Open-cast highwall slope displacement and failure precursor detection
Persistent scatterer InSAR detects sub-centimetre creep on open-pit highwalls weeks before visible failure. Combining Sentinel-1 ascending and descending passes decomposes line-of-sight motion into true horizontal and vertical components, separating thermal expansion from progressive shear.
Sensors
- Sentinel-1 IW SLC (C-band, ESA): 5 m x 20 m ground resolution in interferometric wide-swath mode; 6-day revisit at mid-latitudes with both satellites. C-band (5.6 cm wavelength) is sensitive to millimetre-scale displacement but loses coherence on loose, freshly blasted spoil. Free and open archive from 2014.
- COSMO-SkyMed Second Generation (X-band, ASI): Spotlight mode delivers roughly 1 m resolution; X-band (3.1 cm wavelength) maintains coherence better than C-band on dry, consolidated rock faces. Revisit as short as 4 days with the two-satellite constellation. Commercial tasking required.
- TerraSAR-X / TanDEM-X (X-band, DLR): StripMap mode at 3 m resolution; High-Resolution SpotLight at approximately 1 m. 11-day repeat orbit; shorter effective revisit is achievable by combining ascending and descending passes. Particularly useful for generating high-quality reference DEMs via TanDEM-X bistatic acquisition.
- ICEYE X-band SAR constellation: Sub-metre Spotlight imaging; constellation of more than 30 satellites enables revisit intervals of hours to days on a tasked target. Useful for rapid response after an anomalous displacement event, though the short archive limits long time-series PS-InSAR.
Why millimetres matter on a highwall
Open-pit slope failures rarely arrive without warning. The geomechanical sequence is well established: shear stress builds along a discontinuity or weak zone, micro-slip begins, displacement rate accelerates, and collapse follows. The critical window is the acceleration phase, which can span days to months depending on rock mass quality and geometry. Conventional prism-based total stations and GPS monuments capture this well, but only at the points where instruments have been installed. A highwall several kilometres long and 300 m deep has far more potential failure surfaces than any practical ground-monitoring network can cover.
Persistent scatterer InSAR (PS-InSAR) changes the geometry of the problem. It identifies individual radar-bright pixels, typically rock outcrops, bench corners or installed corner reflectors, that remain coherent across dozens of SAR acquisitions. Each such scatterer carries a phase history that encodes cumulative displacement along the satellite's line of sight to sub-centimetre precision, typically 1 to 3 mm per measurement epoch under good conditions. A single Sentinel-1 time series over a large pit can yield thousands of scatterers, providing spatial coverage that no ground network can match.
Ascending plus descending: decomposing the motion vector
A single SAR geometry measures only the component of displacement projected onto the line of sight between satellite and ground. For a highwall moving predominantly horizontally into the pit, a purely ascending or descending stack will underestimate or misattribute the motion depending on slope aspect relative to the satellite heading. The standard remedy is to process ascending and descending stacks independently and then solve the two-equation system for vertical and east-west horizontal displacement at each scatterer. North-south sensitivity remains poor with polar-orbiting sensors, a genuine limitation that must be acknowledged in any slope-aspect analysis for walls oriented roughly north-south.
This decomposition has been applied and validated at Chuquicamata in Chile, one of the world's largest open-pit copper mines, and at several other large pits documented in peer-reviewed literature. Published results show that the method reliably separates seasonal thermal expansion, which produces reversible, temperature-correlated displacement cycles of a few millimetres, from progressive shear displacement, which accumulates monotonically and follows the inverse-velocity acceleration pattern described by Fukuzono and later extended by Voight. The distinction matters operationally: thermal cycling is noise; monotonic acceleration is a precursor signal.
Where the method struggles
Coherence is the binding constraint. Freshly blasted bench faces, wet or clay-rich spoil, and dense vegetation on rehabilitated slopes all scatter radar energy incoherently, producing no usable scatterers. C-band Sentinel-1 is more susceptible to coherence loss than X-band systems; in wet climates or on actively mined benches, X-band sensors such as COSMO-SkyMed or TerraSAR-X are often necessary to maintain adequate scatterer density.
Temporal resolution is a second limit. Sentinel-1's 6-day repeat is adequate for slow, months-long creep but may miss a rapid failure that accelerates over 24 to 48 hours. Commercial X-band constellations with sub-daily revisit capacity can fill this gap on a triggered basis, but they require pre-arranged tasking agreements and add cost. Processing latency also matters: PS-InSAR requires a time series, so the minimum stack for reliable velocity estimation is typically 20 or more acquisitions, meaning several months of data are needed before a reliable baseline velocity map exists for a new site. Corner reflectors installed on the pit wall can improve scatterer density and provide ground-truth displacement references, but installation on active benches carries its own operational constraints.
