Volcanic lahar flow path and deposit mapping from SAR and optical
Lahars can bury roads, bridges and settlements within hours of an eruption, often under thick volcanic cloud. Sentinel-1 SAR amplitude change detection maps deposit extent regardless of weather; Sentinel-2 optical composites refine boundaries once skies clear.
Sensors
- Sentinel-1 SAR (C-band, ESA): 10 m ground range resolution in Interferometric Wide Swath mode, 6-day repeat at the equator (12-day per orbit direction). C-band backscatter amplitude drops sharply over fine-grained lahar deposits compared to vegetated or rocky surfaces, making fresh deposits detectable within one repeat cycle regardless of cloud cover or time of day.
- Sentinel-2 MSI (ESA): 10 m visible and near-infrared bands, 20 m shortwave infrared. False-colour composites using SWIR-NIR-Red distinguish wet lahar mud, dried deposit, and surviving vegetation with high contrast. 5-day revisit at mid-latitudes, though cloud cover over active volcanoes routinely delays usable acquisitions by days to weeks.
- TanDEM-X (DLR): Global DEM at 12 m posting, absolute vertical accuracy approximately 10 m (90th percentile), better over flat terrain. Used as the pre-event terrain surface for flow-path modelling. Publicly available at 90 m; full 12 m product requires a DLR science proposal or commercial licence.
- SRTM (NASA/NGA): 30 m global DEM, acquired in February 2000. Free and immediately accessible via USGS EarthExplorer. Vertical accuracy is roughly 9 m RMSE globally but degrades on steep volcanic flanks. Older than TanDEM-X and pre-dates many eruptions that have reshaped summit topography.
- Planet SuperDove: 3 m resolution, up to daily revisit on tasked areas. Eight spectral bands including red-edge. Useful for mapping deposit margins and infrastructure burial at sub-parcel scale once cloud breaks, though archive depth and access require a commercial licence.
Why lahars outrun conventional field survey
A lahar is not simply a slow mudflow. On steep volcanic flanks, flows have been recorded exceeding 70 km/h in the proximal channel before decelerating as valleys widen. The 1985 Nevado del Ruiz disaster buried the town of Armero, roughly 50 km from the summit, within two hours of eruption. Field teams cannot safely enter active valleys during or immediately after an event, and helicopter access is typically blocked by the same ash cloud that generated the hazard.
Satellite observation fills the access gap. The critical window is the first 6 to 24 hours, when deposit extent determines which roads are passable, which communities are cut off, and where search-and-rescue assets should concentrate. SAR closes that window even when nothing else can.
What SAR amplitude change actually measures
Sentinel-1 C-band radar illuminates the surface and records the intensity of the return signal. Vegetation, rough rock and built structures return strong, textured signals. A fresh lahar deposit of fine-grained volcanic sediment and water presents a specularly smooth surface at C-band wavelengths (roughly 5.6 cm), scattering energy away from the sensor and producing a pronounced amplitude drop relative to the pre-event baseline. The change is detectable even when the deposit is only a few tens of centimetres thick, though the relationship between deposit thickness and amplitude change is not linear and varies with sediment grain size, water content and surface roughness.
The practical workflow pairs a pre-event Sentinel-1 acquisition (ideally from the same relative orbit to preserve geometry) with the first post-event pass. Pixel-wise amplitude ratios or log-difference images highlight changed areas. Thresholding is straightforward but requires calibration against known stable surfaces to avoid false detections from wind roughening of water bodies or agricultural change. At 10 m resolution, individual buried structures are detectable; deposit boundaries in braided channels are reliable to roughly 20 to 30 m.
Optical composites once the cloud breaks
Sentinel-2 SWIR bands (1610 nm and 2190 nm) are particularly useful for lahar mapping because wet volcanic sediment absorbs strongly in SWIR while retaining moderate reflectance in the visible. A false-colour composite assigning SWIR2 to red, NIR to green, and red to blue renders fresh lahar deposits in dark brown or purple tones, dried deposits in lighter grey-brown, and surviving vegetation in vivid green. The contrast is sufficient to digitise deposit boundaries manually or to automate classification with a simple spectral index.
The honest caveat: over active tropical volcanoes, Sentinel-2 cloud-free acquisitions during or immediately after an eruption are rare. The 5-day revisit is a best case. In practice, analysts should expect to wait 1 to 3 weeks for a usable optical scene, by which time the deposit surface has dried, secondary flows may have occurred, and some boundaries will have been disturbed by early recovery activity. SAR and optical data are complementary, not interchangeable.
Flow-path modelling and the terrain uncertainty problem
Pre-event DEMs from TanDEM-X or SRTM support flow-path modelling using tools such as LAHARZ (a USGS-published method) or simpler D8 flow-routing algorithms. The models ingest a DEM, a source volume estimate, and channel geometry to produce probabilistic inundation envelopes for given lahar volumes. These outputs are valuable for pre-event hazard zonation and for constraining post-event mapping when satellite coverage is delayed.
