Post-reclamation land consolidation settlement monitoring
Hydraulically placed fill consolidates unevenly for years after reclamation ends. Persistent-scatterer and small-baseline InSAR time-series on Sentinel-1 and ALOS-2 PALSAR-2 map settlement velocity fields and separate consolidation trends from seasonal groundwater signals, but decorrelation on bare fill demands careful network design.
Sensors
- Sentinel-1 A/B (ESA): C-band SAR at 5.6 cm wavelength; Interferometric Wide Swath mode delivers 5 x 20 m ground range resolution across a 250 km swath. Six-day repeat at mid-latitudes when both satellites are operational, twelve-day with one. Free archive from 2014. C-band coherence degrades quickly on loose, unvegetated fill, limiting PS density and pushing users toward SBAS or hybrid approaches.
- ALOS-2 PALSAR-2 (JAXA): L-band SAR at 23.6 cm wavelength; StripMap mode offers 3 m x 1 m resolution, ScanSAR up to 350 km swath at coarser resolution. L-band penetrates dry fill surface layers and maintains coherence over longer temporal baselines than C-band, making it substantially better for sparse-scatterer environments. Nominal 14-day repeat; tasking required, not free.
- COSMO-SkyMed SkyMed-2 (ASI): X-band SAR at 3.1 cm wavelength; Spotlight mode achieves sub-metre resolution down to approximately 0.35 m, StripMap at 3 m. High resolution supports dense PS networks on constructed infrastructure edges and corner-reflector targets. Constellation revisit can reach daily under tasking. Commercially licensed; cost scales with archive depth and tasking volume.
- Deployed corner reflector network (ground infrastructure): Not a satellite sensor, but an essential complement. Trihedral corner reflectors with side lengths of 0.5 m or larger produce a known, stable radar cross-section that anchors InSAR velocity fields on otherwise decorrelating bare fill. JAXA and ESA published guidelines recommend spacing of one reflector per 1-2 km² in low-coherence zones to constrain atmospheric phase screen estimation.
Why reclaimed fill is an unusually difficult InSAR target
Hydraulically placed fill is not a single material. It is a slurry of dredged sand, silt and clay pumped behind a bund, then left to drain and consolidate under its own weight and any surcharge. The surface is loose, heterogeneous and often actively regraded during the early years. That combination destroys the phase stability that InSAR depends on.
Coherence, the statistical measure of how reliably a pixel scatters between two SAR acquisitions, drops sharply when surface particles shift even a few millimetres between passes. On bare hydraulic fill, C-band coherence at Sentinel-1's six-day repeat can fall below 0.3 across large areas, leaving too few reliable measurement points for a useful velocity map. L-band is more forgiving because its longer wavelength is less sensitive to small surface changes, but it is not immune. The practical consequence is that a naive application of standard persistent-scatterer InSAR to a freshly reclaimed site will produce a sparse, noisy result that engineers cannot safely act on.
What the physics of consolidation looks like in a time-series
Primary consolidation, the expulsion of pore water from fine-grained layers under load, follows a broadly predictable curve: rapid early settlement that decelerates as excess pore pressure dissipates. Secondary consolidation, creep of the soil skeleton itself, continues at a slower rate for years or decades. Both produce vertical displacement that InSAR can detect, provided the measurement geometry is understood.
Sentinel-1 in its standard ascending and descending geometries measures line-of-sight displacement, not pure vertical. Decomposing that into vertical and east-west horizontal components requires combining ascending and descending passes. On a flat reclamation site where horizontal movement is small, the line-of-sight signal is dominated by vertical settlement, and the approximation is usually acceptable. Typical primary consolidation rates on soft-clay-underlain reclamations range from tens of millimetres per year in the early post-reclamation period to a few millimetres per year once primary consolidation is largely complete. InSAR detection thresholds for well-anchored PS networks are generally cited in the literature at 1 to 2 mm per year for velocity and 3 to 5 mm for individual displacement events, though those figures assume good coherence and dense PS coverage.
Seasonal groundwater fluctuations complicate the picture. Aquifer recharge and drawdown cause reversible elastic deformation of the ground surface that can reach 10 to 20 mm amplitude in some coastal settings. That signal rides on top of the irreversible consolidation trend. Separating them requires time-series long enough to observe at least two full annual cycles, combined with auxiliary groundwater level data where available.
Building a measurement network that actually works
The practical answer to low coherence on bare fill is a hybrid strategy. Permanent infrastructure on or near the site, including retaining walls, drainage structures, road edges and any early buildings, provides natural persistent scatterers. These anchor the velocity field in areas where the fill itself gives nothing reliable. Corner reflectors, installed on stable short piles driven to refusal in the underlying competent stratum, extend coverage into the open fill areas between structures.
Reflector design matters. A trihedral corner reflector with a side length of 0.5 m gives a radar cross-section of roughly 20 to 25 dBm² at C-band, which is detectable against typical fill backgrounds. Larger reflectors improve signal-to-noise but become unwieldy in coastal environments exposed to wind. Orientation must be set to the satellite's incidence angle for each pass direction, and the reflectors must be surveyed by differential GPS to millimetre accuracy at installation so that InSAR-derived displacements can be validated against levelling benchmarks.
SBAS (Small Baseline Subset) processing complements PS methods on reclaimed sites. By forming interferograms only between acquisition pairs with short temporal and spatial baselines, SBAS preserves distributed scatterer coherence better than long-baseline pairs. The two methods can be run in parallel and their outputs compared as a consistency check.
