature and the anomaly can be missed entirely.
This is an honest constraint. A fumarole field with individual vents spaced at 5–10 m intervals is invisible to any of these sensors at their operational resolutions. You will see an integrated thermal anomaly if the field is large enough and hot enough to warm the surrounding ground, but you cannot map individual fumaroles from orbit with current open-access sensors. Airborne or ground-based thermal cameras remain necessary for that level of detail.
Tracking lava flow advance and effusion rate
Lava flow mapping from SWIR imagery is now operationally mature. The Copernicus Emergency Management Service activates Sentinel-2 tasking during major eruptions, and the MIROVA system (Moderate Resolution Imaging Volcanic Activity) processes VIIRS and MODIS data in near-real-time to produce volcanic radiative power estimates. Radiative power, measured in megawatts, correlates with effusion rate through empirically derived relationships, though the relationship varies with lava composition and flow morphology.
Flow front advance rate can be estimated by differencing successive SWIR classifications. At Sentinel-2's 20 m resolution and 5-day revisit, you can detect a flow that has advanced by more than one pixel width between passes. Faster-moving flows on steep slopes, such as those on Nyiragongo or the 2021 La Palma eruption, can travel kilometres in hours and may require VIIRS or even geostationary MSG SEVIRI data at 15-minute intervals to track continuously. Geostationary sensors trade spatial resolution (roughly 3 km at best for SEVIRI thermal bands) for the temporal density needed to catch rapid advance.
Dome growth at silicic volcanoes presents a different problem. Lava domes are often obscured by ash clouds and degassing plumes exactly when they are most active. SAR can see through cloud, but thermal emission from a dome surface is the primary indicator of extrusion rate. Cloud-free SWIR acquisitions may be infrequent. At persistently cloudy volcanoes in the tropics, weeks can pass without a usable optical or SWIR overpass.
Archive depth, latency and what to expect operationally
ASTER data extends from 1999 and is freely available via NASA Earthdata. VIIRS begins in 2012 with Suomi-NPP, 2018 with NOAA-20. Sentinel-2 archive starts in 2015 for most volcanic regions. Landsat provides the longest continuous thermal record, from 1982 with Landsat 4 TM. For a volcano with no historical monitoring, this archive depth is genuinely useful: you can reconstruct decades of thermal anomaly history before any ground instrument was deployed.
Operational latency varies. VIIRS-based volcanic radiative power products from MIROVA are typically available within a few hours of overpass. Sentinel-2 Level-1C data appears in the Copernicus Data Space within 24 hours of acquisition. ASTER requires a tasking request and processing time; expect one to three days. ECOSTRESS latency from the ISS is variable but generally within 24 hours for standard products.
Satellize runs thermal anomaly time series on Sentinel-2 SWIR and VIIRS for clients who need persistent monitoring across multiple volcanic regions, combining open-data pipelines with commercial tasking for high-priority acquisitions. The Tonga programme demonstrated how small-island Pacific contexts, where volcanic and agricultural monitoring overlap, can be served from the same data infrastructure.
Honest limits before you commission a programme
Spatial resolution is the binding constraint for fumarolic and diffuse degassing studies. Nothing in the open sensor stack resolves features below 20 m, and 90 m is more typical for calibrated thermal data. Cloud cover at tropical stratovolcanoes can reduce usable acquisition frequency to a handful of passes per month. Saturation in TIR bands during major effusive events means you lose quantitative temperature data exactly when you most need it.
Effusion rate estimates derived from radiative power carry substantial uncertainty, typically 20–50%, depending on the empirical coefficients applied and assumptions about lava surface emissivity and crust thickness. They are useful for order-of-magnitude tracking and trend detection, not for engineering calculations. For hazard assessment requiring precise flow volume or advance rate, ground-based or airborne survey remains necessary. Space data narrows the search area and provides context; it does not replace fieldwork at an active vent.
Typical figures
| Best SWIR spatial resolution (lava mapping) | 20 m (Sentinel-2 Band 11/12); 30 m (Landsat OLI) |
| Best TIR spatial resolution (crater lake, cooling flows) | 70 m (ECOSTRESS); 90 m (ASTER TIR) |
| Coarse-resolution thermal revisit (VIIRS) | Near-daily global; 375 m (I4 band), 750 m (M13 band) |
| Sentinel-2 SWIR revisit | 5 days at equator (both satellites combined) |
| ASTER tasking latency | 1–3 days from request to data delivery; 16-day free-repeat cycle |
| Spectral bands used | SWIR: 1.6 µm, 2.2 µm; MWIR: 3.7–4.1 µm; TIR: 8–12.5 µm |
| TIR saturation temperature (typical broad-band sensors) | ~340–370 K; SWIR channels remain unsaturated to ~800–1000 K |
| Minimum detectable thermal anomaly (TIR, clear sky) | ~0.3 K above background for large targets (>1 pixel); sub-pixel lava detectable via SWIR if fractional area >~0.001 of pixel |
| Archive depth | Landsat TIR from 1982; ASTER from 1999; VIIRS from 2012; Sentinel-2 from 2015 |
| Effusion rate retrieval uncertainty | Typically 20–50% using radiative power methods (empirical coefficients) |
Analytics Satellize can run
| Volcanic thermal anomaly alert | Contextual threshold detection on VIIRS I4/M13 radiative power (MIROVA-class algorithm); validated against background clutter statistics | Near-real-time alert (within 3–6 hours of overpass) with location, radiative power in MW, and confidence flag; delivered by API or email |
| Lava flow extent polygon | Binary classification of Sentinel-2 Band 12 (2.19 µm) anomaly above scene-specific threshold, morphological filtering to remove noise, vectorisation | GeoJSON or Shapefile of active flow perimeter per acquisition, with area in km² and date-stamped change layer |
| Effusion rate time series | Conversion of VIIRS volcanic radiative power to volumetric effusion rate using published empirical relationships (Wooster et al. method class) | CSV time series with daily or per-overpass estimates, uncertainty bounds, and flagged cloud-contaminated epochs |
| Crater lake surface temperature trend | Multi-sensor TIR retrieval (ASTER, ECOSTRESS, Landsat TIRS) with atmospheric correction; time-series anomaly detection against seasonal baseline | Monthly report with temperature trend chart, anomaly magnitude, and qualitative unrest classification (background / elevated / high) |
| Multi-decade thermal anomaly baseline | Landsat TIR and ASTER archive processing; cloud masking, emissivity correction, compositing to suppress noise | Annual thermal anomaly composites as GeoTIFF, covering available archive depth, with summary statistics per volcanic region of interest |
| Dome growth indicator | Change detection in Sentinel-2 SWIR and VNIR between cloud-free acquisitions; pixel-count growth rate as proxy for extrusion volume | Bi-weekly GIS layer update with dome outline, estimated area change, and data-gap log showing cloud-obscured periods |
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.