Construction and traffic vibration microdeformation monitoring at standing monuments
Persistent-scatterer InSAR detects sub-centimetre differential movement in historic masonry driven by nearby construction, tunnelling or heavy traffic, giving heritage managers quantified early warning before visible damage appears.
Sensors
- Sentinel-1 (C-band SAR, ESA): 5.6 cm wavelength; interferometric wide-swath mode at 5 × 20 m ground resolution; 6-day repeat at mid-latitudes with both satellites active. Free archive from 2014. Sufficient scatterer density on large masonry complexes; less discriminating on small or isolated structures.
- COSMO-SkyMed (X-band SAR, ASI): 3.1 cm wavelength; spotlight mode reaches 1 m resolution, enabling PS identification on individual stone courses and brick piers. Revisit of 1–4 days depending on constellation tasking. Shorter wavelength increases sensitivity to small displacements but also to atmospheric phase noise.
- Capella Space (X-band SAR, commercial): Spotlight resolution down to 0.35 m; on-demand tasking within hours. The fine resolution allows persistent scatterers to be pinned to specific architectural elements such as column drums or buttress faces. Archive depth is shorter than Sentinel-1, limiting long baseline analyses.
- ICEYE (X-band SAR, commercial): Sub-1 m spotlight mode; daily revisit on selected orbits. Useful for monitoring active construction phases where weekly cadence is insufficient to catch rapid displacement events.
Why a passing lorry shows up in a radar archive
Every time a SAR satellite passes over, it measures the two-way travel time of a microwave pulse to each point on the ground. If the ground moves between passes, the phase of the returning signal shifts. For C-band Sentinel-1, one full phase cycle corresponds to roughly 2.8 cm of motion along the satellite's line of sight. X-band systems halve that figure to about 1.5 cm per cycle, which is why COSMO-SkyMed and Capella Space data have been preferred in published monument studies where the expected displacements are small.
The complication is that phase measurements are ambiguous: you cannot tell from a single interferogram whether a point moved 1 mm or 28 mm plus 1 mm. Persistent-scatterer InSAR resolves this by stacking dozens or hundreds of interferograms over time and fitting a displacement model to the phase history of each stable reflector. The technique was formalised by Ferretti et al. at Politecnico di Milano in the late 1990s and has since been applied to heritage structures in peer-reviewed literature covering Rome, Istanbul, Lisbon and several other cities with dense historic fabric.
What a persistent scatterer actually is on a monument
A persistent scatterer (PS) is a point that maintains a stable radar reflection across all acquisition geometries and weather conditions. On bare ground or vegetation, coherence collapses between passes. On masonry, metal fixings, cornices, exposed brick and cut stone, it is maintained. This is why PS-InSAR is particularly well suited to standing monuments: the material properties that make them archaeologically significant, hard, angular, durable surfaces, are exactly the properties that produce good scatterers.
In practice, PS density on a large structure such as the Colosseum or the Theodosian Walls can reach several hundred points per hectare with X-band data. Each point gets an independent displacement time series. That spatial resolution within the structure is the key advantage over conventional geodetic levelling or crack gauges, which sample only where instruments are physically installed. The satellite samples the whole exposed fabric simultaneously, every pass, without touching it.
The line-of-sight problem and how two passes fix it
SAR measures displacement only along the slant range between satellite and target, not in true vertical or horizontal coordinates. A wall leaning toward the sensor looks identical in the interferogram to a wall settling vertically, if the geometry is right. This is the line-of-sight ambiguity, and ignoring it has caused misinterpretation in early published studies.
The standard remedy is to process both ascending and descending passes separately, then decompose the two line-of-sight displacement vectors into vertical and east-west horizontal components. North-south motion remains poorly constrained by this approach because SAR orbits are near-polar and the satellite is almost insensitive to motion along its flight direction. For a monument subject to tunnelling-induced settlement, the vertical component is usually dominant and well recovered. For a structure being pushed laterally by a retaining wall failure or differential foundation loading, the horizontal component matters and requires careful geometric decomposition. Sentinel-1 provides both ascending and descending passes with good regularity; COSMO-SkyMed and Capella can be tasked in both geometries but at higher cost.
Atmospheric phase delay is the other major error source. Water vapour gradients introduce apparent displacement signals of several millimetres that are spatially correlated and can mimic real structural motion. Corrections using ERA5 reanalysis weather data or GACOS (Generic Atmospheric Correction Online Service) are now standard practice and reduce residual atmospheric noise to roughly 1 mm or below in favourable conditions.
Documented cases: Rome and Istanbul
The Colosseum has been studied with both Sentinel-1 and COSMO-SkyMed data in the context of Rome's ongoing metro and road infrastructure works. Published analyses have identified differential settlement patterns across the structure's north and south sides, consistent with the known asymmetry of its foundation conditions, with displacement rates in the range of 1 to 3 mm per year. Whether any component is attributable to traffic vibration rather than long-term consolidation remains an active research question, which illustrates an honest limit of the method: PS-InSAR measures cumulative displacement, not instantaneous vibration. It cannot directly record the dynamic loading event, only its residual deformation.
