Surface heave above deep basement excavation from InSAR
Deep basement excavations relieve overburden stress and disturb groundwater, producing upward ground movement that can crack adjacent structures. Short-interval X-band InSAR time-series resolves this heave at millimetre scale across dense city blocks.
Sensors
- TerraSAR-X / TanDEM-X: X-band (9.65 GHz), stripmap mode at 3 m resolution, spotlight at 1 m. Repeat cycle 11 days, but programmable for 2–3 day revisit via orbit agility. Preferred for isolating individual building footprints in dense urban blocks.
- COSMO-SkyMed Second Generation: X-band, spotlight mode to 1 m resolution, 16-day nominal repeat but constellation of four satellites allows sub-weekly tasking. High coherence on urban hard targets over short intervals.
- ICEYE: X-band SAR constellation of 30+ satellites. Stripmap at roughly 3 m, spot at sub-1 m. Can achieve same-day or next-day revisit on a given city, which is operationally useful during active excavation phases.
- Capella Space: X-band, spotlight mode at 0.5 m resolution. Small constellation with on-demand tasking. Useful for detecting displacement on narrow structures such as party walls and terrace frontages adjacent to a pit.
- Sentinel-1 (ESA): C-band (5.4 GHz), 5 × 20 m IW mode, 6-day repeat over Europe. Free archive back to 2014. Suitable for regional context and for monitoring slower, broader heave bowls, but spatial resolution limits its use on individual buildings within 20–30 m of the pit edge.
What heave actually is, and why it matters more than it looks
When a deep excavation removes tens of thousands of tonnes of soil, the ground around and beneath the pit rebounds elastically. Stress that the overburden was exerting on the surrounding soil mass is suddenly absent. The result is upward displacement, sometimes several millimetres, occasionally more than a centimetre, in the ground adjacent to the excavation wall. This is heave in its elastic form, and it is largely recoverable once the basement structure is built and backfilled.
The dangerous version is different. If groundwater drawdown accompanies dewatering of the pit, the effective stress in adjacent clay or silt layers changes permanently. Consolidation settlement can follow in one direction while the elastic heave signal is still present in another. A monitoring programme that detects only vertical movement without distinguishing these two mechanisms can give a reassuring net-zero reading while both processes are simultaneously damaging adjacent foundations. InSAR time-series, interpreted alongside piezometer records and geotechnical prism data, is one of the few tools that can map the spatial extent of both signals at once.
Why X-band and not C-band for this problem
The choice of radar frequency is not academic. C-band Sentinel-1, at 5 × 20 m ground resolution in its standard interferometric wide-swath mode, cannot reliably isolate a single Victorian terrace house from its neighbour when both sit within 15 m of a retaining wall. X-band sensors at 1–3 m resolution can. A spotlight-mode TerraSAR-X acquisition resolves individual building facades as persistent scatterers, allowing a displacement time-series to be attached to a specific structure rather than to a blurred neighbourhood average.
Wavelength also matters for sensitivity. X-band's 3.1 cm wavelength means a half-wavelength displacement of roughly 15 mm causes a full phase cycle. C-band at 5.6 cm doubles that threshold. For heave signals that often peak at 3–8 mm in the first weeks of excavation, X-band's finer phase gradient is measurable where C-band may be at or below its noise floor. The trade-off is that X-band loses coherence faster over vegetated or disturbed surfaces, which is why a mixed strategy, X-band for the immediate adjacency zone and Sentinel-1 for the wider regional context, is often the most defensible approach.
Reading the displacement signature: what the interferogram shows
A typical InSAR time-series over an active deep basement excavation produces a characteristic spatial pattern. Directly above the pit, coherence is lost entirely because the surface is being physically removed. In an annular zone around the pit, roughly one to two times the excavation depth in radius, the time-series shows upward line-of-sight displacement during the active dig phase. Beyond that zone, displacement tapers to noise level. The geometry of this bowl is diagnostic: a symmetric pattern suggests elastic rebound; an asymmetric one, with stronger heave on one side, often points to differential groundwater drawdown or to anisotropic soil conditions.
Persistent scatterer InSAR (PS-InSAR) and small-baseline subset (SBAS) methods both apply here, but they suit different parts of the problem. PS-InSAR extracts phase histories from stable point reflectors, typically building corners, window frames and lamp posts, and works well in the dense urban fabric surrounding a pit. SBAS uses distributed scatterers across short-baseline image pairs and performs better on open ground or construction fill where point reflectors are sparse. Running both and cross-validating against each other is standard practice in high-stakes monitoring.
Honest limits of the method
InSAR measures displacement in the radar line-of-sight direction, not true vertical. Converting to vertical requires knowing the incidence angle and assuming horizontal motion is negligible, an assumption that holds reasonably well for heave but breaks down if the retaining wall is also deflecting laterally. Reported vertical precision for X-band PS-InSAR in urban environments is typically 1–2 mm per epoch under good coherence conditions, but that figure degrades near construction scaffolding, crane movements and freshly placed concrete, all of which scatter radar energy unpredictably.
