Reservoir-induced seismicity and dam-body deformation monitoring
InSAR can detect millimetre-scale deformation of dam bodies and surrounding bedrock before that motion becomes a seismic or structural event. This page explains the physics, the sensors, and the honest limits of the method.
Sensors
- Sentinel-1 (C-band SAR, ESA): 5.6 cm wavelength, 5 x 20 m ground range resolution in Interferometric Wide swath mode, 6-day repeat at mid-latitudes with both satellites operating. Free and open archive from 2014. Sufficient for detecting deformation at the centimetre scale on coherent concrete faces; coherence degrades on vegetated earthfill embankments, particularly in humid climates.
- COSMO-SkyMed (X-band SAR, ASI): 3.1 cm wavelength, spotlight mode achieves approximately 1 m resolution, with revisit potentially as short as 1-4 days via the four-satellite constellation. X-band is more sensitive to small displacements than C-band but loses coherence faster over rough or vegetated surfaces. Well suited to concrete arch dam faces and rock abutments.
- TerraSAR-X / TanDEM-X (X-band SAR, DLR): 3.1 cm wavelength, Staring Spotlight mode delivers approximately 0.25 m azimuth resolution. Revisit of 11 days for a single satellite, reducible by orbit selection. Provides the highest geometric precision of commercially available SAR for point-target analysis on dam infrastructure. TanDEM-X bistatic pairs also support very high-resolution DEM generation for abutment slope mapping.
- ALOS-2 PALSAR-2 (L-band SAR, JAXA): 23.6 cm wavelength, 3-10 m resolution in stripmap modes, 14-day repeat. L-band penetrates vegetation canopy and maintains coherence on earthfill and soil-covered embankments where X- and C-band fail. Critical for monitoring reservoir rim slopes and the approach embankments of rockfill dams. The longer wavelength is less sensitive per fringe but more coherent over longer time intervals.
What a filling reservoir does to the ground beneath it
When a large reservoir fills, three distinct mechanical processes act on the underlying crust simultaneously. The water column adds a direct elastic load, flexing the crust downward by amounts that can reach tens of millimetres at the dam toe and taper over tens of kilometres. Pore pressure rises as water diffuses into permeable fault zones, reducing the effective normal stress on pre-existing faults. And the shear stress field around the reservoir changes shape as the load geometry evolves with water level.
Reservoir-induced seismicity (RIS) is the documented result. The Zipingpu reservoir in Sichuan, China, impounded in 2004, sits approximately 17 km from the Longmenshan fault system. Several published studies have argued, with varying degrees of certainty, that the stress perturbation from Zipingpu's roughly 300-metre water column advanced the timing of the 2008 Wenchuan Mw 7.9 earthquake, though the causal link remains debated in the literature. What is not debated is that the reservoir produced measurable surface deformation that InSAR could, in principle, have tracked continuously. The Vajont reservoir in northern Italy is the older and grimmer example: slope deformation on Monte Toc was observed before the 1963 landslide that overtopped the dam and killed approximately 2,000 people. Modern InSAR applied retrospectively to Vajont-analogue sites has shown that precursor motion on slopes of that geometry is detectable months in advance.
What a floating roof gives away: the physics of InSAR on dam structures
Synthetic aperture radar interferometry measures the phase difference between two SAR acquisitions of the same scene. One full phase cycle corresponds to a line-of-sight range change of half the radar wavelength: 2.8 cm for C-band Sentinel-1, 1.55 cm for X-band COSMO-SkyMed or TerraSAR-X. Sub-fringe precision, achieved through time-series methods such as Persistent Scatterer InSAR (PS-InSAR) or Small Baseline Subset (SBAS), can resolve displacements of 1-2 mm per year on stable, coherent reflectors.
Concrete dam faces are nearly ideal for PS-InSAR. The smooth, hard surface returns strong, phase-stable radar signals across years of acquisitions. Corner reflectors or radar-bright structural features on the dam crest and downstream face act as natural persistent scatterers. A well-processed X-band PS-InSAR time series over a concrete arch dam can resolve seasonal thermal expansion and contraction of the structure itself, which is a useful calibration signal, as well as any anomalous trend that departs from the expected thermal cycle. Earthfill and rockfill embankments are a different problem. Vegetation, loose surface material, and settlement-induced surface roughening all destroy coherence. L-band ALOS-2 is the practical choice for embankment monitoring; even then, the analyst should expect coherence loss during wet seasons and should validate against in-situ survey points.
Slope instability at the reservoir rim: the Vajont lesson from orbit
Dam failure by overtopping from a landslide into the reservoir is, statistically, a more common catastrophic failure mode than structural collapse of the dam body itself. The reservoir rim, often kilometres of steep valley wall, is rarely instrumented at the density that the dam structure itself receives. InSAR fills that gap economically.
Published studies using Sentinel-1 SBAS processing over reservoir rim slopes in the Alps, the Himalayas, and the Three Gorges reservoir zone have detected precursor creep at rates of a few millimetres per month, accelerating to centimetres per month in the weeks before visible slope movement. The detection threshold depends on coherence, which depends on vegetation and surface roughness. On bare rock or sparsely vegetated slopes, C-band can resolve 3-5 mm of cumulative displacement between acquisitions 12 days apart. On forested slopes, L-band with a longer baseline interval is required, and the effective detection threshold rises to roughly 10-20 mm. Neither sensor sees through dense cloud, a real operational constraint in monsoon-affected or maritime mountain environments. Ascending and descending orbit geometries together help decompose line-of-sight displacement into horizontal and vertical components, which matters for distinguishing slope-parallel creep from vertical settlement.