The fundamental problem is that the terrain a lahar actually flows over may differ substantially from any pre-event DEM. Thick tephra fall during the eruption can deposit metres of loose ash on valley floors and flanks within hours, raising channel beds and redirecting flow paths before the lahar initiates. The 2010 Merapi eruption in Java, for example, produced pyroclastic deposits that significantly altered valley geometries before subsequent rain-triggered lahars. A DEM acquired months or years before the eruption will not capture this. Analysts should treat model outputs as bounding estimates and update them as SAR-derived deposit maps become available. Where TanDEM-X 12 m data is accessible, it is preferable to SRTM for channel geometry, but neither is immune to the post-eruption terrain problem.
Integrating the data stream in an operational response
A practical response workflow has three phases. In the first 0 to 12 hours, SAR amplitude change detection from the next available Sentinel-1 pass provides a deposit extent map and identifies blocked transport corridors. This feeds directly into humanitarian logistics planning. In the following days, repeat SAR acquisitions track deposit growth if rain continues to mobilise material, and flow-path models are updated with observed extents. Once cloud breaks, Sentinel-2 or tasked Planet imagery refines deposit boundaries and enables damage assessment at individual structure level.
The International Charter on Space and Major Disasters can activate rapid-mapping support from multiple agencies within hours of a declared disaster, providing access to commercial SAR and optical tasking beyond the open constellations. Satellize's analytics pipeline runs on open Sentinel data as a baseline, with commercial tasking added on client licence when sub-weekly revisit or sub-5 m resolution is operationally required.
Honest limits of the method
SAR amplitude change detection misses thin deposits (under roughly 10 to 20 cm) that do not substantially alter surface roughness. It also struggles to distinguish lahar deposits from other smooth surfaces that appeared between acquisitions, including standing floodwater or freshly ploughed agricultural fields. Urban areas present a persistent problem: built structures produce strong double-bounce returns that can mask deposit signatures in partially buried settlements.
Flow-path models carry volume uncertainties of at least one order of magnitude when source conditions are poorly constrained, which is almost always the case in the first hours after an eruption. DEM vertical errors of 9 to 10 m translate to significant lateral uncertainty in low-gradient distal channels. No satellite method currently provides reliable deposit thickness estimates from a single pass; thickness inference requires either repeat-pass DEM differencing (which demands a post-event DEM of comparable quality to the pre-event one) or field validation. These are not reasons to avoid satellite mapping; they are reasons to communicate uncertainty explicitly in every product delivered to a response team.
Typical figures
| SAR spatial resolution (Sentinel-1 IW mode) | 10 m range × 10 m azimuth (detected ground range) |
| SAR revisit (Sentinel-1, equatorial) | 6 days (both orbits combined); 12 days per orbit direction |
| Optical resolution (Sentinel-2 SWIR bands) | 20 m; resampled to 10 m for composite products |
| Optical revisit (Sentinel-2) | 5 days at equator under cloud-free conditions; operationally 1–3 weeks over active tropical volcanoes |
| DEM vertical accuracy (TanDEM-X 12 m) | ~10 m absolute (90th percentile); better over low-relief terrain |
| DEM vertical accuracy (SRTM 30 m) | ~9 m RMSE globally; degrades on steep volcanic flanks |
| Minimum detectable deposit thickness (SAR amplitude) | Approximately 10–20 cm for fine-grained sediment; varies with grain size and water content |
| SAR archive depth (Sentinel-1) | From April 2014 (Sentinel-1A launch); continuous |
| Typical latency from SAR acquisition to deposit map | 2–6 hours with automated processing pipeline |
| Delivery formats | GeoTIFF, GeoPackage (GPKG), KMZ, PDF report; WMS/WFS on request |
Analytics Satellize can run
| SAR amplitude change map (deposit extent) | Log-ratio amplitude change detection, Sentinel-1 IW SLC or GRD pairs, thresholded against stable-surface baseline | GeoTIFF and GeoPackage polygon layer of deposit extent, delivered within 6 hours of SAR acquisition |
| Optical deposit boundary delineation | Sentinel-2 SWIR false-colour composite with spectral index classification (SWIR2/NIR ratio); manual QA on deposit margins | Classified GeoTIFF and vector boundary file with deposit age attribution (fresh vs. dried) |
| Flow-path inundation envelope | D8 flow routing or LAHARZ-class volume-based modelling on TanDEM-X or SRTM DEM with stated volume uncertainty range | Probabilistic inundation polygon for low, median and high volume scenarios; PDF technical note stating DEM vintage and terrain uncertainty |
| Infrastructure burial assessment | Overlay of deposit extent polygon against OpenStreetMap or client-supplied road and building layer; intersection analysis | CSV and map report listing road segments and structure counts within deposit boundary, with confidence tier per feature |
| Repeat-pass deposit evolution time series | Multi-date SAR amplitude stack tracking deposit growth or remobilisation across successive Sentinel-1 passes | Animated GIF and GeoTIFF stack with per-date extent statistics; updated on each new acquisition |
| Post-event DEM differencing for deposit thickness estimation | Differencing of pre-event TanDEM-X or SRTM DEM against post-event DEM (where available from commercial tasking or Structure-from-Motion); uncertainty propagated from source DEM errors | Thickness raster with stated vertical uncertainty band; suitable for volume estimation with explicit confidence interval |
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.