Istanbul's Theodosian Walls, a UNESCO World Heritage site, sit alongside busy arterial roads and have been monitored using Sentinel-1 time series. Studies have mapped sections showing accelerated displacement coinciding with road-widening works, providing the kind of before-and-after evidence that heritage authorities need to support enforcement action. The walls' long linear geometry is well suited to PS analysis because the ascending and descending decomposition is geometrically tractable along most of the circuit.
Honest limits of the method
PS-InSAR is not a vibration sensor. A lorry crossing a bridge induces dynamic strains at frequencies of 1 to 20 Hz; a SAR satellite samples the same point perhaps once every six days. What the method captures is the slow accumulation of irreversible deformation after many loading cycles, not the instantaneous dynamic event. For dynamic monitoring, accelerometers or fibre-optic strain gauges installed on the structure are the appropriate tool. PS-InSAR complements them by providing spatial coverage they cannot match.
Scatterer density drops sharply on rubble-core walls with rough or vegetated surfaces, on earthen structures, and on any masonry that has been repointed with modern mortar that absorbs rather than reflects radar. Monuments in dense urban canyons can suffer geometric distortion from layover and shadow, particularly with steep-incidence X-band data. A site assessment, checking scatterer density on existing archive data before committing to a monitoring programme, is essential and takes a few days of processing.
Satellize runs PS-InSAR workflows on Sentinel-1 open data and can add COSMO-SkyMed or Capella tasking under client licence for sites where C-band scatterer density is insufficient or where faster revisit is needed during active construction phases.
Setting up a practical monitoring programme
A credible programme needs a baseline period of at least twelve months of pre-construction data to establish natural seasonal displacement patterns, which can reach 2 to 4 mm peak-to-peak on structures with clay-rich foundations responding to moisture. Without that baseline, construction-induced signals cannot be separated from seasonal noise.
Alert thresholds are typically set in consultation with structural engineers. A common approach is to flag any PS showing displacement exceeding two standard deviations of its pre-construction rate, or any spatial cluster of PS points showing coherent movement above 0.5 mm per month. These thresholds are not universal; they depend on foundation type, structure age and the sensitivity of the heritage authority. The output is a GIS layer updated on each satellite pass, with time-series plots for flagged points delivered as a PDF or web dashboard report. That is the product, not a promise of safety, but a quantified early warning that something has changed and warrants ground investigation.
Typical figures
| Spatial resolution (C-band, Sentinel-1 IW) | 5 × 20 m ground pixel; PS points resolved to sub-pixel positions by phase centroid fitting |
| Spatial resolution (X-band spotlight, COSMO-SkyMed / Capella) | 0.35 to 1 m ground pixel; PS density typically several hundred points per hectare on masonry |
| Revisit period | 6 days (Sentinel-1, both satellites); 1 to 4 days (COSMO-SkyMed tasked); sub-daily possible (ICEYE) |
| Minimum detectable displacement (line-of-sight, per epoch) | Approximately 1 to 3 mm per epoch after atmospheric correction; cumulative rates below 0.5 mm per year detectable over multi-year stacks |
| Radar frequency | C-band 5.405 GHz (Sentinel-1); X-band ~9.6 GHz (COSMO-SkyMed, Capella, ICEYE) |
| Archive depth | Sentinel-1 from April 2014; COSMO-SkyMed from 2007 (tasked archive); Capella from 2019 |
| Displacement components recoverable | Vertical and east-west horizontal (ascending + descending decomposition); north-south poorly constrained |
| Atmospheric correction | ERA5 or GACOS tropospheric delay models; residual noise approximately 1 mm in favourable conditions |
| Deliverable formats | GeoTIFF displacement maps, CSV or GeoJSON PS time-series tables, PDF monitoring reports, web dashboard layers |
Analytics Satellize can run
| PS-InSAR displacement time series per structure | Persistent-scatterer interferometry (Ferretti et al. method class); multi-temporal stack of 30 or more interferograms | GeoJSON point layer with per-PS displacement history; updated on each satellite pass |
| Ascending/descending decomposition into vertical and horizontal components | Geometric decomposition of two line-of-sight vectors; standard two-geometry inversion | GeoTIFF rasters of vertical and east-west velocity fields; PDF interpretation note |
| Construction-phase change detection alert | Statistical comparison of post-construction PS rates against pre-construction baseline; two-sigma threshold flagging | Automated alert report listing flagged PS clusters with map and time-series plot; delivered within 48 hours of new acquisition |
| Seasonal baseline characterisation | Harmonic regression on pre-construction PS time series to isolate annual moisture and thermal cycles | Baseline report quantifying natural displacement amplitude per structure zone; used to set alert thresholds |
| Scatterer density feasibility assessment | Coherence and amplitude stability analysis on existing archive data before programme commitment | Site suitability report with PS density map and sensor recommendation (C-band vs X-band); delivered within one week |
| Multi-epoch differential settlement map | Spatial clustering of PS velocity vectors to identify zones of differential movement across a structure's footprint | Annotated GIS layer showing settlement gradient vectors; suitable for structural engineer review |
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.