Distinguishing consolidation from everything else that moves
A velocity map on its own is not an answer. A site engineer needs to know whether a measured 15 mm/year subsidence rate represents ongoing primary consolidation that will slow, secondary creep that will persist, a localised weak zone that demands ground improvement, or an artefact of atmospheric delay that should be discarded.
Atmospheric phase screen correction is the most consequential processing step. Tropospheric water vapour introduces path delay that can mimic several centimetres of displacement in a single interferogram. ERA5 reanalysis data from the Copernicus Climate Change Service, or empirical corrections derived from the InSAR data itself using methods such as the power-law approach of Bekaert et al., substantially reduce this noise. Ionospheric delay is less significant at C-band for mid-latitude sites but becomes relevant for L-band ALOS-2 data in equatorial regions.
Once atmospheric artefacts are suppressed, the residual time-series can be fitted to consolidation models. A simple hyperbolic or logarithmic fit to the settlement-time curve gives an estimate of when primary consolidation will be substantially complete. Deviations from the expected curve, particularly accelerations or spatial discontinuities, are the signals that warrant engineering investigation.
What the data cannot tell you, and what fills the gap
InSAR measures surface displacement at the locations of coherent scatterers. It says nothing directly about what is happening at depth. A 20 mm surface settlement could be driven by a thin compressible clay lens at 3 m depth or by a thick deposit at 15 m. Distinguishing those scenarios requires geotechnical borehole data and piezometer records. Satellite monitoring is most valuable when it is integrated with a subsurface model, not treated as a substitute for one.
There is also a resolution floor. Standard Sentinel-1 IW mode at 5 x 20 m cannot resolve differential settlement between adjacent pile caps separated by 2 m. For that scale, X-band systems such as COSMO-SkyMed in Spotlight mode, or ground-based radar, are more appropriate. The satellite time-series is best understood as a site-wide screening tool that identifies zones requiring closer investigation, not as a replacement for precision levelling on individual foundation elements.
Satellize processes Sentinel-1 and ALOS-2 archives for consolidation monitoring programmes, incorporating corner-reflector networks where clients can deploy them on site. The analytics pipeline outputs velocity maps, time-series plots per zone and threshold alerts when settlement rates exceed agreed limits, delivered as GIS layers compatible with standard engineering software. The approach is the same whether the site is a single reclaimed island or a multi-phase port expansion covering several square kilometres.
Typical figures
| Spatial resolution (Sentinel-1 IW) | 5 m range x 20 m azimuth (ground range); PS/SBAS products typically gridded at 20-100 m |
| Spatial resolution (ALOS-2 StripMap) | 3 m x 1 m; PS products gridded at 10-30 m depending on PS density |
| Spatial resolution (COSMO-SkyMed Spotlight) | 0.35-1 m; supports dense PS networks on engineered structures |
| Revisit interval | Sentinel-1: 6 days (dual satellite); ALOS-2: 14 days (tasked); COSMO-SkyMed: 1-4 days (tasked) |
| Minimum detectable velocity (well-anchored PS network) | 1-2 mm/year for long time-series (>2 years); individual epoch precision 3-5 mm |
| SAR frequency / wavelength | C-band 5.6 cm (Sentinel-1); L-band 23.6 cm (ALOS-2); X-band 3.1 cm (COSMO-SkyMed) |
| Archive depth | Sentinel-1: from 2014; ALOS-2: from 2014; COSMO-SkyMed: from 2007 (commercial access) |
| Atmospheric correction | ERA5 reanalysis or empirical power-law correction; residual tropospheric noise typically 3-8 mm per interferogram after correction |
| Typical processing latency | 5-15 days from acquisition to validated velocity update for operational monitoring |
| Delivery formats | GeoTIFF velocity maps, CSV time-series per zone, GeoPackage/Shapefile PS point layers, PDF engineering summary |
Analytics Satellize can run
| Site-wide settlement velocity map | Persistent-scatterer InSAR (PSInSAR) or SBAS time-series analysis on Sentinel-1 or ALOS-2 stack | GeoTIFF raster and PS point layer showing mm/year line-of-sight velocity, updated quarterly or monthly |
| Consolidation trend vs seasonal signal decomposition | Time-series decomposition fitting hyperbolic/logarithmic consolidation model plus sinusoidal seasonal term to PS displacement history | Zone-level report with fitted curves, estimated primary consolidation completion date and residual creep rate |
| Corner reflector displacement time-series | PS analysis anchored to reflectors surveyed by differential GPS; displacement validated against levelling benchmarks | CSV time-series per reflector with uncertainty bounds, plus comparison table against ground-truth levelling data |
| Differential settlement risk zones | Spatial gradient analysis of velocity field; zones where adjacent PS points exceed a defined differential threshold flagged | GIS polygon layer of risk zones with severity classification, fed into client engineering review workflow |
| Settlement rate threshold alert | Automated monitoring of PS velocity estimates against client-defined engineering limits (e.g. >10 mm/year triggers review) | Email or API alert with zone identifier, current rate, trend direction and link to updated GIS layer |
| Vertical vs horizontal displacement decomposition | Combination of ascending and descending Sentinel-1 passes to separate vertical and east-west displacement components | Paired GeoTIFF maps (vertical, east-west) with uncertainty estimates; relevant where lateral fill movement is suspected |
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.