Differential settlement and tilt monitoring of individual buildings

Persistent Scatterer InSAR resolves line-of-sight displacement on individual buildings to sub-millimetre precision, detecting differential settlement weeks or months before visible structural damage appears. This page explains the method, its real limits, and what published studies in Mexico City, Shanghai and Rotterdam have demonstrated.

Sensors

TerraSAR-X / TanDEM-X: X-band (9.65 GHz), spotlight mode delivers 1 m ground resolution; 11-day repeat; dense PS networks on individual buildings; phase precision supports sub-millimetre line-of-sight measurement on strong scatterers.
Sentinel-1 A/B: C-band (5.405 GHz), IW mode at 5 × 20 m resolution; 6-day repeat over Europe (12-day globally with one satellite); free archive from 2014; sufficient PS density in urban areas for building-level analysis at lower spatial resolution than X-band.
COSMO-SkyMed (first and second generation): X-band; spotlight modes to 1 m; Italian Space Agency constellation of four satellites allows sub-daily revisit on tasked targets; second generation adds improved radiometric stability useful for long time-series PS analysis.
Capella Space (X-band SAR): Commercial X-band constellation; spotlight imaging to approximately 0.5 m; on-demand tasking within hours; archive depth shorter than Sentinel-1 but useful for rapid response or high-resolution PS densification on specific structures.

What a persistent scatterer actually measures

Synthetic aperture radar measures the two-way travel time of microwave pulses to the ground. When a surface moves toward or away from the satellite along the line of sight, the phase of the returned signal shifts. One millimetre of displacement at X-band (wavelength roughly 3.1 cm) corresponds to a phase shift of about 23 degrees. That is detectable, but only if you can separate it from atmospheric delay, orbital error and thermal noise. Persistent Scatterer InSAR does this by identifying pixels that behave as stable point reflectors across dozens or hundreds of acquisitions: corners of buildings, window frames, metal fixtures, parapets. Their phase histories are coherent enough that the displacement signal can be isolated from everything else.

The result is a sparse network of measurement points, each tied to a physical feature on a structure. Typical PS densities in dense urban areas run from a few hundred to several thousand points per square kilometre with Sentinel-1, and higher with X-band systems. On a single large building, TerraSAR-X spotlight data can yield tens of individual PS points, enough to fit a plane to the building face and estimate both mean settlement rate and tilt gradient across the structure.

Rigid-body tilt versus differential deformation: why the distinction matters

A building settling uniformly into soft ground is a nuisance. A building whose east foundation sinks faster than its west foundation is a structural emergency. PS-InSAR can distinguish the two, but only with care. Rigid-body tilt produces a linear phase gradient across PS points on a single facade: all points move coherently, with displacement proportional to height above the tilting axis. Differential deformation, by contrast, produces residuals from that linear fit. Cracking, spandrel separation and localised foundation failure all leave non-linear signatures in the PS point cloud.

The distinction requires enough PS points on the building to fit and then test a rigid-body model. This is straightforward with X-band data on large reinforced-concrete or steel-frame structures. It becomes ambiguous on narrow terraced housing or low-rise masonry where only two or three PS points exist per facade. Analysts should state this limit explicitly rather than infer differential deformation from sparse data.

What published studies have actually shown

Mexico City sits on drained lake sediments that compress under load and groundwater extraction. Published work using Sentinel-1 and TerraSAR-X has documented building-scale settlement rates exceeding 30 cm per year in the most affected districts, with differential rates between adjacent structures reaching several centimetres per year. The PS approach there is complicated by the sheer magnitude of motion, which can exceed a fraction of the radar wavelength per acquisition interval and cause phase unwrapping errors. Researchers have addressed this by using shorter-wavelength X-band data and multi-temporal unwrapping algorithms.

In Shanghai, studies using COSMO-SkyMed and TerraSAR-X over the Pudong district have resolved differential settlement between adjacent high-rise towers founded on different pile configurations, with measured line-of-sight rates differing by 3 to 8 mm per year between neighbouring structures. Rotterdam has been studied with Sentinel-1 PS-InSAR to monitor buildings on timber pile foundations susceptible to groundwater-table fluctuation; published results identified structures with tilt rates detectable at the 1 to 2 mm per year level. These are not outliers. They represent what the method delivers when archive depth exceeds three years and PS density is adequate.

The limits you should know before commissioning an analysis

Line-of-sight geometry is the first constraint. Sentinel-1 ascending and descending passes together constrain vertical and east-west displacement reasonably well, but north-south motion remains poorly observed because the satellite flies near-polar orbits. A building tilting predominantly northward will be underestimated. Second, the minimum detectable displacement rate depends on archive length and atmospheric conditions. In humid tropical cities, atmospheric water vapour introduces path-delay noise that can mask signals below 3 to 5 mm per year even with long archives. Arid continental cities are more favourable.

