Bridge deformation monitoring with InSAR
Spaceborne SAR interferometry can resolve millimetre-scale deformation on individual bridge decks, separating daily thermal expansion from the slower, more dangerous signals of structural settlement or fatigue.
Sensors
- TerraSAR-X Staring Spotlight: Ground resolution of approximately 0.25 m in azimuth by 0.25 m in range, revisit of 11 days in standard repeat-pass mode (shorter with off-axis tasking). The only spaceborne SAR mode that can isolate individual bridge piers and deck segments for persistent scatterer analysis.
- COSMO-SkyMed Second Generation (CSG): Spotlight mode delivers 0.35 m resolution; the two-satellite constellation reduces revisit to roughly 4 to 8 days. X-band wavelength (3.1 cm) gives high sensitivity to small displacements but also means phase wrapping occurs at about 1.55 mm of displacement per half-cycle, requiring careful unwrapping on structures with large thermal excursions.
- ICEYE X-band SAR: Spotlight resolution of approximately 0.5 m, with a large constellation enabling revisit intervals of 1 to 3 days. Useful for rapid-cadence monitoring after a seismic event or suspected structural incident, though the shorter archive depth compared to TerraSAR-X limits long-baseline time-series analysis.
- Sentinel-1 SAR: 6-day repeat (12-day per satellite before the loss of Sentinel-1B in 2021, recovering with Sentinel-1C). IW mode resolution of 5 m by 20 m is insufficient to isolate a single bridge deck, but the free archive back to 2014 makes it useful for detecting gross settlement trends on wide structures or approach embankments, and for calibrating longer-term drift before high-resolution tasking begins.
What a bridge actually looks like to a SAR satellite
A SAR satellite does not photograph a bridge; it measures the round-trip travel time of microwave pulses to every surface that reflects energy back towards the sensor. On a bridge deck, the most reliable reflectors are corner-like geometries: expansion joints, parapet edges, lamp-post bases and the junctions between girders and deck plates. These are called persistent scatterers (PS), and their phase history across dozens of acquisitions is what produces a displacement time series.
The geometry matters enormously. SAR sensors image from a side-looking angle, typically 20 to 45 degrees from vertical. A bridge oriented perpendicular to the satellite's flight path presents its full deck width to the sensor and yields many PS candidates. A bridge running parallel to the flight path may appear foreshortened or, in extreme cases, partially layover-affected, collapsing deck and parapet returns into a single pixel. Selecting the right satellite pass geometry for a given bridge orientation is not optional; it is the first design decision in any monitoring programme.
Thermal expansion is not a structural problem, but it looks like one
A 500-metre concrete bridge span can expand by 15 to 25 mm between a cold winter night and a hot summer afternoon. That is a legitimate engineering quantity, not a defect. But InSAR measures displacement in the satellite's line-of-sight direction, which is a projection of all three-dimensional movement. Thermal expansion along the bridge axis, vertical deflection under traffic load and genuine settlement of a pier foundation all contribute to the same phase measurement. Separating them requires either multi-geometry acquisitions (ascending and descending passes, ideally with different incidence angles) or a thermal correction model built from air temperature or bridge-mounted sensor records.
The ambiguity is real and should not be minimised. A single-geometry PS-InSAR time series showing a 3 mm seasonal oscillation could be a healthy bridge breathing normally or an early indicator of bearing failure allowing anomalous lateral movement. The diagnostic value comes from the residual after the thermal model is removed. If that residual shows a monotonic trend, accelerating displacement or a step change coinciding with a heavy-load event, structural investigation is warranted. If it is noise around zero, the bridge is behaving as designed.
Corner reflectors: the calibration tool most programmes skip
A trihedral corner reflector, a metal corner typically 30 to 60 cm along each face, produces a radar cross-section orders of magnitude larger than natural scatterers at the same wavelength. Installed on a bridge deck at known positions and surveyed by differential GPS, it provides a ground-truth displacement reference that ties the InSAR phase measurements to absolute millimetric accuracy. Without it, PS-InSAR results are relative: displacement of one scatterer relative to another, with any common-mode error (atmospheric phase screen, orbital uncertainty) absorbed into the network.
Corner reflectors are particularly valuable on bridges where natural PS density is low, such as modern composite-deck structures with smooth surfaces and few metallic protrusions. They are also the only reliable way to validate that the atmospheric correction applied to a time series is working correctly over a water crossing, where standard tropospheric models perform poorly due to the abrupt humidity discontinuity between river surface and deck. The cost of deploying two or three reflectors is trivial against the cost of a missed structural alert.
What the numbers look like in practice
TerraSAR-X Staring Spotlight PS-InSAR studies published in peer-reviewed literature have reported displacement precision of 0.3 to 1.0 mm per acquisition in line-of-sight, depending on PS density, atmospheric conditions and the length of the time series used to estimate the atmospheric phase screen. That is sufficient to detect the early stages of pier settlement, which in published case studies of failing infrastructure typically manifests as trends of 1 to 5 mm per year before visible cracking appears.
