Cofferdam dewatering and foundation pit settlement from InSAR
Cofferdam dewatering consolidates surrounding soils, settling adjacent structures by millimetres that matter. Sentinel-1 and COSMO-SkyMed interferometric stacks resolve that motion in space and time, before cracks appear.
Sensors
- Sentinel-1 (ESA, C-band, 5.6 cm wavelength): Interferometric Wide Swath mode delivers 5 × 20 m ground resolution across a 250 km swath, with 6-day repeat at mid-latitudes when both satellites are active. Free archive from 2014 enables long baseline stacks for PS-InSAR and SBAS processing. C-band coherence degrades over freshly disturbed fill and standing water inside the pit.
- COSMO-SkyMed (ASI, X-band, 3.1 cm wavelength): Stripmap mode reaches 3 m resolution; Spotlight reaches approximately 1 m. The four-satellite constellation can revisit a site every 1 to 4 days under tasking. X-band's shorter wavelength improves sensitivity to small displacements but loses coherence faster over vegetation and loose spoil.
- TerraSAR-X / TanDEM-X (Airbus, X-band): High-Resolution Spotlight delivers approximately 1 m resolution. Repeat pass is 11 days for a single satellite, reducible with staggered acquisitions. Well-suited to dense urban persistent scatterer networks where buildings provide stable reflectors adjacent to the excavation.
- ALOS-2 PALSAR-2 (JAXA, L-band, 23.6 cm wavelength): L-band penetrates light vegetation and maintains coherence over softer ground covers that defeat X- and C-band stacks. Spatial resolution in Fine Beam mode is approximately 3 m. Revisit is 14 days. Useful where the settlement bowl extends into parkland or unpaved ground adjacent to the pit.
Why dewatering moves the ground outside the pit
A cofferdam keeps an excavation dry by lowering the piezometric head inside the sheet-pile or diaphragm-wall perimeter. That drawdown does not stop at the wall. Seepage gradients propagate outward, reducing pore-water pressure in the surrounding aquifer. Where soils are compressible, especially soft clays and silts, effective stress rises as pore pressure falls, and the skeleton consolidates. The result is a settlement bowl centred on the pit, typically widest at the surface and deepest closest to the dewatering boundary.
The geometry of that bowl depends on soil stratigraphy, the depth and permeability of the pumped layer, the efficiency of the cut-off wall, and how long pumping continues. In soft alluvial settings, drawdown of just a few metres can produce surface settlements of 10 to 50 mm over a radius several times the excavation width. Adjacent piled foundations may be partially protected, but shallow strip footings and buried utilities are not. Knowing the bowl's shape and how it evolves through the dewatering campaign is the core monitoring problem.
How an interferometric stack resolves millimetric motion
A single SAR interferogram measures the change in radar path length between two passes as a phase shift. One full cycle of phase corresponds to half the radar wavelength of displacement in the line-of-sight direction: roughly 1.4 mm per fringe for C-band Sentinel-1, and about 0.8 mm for X-band COSMO-SkyMed or TerraSAR-X. Single interferograms are noisy. Stacking tens or hundreds of image pairs using Persistent Scatterer (PS-InSAR) or Small Baseline Subset (SBAS) methods suppresses atmospheric delay and thermal noise, recovering displacement time-series with standard deviations typically in the 1 to 3 mm range for well-conditioned urban stacks.
Persistent scatterers in urban areas are plentiful: lamp posts, kerb edges, parapet walls, window sills. Each acts as a stable phase reference. The PS network around an active excavation can be dense enough to resolve the settlement bowl spatially with point spacing of 5 to 20 m, depending on building density and sensor resolution. SBAS methods work better where the PS density is lower, trading spatial resolution for coherence by averaging over distributed scatterers.
Displacement is measured in the radar line-of-sight, which for Sentinel-1 ascending geometry is roughly 34 to 38 degrees from vertical. Vertical settlement dominates the signal for consolidation, so the line-of-sight component captures most of it. Combining ascending and descending passes allows decomposition into vertical and east-west horizontal components, which matters for detecting lateral wall movement as well as settlement.
What coherence loss tells you, and where it misleads you
Coherence is the correlation of phase between two SAR acquisitions. It drops toward zero over surfaces that change between passes: water, freshly placed fill, vegetation, and the disturbed ground inside the pit itself. That is precisely the problem. The area of greatest geotechnical interest, the pit floor and the immediate surcharge zone, is often incoherent. InSAR cannot monitor settlement of the excavation base directly during active works.
The useful signal lives in the surrounding built fabric. Provided buildings within roughly one to three excavation widths remain standing and undamaged, their PS points give a reliable settlement time-series. If a structure is demolished or scaffolded with metal sheeting, its scatterers change character and must be masked. Seasonal vegetation on verges can suppress coherence in C-band stacks during summer months, creating gaps in the spatial coverage of the bowl's outer edge.
Atmospheric delay is the other persistent nuisance. A 1 km column of troposphere varying by 1 per cent in water vapour content introduces several millimetres of apparent path-length change. Stacking over many acquisitions averages this down, but a single interferogram spanning a thunderstorm event is essentially useless for settlement detection at the millimetre level. This is why time-series methods, not individual interferograms, are the operational standard.
