Methodology
How satellites measure ground movement — and how far to trust it
No black boxes. This page explains the physics, the processing, and — just as carefully — the limitations that every honest report must document.
Foundations
Synthetic Aperture Radar (SAR)
Radar satellites illuminate the Earth with microwave pulses and record the echo. Because they carry their own illumination, they observe day and night, and at the wavelengths used they can image through cloud cover in most conditions — a decisive advantage over optical imagery for systematic monitoring.
"Synthetic aperture" refers to how the sensor combines echoes along its orbit to synthesise a much larger antenna, achieving fine ground resolution from ~700 km altitude. Each acquisition records not just the strength of the echo but its phase — a precise measure of the distance travelled by the signal. Phase is what makes deformation measurement possible.
Interferometry
InSAR
Interferometric SAR compares the phase of two radar images of the same area taken at different times. If the ground moved toward or away from the satellite between acquisitions, the echo's travel distance changed, and the phase difference records it — at a sensitivity of a fraction of the radar wavelength (centimetres).
The raw phase difference mixes ground movement with topography, orbital geometry, and atmospheric delay. InSAR processing is largely the discipline of separating these: removing what is not movement, and quantifying how well the separation worked. The result is an estimate of surface displacement along the satellite's line of sight for each resolution cell.
Time series
Persistent Scatterer InSAR (PSInSAR)
A single interferogram is noisy. PSInSAR overcomes this by analysing large stacks — often a hundred or more acquisitions — and identifying persistent scatterers: points such as building corners, roof edges, and exposed rock whose radar reflection stays stable across years.
For each of these points, the full acquisition history is combined into a displacement time series, with atmospheric and orbital errors estimated statistically across the stack. In built environments this yields hundreds to thousands of measurement points per square kilometre, each with a multi-year movement history and a quality estimate.
Data sources
Satellite missions
Analyses may draw on Sentinel-1 (the free, systematic backbone: global coverage every 6–12 days since late 2014), TerraSAR-X and COSMO-SkyMed (high-resolution X-band commercial missions), and other suitable public or commercial SAR missions where a project justifies them.
Commercial data is proposed only when it is actually available and licensed for the specific project — availability is confirmed during the data review, never assumed.
Workflow
From archive to report in ten steps
Area definition
The site, corridor, or region of interest is defined precisely, with a surrounding reference zone.
Data search
Available radar archives are searched across missions, geometries, and years.
Acquisition review
Individual acquisitions are screened for suitability and continuity.
Interferometric processing
Phase differences between acquisitions are computed across the image stack.
Persistent scatterer selection
Stable radar-reflective points are identified for reliable long-term measurement.
Error mitigation
Atmospheric and orbital contributions are estimated and reduced.
Time-series estimation
A displacement history is derived for every retained measurement point.
Quality control
Statistical and spatial checks remove unreliable points and quantify confidence.
Spatial interpretation
Patterns are interpreted in the context of the asset, geology, and land use.
Report preparation
Findings, maps, uncertainty, and recommendations are written for the decision at hand.
Important limitations
What every honest report must say
These are not footnotes — they determine how results may be used. Each report documents which of these apply to the specific site and how strongly.
- Line-of-sight measurement
- Movement is measured along the satellite's viewing direction, not as pure vertical or horizontal displacement. Combining ascending and descending geometries helps separate components; north–south horizontal motion remains poorly observed.
- Data gaps
- Archives are not uniform. Some regions and periods have dense coverage; others have gaps that limit temporal resolution or the analysable period.
- Vegetation and temporal decorrelation
- Vegetated, ploughed, or frequently changing surfaces scramble the radar signal between acquisitions, leaving few or no reliable measurement points.
- Atmospheric effects
- Water vapour delays the radar signal and can mimic movement. Multi-temporal methods mitigate this statistically, but residual atmospheric noise contributes to uncertainty.
- Geometric distortions
- Steep terrain and tall structures cause layover and shadowing that can hide surfaces from the radar or distort their apparent position.
- Ascending vs descending
- The two viewing directions see the same ground differently. Results from one geometry alone carry directional blind spots that the report must state.
- Reference point dependence
- All InSAR movement is relative to a chosen reference area assumed stable. Reference choice is documented; absolute movement requires external data such as GNSS.
- Need for engineering context
- Surface movement patterns acquire meaning only in context — foundation type, geology, drainage, and site history. Interpretation without context invites error.
- Variable resolution
- Ground resolution and measurement density differ between missions (tens of metres for Sentinel-1; metres for high-resolution commercial sensors) and between surfaces.
- Historical archive limits
- Systematic free coverage begins in late 2014 (Sentinel-1). Earlier analysis depends on older missions having imaged the site, which varies by location.
- Sensor coverage differences
- Not every mission covers every site in every geometry. What is achievable is established per site during the data availability review.
- Measurement uncertainty
- Every velocity and displacement estimate carries uncertainty from noise, processing choices, and reference stability. Reports state it rather than hiding it.
- Ground-truth integration
- For some questions, GNSS, levelling, geotechnical, geological, or field data are necessary complements — and the report says when that is the case.
Quality and transparency
Documented in every report
A result you cannot audit is a result you cannot use. Reports therefore always document:
- Sensor(s) and acquisition period used
- Number of observations, where appropriate
- Processing approach and key parameters
- Reference framework and reference-area choice
- Data quality and measurement density
- Interpretation assumptions
- Known limitations for this specific site
- Confidence level of each conclusion
Technical questions
Frequently asked
What is InSAR?
InSAR (Interferometric Synthetic Aperture Radar) is a technique that compares radar images acquired by satellites at different times. By analysing the phase difference between acquisitions, it is possible to estimate how the ground surface has moved along the satellite's line of sight — often at the level of millimetres to centimetres per year, depending on conditions.
How accurate are the measurements?
Achievable precision depends on data availability, surface characteristics, atmospheric conditions, viewing geometry, processing method, temporal coverage, and validation. Under favourable conditions, multi-temporal InSAR velocity estimates can reach millimetre-per-year precision, but this cannot be assumed universally. Every report documents the confidence level and uncertainty of its measurements.
How much historical data is available?
Sentinel-1 provides free, systematic global coverage from late 2014 onwards, typically every 6–12 days in most regions. Earlier or higher-resolution archives (e.g. TerraSAR-X, COSMO-SkyMed, ERS/Envisat) exist for many areas but vary by location. The data availability review establishes exactly what archive exists for your site.
Do you use public or commercial satellite data?
Both, depending on the project. Sentinel-1 (public, free) is the default backbone for most screening work. Where higher resolution or a longer archive is needed and justified, commercial missions such as TerraSAR-X or COSMO-SkyMed may be proposed — always as an explicit, costed option, never assumed.
What happens if suitable satellite data are unavailable?
We tell you, before you commit. If the data availability review finds insufficient coverage, poor measurement conditions, or unsuitable surface characteristics, we explain why and — where possible — suggest alternatives such as different sensors, a modified area of interest, or ground-based techniques. Not every site can be analysed successfully, and we say so.
Can results be integrated into GIS systems?
Yes. Outputs can be delivered as GeoJSON, Shapefile, GeoTIFF, CSV with coordinates, or KML, ready for QGIS, ArcGIS, or web mapping platforms, alongside the PDF report.
Start a conversation
Want to know what the data can support for your site?
The data availability review answers that question specifically — before any analysis is quoted.