Traffic-load deflection is harder. A heavy lorry crossing a span causes deflection on the order of a few millimetres over seconds, far faster than the minutes-long synthetic aperture integration time. What InSAR captures is not the instantaneous deflection but the mean displacement state during the acquisition window, which varies with traffic density at the moment of overpass. This is a genuine limitation. Continuous structural health monitoring sensors remain necessary for dynamic load assessment; InSAR is the complement, not the replacement, providing the slow-drift and long-term settlement picture that in-situ sensors rarely sustain over decades.
Designing a monitoring programme that will still be useful in ten years
The Sentinel-1 archive, free and consistent since 2014, is the natural baseline layer. Running a PS or SBAS (Small Baseline Subset) analysis over the archive gives the pre-programme deformation history at coarse resolution, identifying whether a structure has been stable or has exhibited trends before high-resolution tasking begins. This matters for project finance and insurance underwriting: a clean ten-year Sentinel-1 record is meaningful due-diligence evidence.
High-resolution tasking with TerraSAR-X or CSG should then be scoped around the specific spans of concern, the acquisition geometry that maximises PS density for that bridge's orientation, and a revisit cadence matched to the expected deformation rate. For a bridge under active load monitoring after a detected anomaly, monthly acquisitions may be insufficient; for routine long-term surveillance of a stable crossing, quarterly is defensible. Satellize structures these monitoring cadences as part of its analytics engagements, applying the same time-series methodology it uses in its Tonga crop-estimation programme to infrastructure displacement problems.
One practical note on archive depth: COSMO-SkyMed first-generation data extends back to around 2007 for some geometries, and TerraSAR-X to 2007 as well. Accessing this archive for a specific bridge requires a formal data request and is not always guaranteed, but it is worth attempting before commissioning new acquisitions.
Typical figures
| Best spatial resolution (single-look) | ~0.25 m (TerraSAR-X Staring Spotlight); ~0.35 m (COSMO-SkyMed Second Generation Spotlight) |
| Displacement precision (PS-InSAR, per acquisition) | 0.3 to 1.0 mm line-of-sight, dependent on PS density and atmospheric correction quality |
| Revisit interval | 11 days (TerraSAR-X standard); 4 to 8 days (CSG two-satellite); 1 to 3 days (ICEYE); 6 days (Sentinel-1C) |
| SAR frequency / wavelength | X-band, 9.6 GHz / 3.1 cm (TerraSAR-X, CSG, ICEYE); C-band, 5.4 GHz / 5.6 cm (Sentinel-1) |
| Minimum detectable long-term trend | ~0.5 to 1 mm/year over a multi-year time series with sufficient PS density |
| Phase ambiguity (half-wavelength) | ~1.55 mm per fringe at X-band; ~2.8 mm per fringe at C-band |
| Archive depth | TerraSAR-X from ~2007; Sentinel-1 from 2014; ICEYE from ~2019; CSG from ~2019 |
| Typical scene width (high-resolution modes) | 4 to 10 km swath (Spotlight modes); 250 km (Sentinel-1 IW) |
| Delivery formats | GeoTIFF displacement maps, CSV/GeoJSON PS time-series, KMZ for GIS import, PDF structural summary report |
| Weather / cloud sensitivity | None for SAR; all-weather, day-and-night acquisition. Atmospheric water vapour introduces phase noise requiring correction. |
Analytics Satellize can run
| Persistent Scatterer displacement time series | PS-InSAR (Ferretti et al. method class) applied to multi-temporal TerraSAR-X or CSG Spotlight stack | GeoJSON point layer with per-scatterer velocity and seasonal amplitude, updated per new acquisition |
| Thermal deformation model and structural residual | Linear regression of PS phase against recorded or ERA5-derived temperature, residual extraction | Time-series chart separating thermal, traffic-correlated and trend components; PDF technical note |
| Baseline deformation history from Sentinel-1 archive | SBAS processing of free Sentinel-1 IW archive (2014 to present) for approach embankments and gross deck trends | GeoTIFF mean velocity map and displacement time series for the bridge corridor, supplied as project baseline layer |
| Corner reflector calibration report | Phase comparison of installed trihedral reflectors against GPS-surveyed ground truth to validate absolute displacement accuracy | Calibration certificate with residual error statistics, referenced to national geodetic datum |
| Anomaly alert | Statistical threshold detection on PS residual time series; step-change and acceleration detection | Automated alert (email or API webhook) triggered when displacement rate or step exceeds client-defined threshold |
| Multi-geometry 3D decomposition | Combination of ascending and descending pass line-of-sight measurements to decompose vertical and horizontal displacement components | Separate vertical and along-bridge horizontal displacement maps, GeoTIFF format |
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.