From displacement map to operational alert
The analytic workflow moves through several stages. First, a reference velocity map is established from the historical archive, typically 12 to 24 months of Sentinel-1 SLC data, to characterise background displacement rates across all sectors of the highwall. Sectors showing anomalous velocity relative to their neighbours, or relative to their own seasonal baseline, are flagged for closer attention. Second, ongoing acquisitions are processed within a few days of downlink and compared against the baseline. Third, scatterers in flagged zones are tracked using inverse-velocity analysis: if the reciprocal of displacement rate is declining linearly toward zero, the Fukuzono-Voight criterion suggests a finite time to failure can be estimated, with all the uncertainty that implies.
Outputs are typically delivered as georeferenced GIS layers showing displacement velocity and cumulative displacement per scatterer, overlaid on a high-resolution pit DEM. Alert thresholds are set in consultation with the mine's geotechnical team, who understand the site's specific geology and operational risk tolerance. The satellite analysis complements, rather than replaces, ground instrumentation; the two data streams should be reconciled regularly.
Practical considerations for procurement
A Sentinel-1-only programme is the lowest-cost entry point and is appropriate for pits in dry climates with consolidated rock faces and relatively slow anticipated displacement rates. The 6-day revisit and free archive make it practical for most geotechnical teams to establish a baseline within three to six months of programme initiation. For pits in humid climates, pits with active blasting on monitored benches, or situations where rapid acceleration is a credible scenario, supplementing with commercial X-band tasking is worth the additional expenditure.
Satellize runs PS-InSAR analytics on open constellations and can add commercial tasking on client licence, which is relevant for mines that need X-band coverage without establishing their own data agreements. The workflow for highwall monitoring shares infrastructure with the tailings dam deformation monitoring covered on a separate page in this library, so combined programmes can share baseline processing costs. Clients should budget for a geotechnical interpreter to contextualise displacement outputs; the satellite data identifies where and how fast, but the failure mechanism and risk consequence require on-site geological knowledge.
Typical figures
| Spatial resolution (Sentinel-1 IW) | 5 m range x 20 m azimuth (SLC); PS scatterer spacing depends on scene coherence, typically 10-50 m in practice |
| Spatial resolution (X-band commercial) | 1-3 m (Spotlight / StripMap); denser scatterer networks on consolidated rock |
| Displacement measurement precision | 1-3 mm per epoch under good coherence conditions; degrades on incoherent surfaces |
| Minimum detectable velocity | Approximately 5-10 mm/year for PS-InSAR over a 12-month stack; faster motions detectable sooner |
| Revisit interval | 6 days (Sentinel-1, two satellites); 4 days (COSMO-SkyMed SG); 11 days (TerraSAR-X); hours to 1 day (ICEYE tasked) |
| SAR frequency / wavelength | C-band 5.6 cm (Sentinel-1); X-band 3.1 cm (COSMO-SkyMed, TerraSAR-X, ICEYE) |
| Archive depth | Sentinel-1: from 2014 (Europe/global coverage varies); TerraSAR-X: from 2007; COSMO-SkyMed: from 2007 |
| Processing latency | 3-5 days after acquisition for routine update; near-real-time alert possible with pre-processed baseline |
| Geometric limitation | North-south displacement component poorly constrained with polar-orbiting sensors; ascending+descending resolves vertical and east-west only |
| Delivery formats | GeoTIFF displacement rasters, vector PS point layers (GeoPackage / Shapefile), CSV time series per scatterer, PDF technical report |
Analytics Satellize can run
| Baseline displacement velocity map | PS-InSAR time-series analysis (StaMPS, MintPy or equivalent open-source processor) on 12-24 month Sentinel-1 SLC stack | Georeferenced GIS layer of mean line-of-sight velocity per persistent scatterer, delivered as GeoPackage with attribute table |
| Ascending/descending motion decomposition | Two-geometry inversion to separate vertical and east-west horizontal displacement components, following published methods validated at large open pits | Paired raster layers (vertical velocity, horizontal velocity) at PS point locations, with uncertainty estimates |
| Thermal vs. progressive displacement separation | Regression of displacement time series against surface temperature (ERA5 or on-site records) to isolate reversible thermal component; residual treated as structural signal | Per-scatterer decomposition report flagging zones where residual (non-thermal) velocity exceeds defined threshold |
| Inverse-velocity acceleration monitoring | Fukuzono-Voight inverse-velocity criterion applied to flagged scatterer clusters; linear regression on 1/v to estimate time-to-failure window | Weekly alert bulletin for anomalous sectors, with inverse-velocity plots and confidence interval on projected failure window |
| Sector-level risk ranking | Spatial clustering of scatterers by velocity percentile and acceleration trend; ranked against pit DEM geometry and bench configuration | Quarterly risk-ranked highwall sector map for geotechnical review, in PDF and GIS format |
| Triggered X-band rapid response | Commercial COSMO-SkyMed or ICEYE tasking activated on alert threshold breach; differential interferogram generated within 24-48 hours of acquisition | Single-pair interferogram and displacement estimate for the flagged zone, delivered as GeoTIFF with interpretation note |
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.