Revisit cadence is a genuine constraint. Even with ICEYE's constellation, a two-to-three day repeat is achievable but not guaranteed for every orbit geometry. During the most critical phase of excavation, when heave rates can change within 24 hours after a large soil lift, satellite InSAR alone is insufficient. It should be treated as a spatial mapping layer that contextualises the point data from optical levelling targets and in-ground inclinometers, not as a replacement for them. Cloud cover does not affect SAR, which is one genuine advantage over optical methods, but atmospheric water vapour introduces phase delays that can mimic millimetre-scale displacement and must be corrected using ERA5 reanalysis or GACOS tropospheric models.
Putting the data to work on a construction project
The practical workflow starts before the first soil is moved. A baseline stack of X-band acquisitions over the pre-excavation site, ideally six to twelve months of archive imagery, establishes which adjacent structures are already moving and at what rate. This pre-existing motion baseline is essential for any legal or insurance dispute that arises later. Once excavation begins, new acquisitions are processed against the baseline and displacement maps are issued at each construction stage gate.
Threshold alerting is the most operationally useful output. A geotechnical engineer sets a heave trigger level, say 5 mm cumulative upward movement at a specific building facade, and the analytics pipeline flags any persistent scatterer that crosses that threshold between epochs. The alert includes the displacement time-series, the spatial map of the affected zone and a comparison against the predicted heave envelope from the original geotechnical design. Satellize structures this kind of monitoring layer as a GIS-deliverable feed that can sit alongside a project's existing BIM or geotechnical dashboard. The Tonga crop-estimation programme uses a similar epoch-triggered alert architecture, applied to a very different physical signal.
For project finance and insurance underwriting, the value is in the independent audit trail. A lender or insurer who needs to confirm that a contractor's claim of 'no measurable ground movement' is accurate can commission a retrospective InSAR analysis of archived X-band imagery. The archive for COSMO-SkyMed and TerraSAR-X extends back to 2007 and 2008 respectively, long enough to cover most contemporary disputes.
Typical figures
| Spatial resolution (X-band spotlight) | 0.5–3 m (Capella, ICEYE, TerraSAR-X, COSMO-SkyMed Second Generation) |
| Spatial resolution (C-band reference) | 5 × 20 m (Sentinel-1 IW mode); useful for regional context only |
| Revisit cadence | 2–3 days (ICEYE or COSMO-SkyMed constellation tasking); 11 days nominal for TerraSAR-X; 6 days for Sentinel-1 over Europe |
| Displacement precision (X-band PS-InSAR, urban) | 1–2 mm per epoch under good coherence; degrades near active construction plant |
| Minimum detectable heave signal | Approximately 2–3 mm cumulative for X-band; ~5 mm for C-band in typical urban coherence conditions |
| Radar frequency | X-band 9.65 GHz (3.1 cm wavelength) for primary monitoring; C-band 5.4 GHz (5.6 cm) for context |
| Archive depth | TerraSAR-X from 2008; COSMO-SkyMed from 2007; Sentinel-1 from 2014 (free) |
| Atmospheric correction | ERA5 reanalysis or GACOS tropospheric model required; uncorrected phase error can reach 3–5 mm equivalent displacement |
| Delivery formats | GeoTIFF displacement maps, GeoJSON persistent scatterer point clouds, CSV time-series per scatterer, threshold-alert feed |
| Processing methods | PS-InSAR (point scatterers on buildings), SBAS (distributed scatterers on open ground), combined hybrid stacks |
Analytics Satellize can run
| Pre-excavation baseline displacement map | PS-InSAR time-series on archived X-band imagery (6–12 months pre-construction) | GeoTIFF and GeoJSON scatterer layer showing pre-existing motion rates on all structures within 100 m of the planned pit boundary |
| Epoch-by-epoch heave progression map | Short-baseline interferogram stack processed with SBAS and PS-InSAR; line-of-sight to vertical conversion at sensor incidence angle | Displacement map issued at each construction stage gate, overlaid on cadastral building footprints |
| Threshold-breach alert | Automated comparison of per-scatterer cumulative displacement against engineer-defined trigger levels | Email or API alert with scatterer ID, coordinates, displacement time-series chart and spatial context map |
| Heave vs. settlement discrimination layer | Sign-separated time-series analysis isolating upward and downward displacement vectors; cross-referenced against piezometer or dewatering schedule data supplied by client | Annotated GIS layer distinguishing elastic heave zones from consolidation settlement zones around the pit perimeter |
| Retrospective archive audit | PS-InSAR reprocessing of historical X-band archive covering the construction period | PDF technical report with displacement chronology, suitable for insurance claims or legal proceedings |
| Atmospheric-corrected displacement time-series | GACOS or ERA5 tropospheric phase screen subtraction applied to each interferogram before stacking | Corrected CSV time-series per persistent scatterer, with before/after correction comparison plots |
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.