Syn-eruption tephra deposit thickness and extent mapping
Pre- and post-eruption DEM differencing and SAR amplitude change let analysts estimate tephra volume and thickness, while SWIR optical bands separate fresh ash from cloud and snow during the eruption window itself.
Sensors
- TanDEM-X: X-band bistatic SAR pair producing DEMs at 12 m posting (global) and 6 m on tasked acquisitions. DEM differencing against the pre-eruption WorldDEM baseline (absolute vertical accuracy ~1 m LE90 over flat terrain) is the primary method for thickness estimation. Repeat tasking is typically days to weeks after an event, not real-time.
- Sentinel-1 C-band SAR: 6-day repeat (12-day per orbit, combined A+B), 10 m ground range resolution in IW mode. Amplitude change between pre- and post-eruption acquisitions indicates deposit presence; coherence is typically lost over fresh tephra, making this a qualitative extent indicator rather than a thickness tool. Free and open archive from 2014.
- Sentinel-2 MSI: 10–20 m resolution, 5-day revisit at mid-latitudes (combined A+B). SWIR bands at 1610 nm (Band 11) and 2190 nm (Band 12) are the primary optical discriminators: fresh volcanic ash has a distinctive reflectance signature that separates it from meteorological cloud and snow when viewing geometry and solar angle are favourable. Cloud cover during eruptions is a real constraint.
- Landsat 9 OLI-2 / TIRS-2: 30 m multispectral, 100 m thermal, 16-day repeat. SWIR bands at 1610 nm and 2200 nm mirror Sentinel-2 capability. TIRS-2 adds thermal contrast: freshly emplaced ash on warm ground can show a distinct thermal signature in the first hours. Useful as a cross-check and for longer archive continuity back through Landsat 8 to 2013.
What a DEM difference actually measures, and what it does not
The principle is straightforward: subtract a pre-eruption elevation model from a post-eruption one, and the positive residual is, in theory, tephra thickness. TanDEM-X bistatic acquisitions are the best satellite source for this because both antennas fire simultaneously, eliminating temporal decorrelation and giving a clean surface model rather than a phase-unwrapping product. The global WorldDEM baseline, derived from the 2010–2015 TanDEM-X science phase, provides the pre-eruption reference for most volcanoes.
The honest caveat is that X-band radar does not penetrate loose ash the way it penetrates dry snow. The surface return comes from the top of the deposit, so the elevation difference gives the geometric thickness of the unconsolidated pile at the moment of acquisition. That is not the same as the mass deposited. Bulk density of fresh tephra ranges from roughly 500 to 1,000 kg/m³ depending on grain size and vesicularity; converting thickness to mass requires a density assumption that introduces a factor-of-two uncertainty unless field samples are collected. The DEM approach is best treated as a volume estimate with explicit error bounds, not a precise mass budget.
Compaction, rain, and the shrinking deposit
Tephra does not stay where it falls. Within hours of emplacement, rainfall begins remobilising fine ash into lahars and rilling the deposit surface. Compaction under its own weight reduces thickness by 20–40% in the first weeks for coarse lapilli layers, and more for fine ash. This means that the timing of the post-eruption DEM acquisition matters enormously. A TanDEM-X pass acquired two weeks after the eruption will underestimate the original deposit thickness relative to one acquired within 48 hours.
For operational hazard response, the practical goal is usually to acquire a post-eruption SAR DEM as quickly as possible, ideally within the first 72 hours. After that, each additional day of rain, wind, and compaction degrades the thickness signal. Where repeat TanDEM-X tasking is not available at that cadence, Sentinel-1 amplitude change can at least delineate the deposit extent, giving emergency managers a spatial footprint even when thickness retrieval is not possible.
Optical SWIR bands during the eruption window when InSAR fails
InSAR coherence collapses the moment a fresh tephra layer lands. The deposit surface scatters radar phase randomly, so the interferogram is noise. During the active eruption and immediately after, optical imagery becomes the primary extent-mapping tool, with SWIR bands doing the discrimination work that visible bands cannot.
Fresh volcanic ash has a relatively low reflectance in the 1600–2200 nm range compared with meteorological ice cloud, which is bright in those bands due to ice crystal scattering. Snow also appears bright in SWIR. Ash, being a silicate glass and lithic mixture with little free water, absorbs more in SWIR and plots in a distinct region of a Band 11 versus Band 12 scatter plot. The separation is real but not perfect: thick ash clouds can be optically opaque at all wavelengths, and thin ash veneers over bright surfaces are easily missed at 20–30 m resolution. Published work on the 2010 Eyjafjallajökull and 2011 Puyehue-Cordón Caulle eruptions used this approach to map deposit boundaries within days of emplacement.
Landsat 9 TIRS-2 adds a useful dimension in the first hours. Warm ground beneath a thin ash veneer can still radiate through a deposit of a few centimetres, creating a detectable thermal contrast. Once the deposit thickens beyond roughly 10–15 cm, the thermal signal from the substrate is suppressed and TIRS-2 loses discriminating power.