Time-series design: what an operational monitoring programme actually needs
A credible InSAR monitoring programme for a large dam requires several design decisions that are not always obvious to infrastructure owners. First, the archive baseline matters. Processing begins with the pre-impoundment or early-impoundment SAR archive to establish the natural deformation state of the site before the reservoir load was applied. Sentinel-1's archive from 2014 onward covers most dams commissioned or significantly modified since then. For older dams, ERS-1/2 and Envisat data from the 1990s and 2000s extend the record, though with lower resolution and longer revisit.
Second, the choice between PS-InSAR and SBAS is not a matter of preference but of surface type. PS-InSAR suits the concrete dam face and rock abutments. SBAS, which averages over distributed scatterers, suits the embankment shoulders and rim slopes. A complete monitoring product runs both chains and merges the outputs into a single deformation map referenced to a stable benchmark outside the deformation zone. Third, the deformation signal must be separated from atmospheric delay artefacts, which can mimic real ground motion at the 5-20 mm level. This requires either a high-quality numerical weather model correction or the use of a network of GPS reference stations near the site. Without atmospheric correction, short-period signals over steep topography are unreliable.
Satellize's processing pipeline for InSAR time-series products applies SBAS and PS methods on Sentinel-1 open data, with optional commercial tasking on COSMO-SkyMed or ALOS-2 for sites where coherence or resolution demands it.
Where the method reaches its limits
InSAR is not a replacement for in-situ instrumentation. A concrete arch dam under active monitoring will carry extensometers, pendulums, piezometers, and seismometers that respond faster and more precisely than any satellite revisit cycle permits. The satellite contribution is complementary: it covers the full spatial extent of the deformation field, including areas that are not instrumented, and it provides an independent check on point sensors that may drift or fail.
The 6-day Sentinel-1 revisit is adequate for slow creep monitoring but will miss a rapid slope failure that develops over hours. Commercial constellations with 1-4 day revisit reduce but do not eliminate this gap. Rapid drawdown events, which can destabilise reservoir rim slopes by reducing the buttressing water pressure faster than pore pressure equalises, can trigger failure on timescales shorter than any current SAR revisit. The honest position is that InSAR provides early warning for processes developing over weeks to months, not for events that accelerate suddenly from apparent stability. Integrating InSAR trends with water-level telemetry and seismic network data is the architecture that makes early warning credible.
Typical figures
| Spatial resolution (Sentinel-1 IW mode) | 5 x 20 m (range x azimuth); multi-looked to ~14 m for interferometry |
| Spatial resolution (COSMO-SkyMed Spotlight) | ~1 m; TerraSAR-X Staring Spotlight ~0.25 m azimuth |
| Revisit interval | 6 days (Sentinel-1 two-satellite); 1-4 days (COSMO-SkyMed); 11 days (TerraSAR-X); 14 days (ALOS-2) |
| Minimum detectable displacement (PS-InSAR, annual rate) | 1-2 mm/year on coherent concrete or rock; 5-10 mm/year on SBAS over vegetated earthfill |
| Radar frequency / wavelength | C-band 5.6 cm (Sentinel-1); X-band 3.1 cm (COSMO-SkyMed, TerraSAR-X); L-band 23.6 cm (ALOS-2) |
| Atmospheric correction accuracy | Residual error 5-20 mm without correction; reducible to 2-5 mm with ERA5 or GACOS tropospheric models |
| Archive depth | Sentinel-1 from April 2014; ERS/Envisat from 1992 for legacy analysis |
| Cloud penetration | SAR is all-weather; optical sensors (for visual change detection) are blocked by cloud cover |
| Coherence constraint | Temporal decorrelation limits usable baselines to ~30 days (X-band, vegetated); months to years on rock or concrete |
| Delivery formats | GeoTIFF displacement maps, time-series CSV per scatterer, KMZ for GIS overlay, PDF monitoring report |
Analytics Satellize can run
| Dam-face deformation time series | PS-InSAR on X-band or C-band SAR stack; persistent scatterer identification on downstream face and crest | Monthly GeoTIFF and per-point CSV showing line-of-sight displacement history; anomaly flag when trend departs from seasonal thermal baseline |
| Reservoir rim slope velocity map | SBAS InSAR on L-band (ALOS-2) or C-band (Sentinel-1) depending on vegetation cover; ascending and descending orbit fusion for 2D decomposition | Quarterly velocity raster (mm/year) covering full reservoir perimeter; ranked list of slope sectors by displacement rate |
| Pore-pressure diffusion deformation model | Comparison of observed InSAR surface deformation with elastic loading and pore-pressure diffusion forward models; residual field isolates non-elastic or fault-related motion | Interpreted deformation anomaly report identifying zones inconsistent with elastic loading alone; input layer for seismic hazard review |
| Abutment rock-mass displacement monitoring | TerraSAR-X or COSMO-SkyMed PS-InSAR on rock abutments; sub-millimetre sensitivity on stable corner-reflector targets if installed | Alert feed triggered when cumulative abutment displacement exceeds defined threshold; GIS polygon showing affected zone |
| Pre-impoundment baseline deformation survey | SBAS time series over 12-24 months of pre-fill SAR archive to characterise natural slope creep and seasonal signals before reservoir load is applied | Baseline report establishing site-specific natural variability; used as reference for all subsequent operational monitoring |
| Rapid drawdown slope-stability change detection | Differential InSAR pairs bracketing drawdown events; comparison of slope displacement rate before, during, and after water-level change | Event-triggered change detection report within 48 hours of SAR acquisition; displacement magnitude map for affected rim sectors |
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.