Third, PS-InSAR is a relative measurement. Displacement is computed relative to a reference point assumed stable. If the entire neighbourhood is subsiding, you need an independent geodetic anchor, typically a continuous GNSS station or a bedrock outcrop, to establish absolute rates. Fourth, the method gives no information about the cause of settlement. Distinguishing groundwater extraction from construction loading from foundation deterioration requires integration with hydrogeological, structural and geotechnical data. The satellite tells you the building is moving. It does not tell you why.

Turning a displacement map into a structural risk signal

Raw PS velocity maps are not directly usable by structural engineers or insurers. The analytical step that matters is converting line-of-sight rates into physically interpretable quantities: vertical settlement rate, tilt angle, differential settlement between foundation points, and acceleration (change in rate over time). Acceleration is particularly significant. A building settling at a steady 2 mm per year for a decade is probably in equilibrium with its environment. The same building accelerating from 2 to 8 mm per year over eighteen months warrants inspection.

Satellize processes Sentinel-1 and commercial X-band archives using published PS-InSAR algorithms to produce per-building displacement time-series, tilt estimates and acceleration alerts, delivered as GIS layers and structured reports. The workflow is the same class of analysis applied to the Tonga crop-estimation programme: open-constellation data processed with documented methods, results expressed in terms a non-specialist buyer can act on. Structural engineers then decide whether field inspection or geotechnical investigation is warranted. The satellite data narrows the list of structures that need attention; it does not replace the engineer.

Archive depth and sensor choice for different project types

For a city-wide risk screen, Sentinel-1 is the practical starting point. The archive runs from April 2014 in Europe and is freely accessible. A ten-year time-series gives enough epochs to detect slow chronic settlement and to identify structures that have changed behaviour. The 5 × 20 m resolution means individual PS points cannot always be assigned to a specific building in dense urban fabric, but at the city block level the method is well validated.

For a specific structure under legal dispute, insurance claim or pre-acquisition due diligence, X-band tasking with TerraSAR-X or COSMO-SkyMed is appropriate. The higher spatial resolution, shorter wavelength and tighter phase noise allow per-floor displacement estimation on tall buildings and rigorous tilt-gradient fitting. Expect to pay for tasking and allow six to twelve months of new acquisitions if no archive exists over the target, or commission a historical archive analysis if the satellite has overflown the site on routine passes. Capella Space offers rapid tasking for urgent cases where the archive gap cannot be tolerated.

Typical figures

Spatial resolution (Sentinel-1 IW)	5 × 20 m ground range; PS points tied to sub-pixel scatterers
Spatial resolution (TerraSAR-X spotlight)	~1 m; enables per-floor PS attribution on large structures
Revisit interval	6 days (Sentinel-1, Europe, two satellites); 11 days (TerraSAR-X); sub-daily possible with COSMO-SkyMed tasking
Minimum detectable displacement rate	0.5 to 2 mm/year (X-band, dry climate, >3-year archive); 2 to 5 mm/year (C-band, humid urban, shorter archive)
Line-of-sight precision per epoch	1 to 3 mm on strong PS after atmospheric correction
Frequency / wavelength	C-band 5.405 GHz / 5.6 cm (Sentinel-1); X-band 9.65 GHz / 3.1 cm (TerraSAR-X, COSMO-SkyMed, Capella)
Archive depth	Sentinel-1: from April 2014 (Europe); TerraSAR-X: from 2007; COSMO-SkyMed: from 2007
Coverage	Global (Sentinel-1 with gaps); targeted tasking worldwide (X-band commercial)
Delivery formats	GeoTIFF velocity maps, GeoPackage PS point clouds, CSV time-series, PDF structural risk reports
North-south sensitivity	Poor on all near-polar SAR systems; ascending plus descending combination required for 2D decomposition

Analytics Satellize can run

Per-building PS velocity map	Persistent Scatterer InSAR (StaMPS or similar published algorithm) applied to Sentinel-1 or X-band archive	GeoPackage point layer with mean line-of-sight velocity and uncertainty per PS point, attributed to building footprint polygons
Tilt angle and tilt gradient estimate	Rigid-body plane fit to PS points on individual building facades using published least-squares methods	Tabular report per structure: tilt magnitude (mm/m), azimuth of maximum tilt, 95% confidence interval
Differential settlement between foundation points	Vertical decomposition from ascending and descending PS time-series; differencing between PS clusters at opposite ends of building footprint	Time-series plot and summary statistic (mm differential, rate, acceleration) per structure
Displacement acceleration alert	Piecewise linear or polynomial fit to PS time-series; change-point detection on rate	Automated alert (email or API) when rate increase exceeds user-defined threshold, with supporting time-series chart
City-wide settlement risk ranking	Sentinel-1 PS-InSAR over full urban extent; spatial join to building cadastre; ranking by velocity, acceleration and tilt	GIS layer with risk tier (high / medium / low) per building polygon, suitable for planning or insurance portfolio screening
Pre- and post-construction baseline comparison	Historical archive PS analysis before construction start; ongoing monitoring during and after; statistical comparison of rates	Structured PDF report with before/during/after velocity panels and differential settlement tables, suitable for legal or regulatory submission

Sources

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.