Turning extent and thickness into a volume estimate
Volume estimation from satellite data follows two parallel paths. The first is DEM differencing: integrate the thickness field over the mapped area, propagating the vertical accuracy of the DEM (typically ±1–2 m for TanDEM-X over rough volcanic terrain) into a volume uncertainty. For a deposit covering 100 km² at a mean thickness of 0.5 m, the satellite-derived volume is 0.05 km³, but the uncertainty on that figure could easily be ±30% before field validation.
The second path is isopach modelling. If deposit extent can be mapped optically at several concentric distance intervals from the vent, classical Pyle-method isopach fitting (exponential thinning with distance) allows extrapolation of the volume including the distal tail that may be below DEM detection limits. This method requires at least three or four mappable isopach contours and works best when the deposit is roughly elliptical, which is the case for most Plinian and sub-Plinian fall deposits. Combining both approaches, where data allow, gives a more defensible volume estimate than either alone.
Satellize runs both workflows on open-constellation data and can add commercial TanDEM-X tasking under client licence. The Tonga crop-estimation programme showed the team's ability to integrate multi-sensor optical and radar products in island-arc environments where cloud cover and rapid landscape change are the norm, conditions that also characterise post-eruption monitoring in the Pacific.
Resolution floors and what satellite data genuinely cannot tell you
A 12 m TanDEM-X DEM cannot detect tephra thinner than roughly 20–30 cm in practice, because the noise floor of the DEM difference (driven by phase noise, baseline uncertainty, and surface roughness) is of that order over rough volcanic terrain. Distal ash deposits, which may be agronomically and infrastructurally important at thicknesses of 1–5 cm, are invisible to this method. Ground-based isopach surveys remain essential for the thin distal margin.
Optical mapping at 10–20 m resolution cannot distinguish a 2 cm ash veneer from a 20 cm deposit by reflectance alone; it maps presence or absence, not thickness. The SWIR approach also fails completely under persistent volcanic cloud, which is common over active vents. For eruptions in persistently cloudy regions (much of the tropical Pacific, Indonesia, Central America), usable optical windows may occur only once every several days.
Neither satellite method resolves grain-size variation across the deposit, which controls both the density assumption in mass calculations and the agricultural and infrastructure impact. Satellite data frames the problem and constrains the volume; field sampling closes it.
Typical figures
| DEM spatial resolution (TanDEM-X) | 12 m posting (global WorldDEM); 6 m on dedicated bistatic tasking |
| DEM vertical accuracy (flat terrain) | ~1 m LE90; degrades to 2–5 m over steep, rough volcanic terrain |
| Minimum detectable thickness (DEM differencing) | ~20–30 cm over rough surfaces; thinner deposits below noise floor |
| Sentinel-1 SAR revisit | 6 days (combined Sentinel-1A + 1B); 10 m ground range in IW mode |
| Sentinel-2 SWIR revisit | 5 days at mid-latitudes (combined A+B); 20 m resolution at 1610 nm and 2190 nm |
| Landsat 9 OLI-2 SWIR revisit | 16 days; 30 m resolution at 1610 nm and 2200 nm; TIRS-2 thermal at 100 m |
| Archive depth | Sentinel-1 from 2014; Sentinel-2 from 2015; Landsat back to 1972 (TM/ETM+/OLI); TanDEM-X baseline from 2010–2015 |
| Optical cloud constraint | SWIR mapping fails under thick volcanic or meteorological cloud; usable windows may be days apart in tropical regions |
| Delivery formats | GeoTIFF thickness grids, isopach shapefiles, volume estimate reports with uncertainty bounds |
Analytics Satellize can run
| Post-eruption tephra thickness grid | TanDEM-X bistatic DEM differencing against WorldDEM baseline; vertical uncertainty propagation | GeoTIFF raster of thickness in metres with per-pixel uncertainty layer |
| Deposit extent map from SWIR classification | Sentinel-2 and/or Landsat 9 Band 11/12 ratio thresholding and spectral unmixing to separate ash from cloud and snow | Polygon shapefile of mapped ash extent with confidence class attribute |
| Isopach contour set and volume estimate | Pyle exponential thinning fit to mapped thickness or extent contours; dual-method volume with stated uncertainty range | PDF report with isopach map, fitted thinning curve, and volume range in km³ |
| SAR amplitude change map for rapid extent delineation | Sentinel-1 pre/post amplitude ratio in dB; threshold-based deposit boundary extraction | GeoTIFF and shapefile delivered within 24 hours of post-eruption Sentinel-1 pass |
| Temporal deposit evolution series | Multi-date Sentinel-2 SWIR classification stack showing deposit shrinkage from remobilisation and compaction | Time-series GIF and tabulated area statistics at each acquisition date |
| Agricultural impact footprint | Overlay of thickness grid and deposit extent against land-cover classification; thickness thresholds for crop damage from published agronomic literature | GIS layer of affected agricultural area by thickness class, with area statistics by crop type |
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.