Minimum detectable rates and alert thresholds
For a well-conditioned PS stack over a dense urban area, the minimum detectable settlement rate is generally cited in the literature as 1 to 2 mm per year for long-duration monitoring. Over a dewatering campaign lasting weeks to a few months, the relevant question is not annual rate but cumulative displacement per campaign phase. A 6-day Sentinel-1 revisit means roughly five to six acquisitions per month, enough to detect cumulative settlements approaching 5 mm within a single month if the signal is clean.
Commercial X-band sensors with 1 to 4 day revisit improve temporal resolution considerably, allowing detection of accelerating settlement events that might precede structural distress. Alert thresholds are set by the geotechnical engineer, not by the satellite; typical trigger levels for adjacent structures range from 10 mm total settlement to 1 in 500 differential settlement over a sensitive foundation. The satellite time-series feeds those triggers with spatial coverage that no inclinometer or settlement point array can match economically.
Integrating satellite data into a dewatering monitoring programme
InSAR does not replace ground instruments. Vibrating wire piezometers inside the pit give real-time pore-pressure data. Settlement pins and precise levelling surveys on adjacent structures provide ground truth. What InSAR adds is spatial completeness: hundreds of measurement points across a several-hundred-metre radius, updated every few days, without site access. The combination is more informative than either alone.
A practical workflow starts with a pre-construction baseline stack, ideally six to twelve months of archive Sentinel-1 data, to establish background motion rates and identify any pre-existing subsidence unrelated to the works. Once dewatering begins, the stack is extended in near-real time. Velocity maps and cumulative displacement rasters are updated after each new acquisition. Where the bowl's edge approaches a trigger threshold at a specific building, a targeted commercial tasking order on COSMO-SkyMed or TerraSAR-X can provide higher-resolution confirmation within days.
Satellize runs this type of interferometric stack workflow on open Sentinel-1 archives and adds commercial tasking where resolution or revisit demands it. The analytics output is a georeferenced displacement time-series per PS point, delivered as a GIS layer with associated velocity and acceleration attributes, ready for overlay against the geotechnical monitoring database.
Honest limits a project team should carry into procurement
InSAR cannot see through the pit itself. It cannot detect settlement beneath a newly poured base slab or under a temporary works platform. It is a surface observation technique, and its surface must remain coherent. A site surrounded by low-rise masonry buildings in a dry climate is an ideal target. A site bordered by glass curtain-wall towers, active scaffolding, and a canal is a difficult one.
Latency matters too. Open Sentinel-1 data is typically available within 1 to 3 days of acquisition via the Copernicus Data Space. Processing a new interferogram and updating the PS time-series adds further hours. For a rapidly deteriorating situation, this is not a real-time early-warning system. It is a high-frequency audit trail that catches trends before they become emergencies, provided the monitoring programme is set up before dewatering starts, not after the first crack appears.
Typical figures
| Spatial resolution (Sentinel-1 IW) | 5 × 20 m (ground range × azimuth); PS point spacing typically 5–20 m in dense urban areas |
| Spatial resolution (COSMO-SkyMed Spotlight / TerraSAR-X HS) | Approximately 1 m; PS point spacing can approach 2–5 m |
| Revisit interval | 6 days (Sentinel-1, dual satellite); 1–4 days (COSMO-SkyMed tasked); 11 days (TerraSAR-X single satellite) |
| Line-of-sight displacement precision (PS stack) | 1–3 mm standard deviation for well-conditioned urban stacks |
| Minimum detectable cumulative settlement | Approximately 5 mm over a one-month campaign window at 6-day revisit; lower with X-band and denser acquisitions |
| 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) |
| Archive depth (Sentinel-1) | From April 2014 (Sentinel-1A); enables pre-construction baseline of 6–12 months or longer |
| Data latency (Sentinel-1 open access) | 1–3 days post-acquisition via Copernicus Data Space |
| Coherence limitations | Loss over open water, freshly placed fill, active scaffolding, and dense summer vegetation; pit interior typically incoherent |
| Delivery formats | GeoTIFF displacement rasters, GeoPackage or Shapefile PS point layers, CSV time-series per point, PDF campaign reports |
Analytics Satellize can run
| Pre-construction baseline velocity map | PS-InSAR or SBAS stack over 6–12 months of archive Sentinel-1 acquisitions | GeoTIFF velocity raster and PS point GeoPackage showing background motion rates before dewatering begins |
| Campaign-phase cumulative displacement map | Incremental SBAS update after each new acquisition, referenced to pre-dewatering baseline | Updated GIS layer of cumulative line-of-sight displacement, refreshed every 6 days (Sentinel-1) or per tasked acquisition |
| Settlement bowl spatial extent and depth profile | Kriging or radial basis function interpolation of PS velocity points around the pit perimeter | Contour GeoTIFF of settlement bowl with radial cross-sections, updated monthly |
| Per-structure displacement time-series | PS point clustering by cadastral parcel or structure footprint; median displacement per structure per acquisition | CSV and PDF time-series report per monitored building, with velocity, acceleration and threshold status flags |
| Threshold breach alert | Automated comparison of cumulative displacement or rate against engineer-specified trigger levels | Email or API alert with map attachment identifying the breaching PS cluster and nearest structure |
| Vertical / horizontal displacement decomposition | Ascending and descending pass combination following published two-geometry decomposition methods | Separate vertical and east-west displacement GeoTIFF layers, supporting lateral wall movement assessment alongside settlement |
| Coherence change map for disturbed-ground extent | Mean coherence differencing between pre- and during-construction interferogram stacks | GeoTIFF coherence-loss layer indicating areas where ground has been disturbed and InSAR coverage is degraded |
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.