Peatland subsidence monitoring for buried cable route integrity
Organic peat soils can subside several centimetres per year under drainage and drying, imposing differential settlement on buried ducts and direct-buried fibre. InSAR time-series methods applied to Sentinel-1 and ALOS-2 PALSAR-2 data can detect and map these movements along cable corridors, within the coherence limits that wet peat imposes.
Sensors
- Sentinel-1 A/B (C-band SAR, 5.6 cm wavelength): Interferometric Wide Swath mode at 5 x 20 m ground range resolution, 6-day revisit over Europe with two satellites. C-band is sensitive to surface and near-surface dielectric change on wet peat, but temporal decorrelation over vegetated bog surfaces is severe, typically limiting coherent pairs to 6-12 day baselines in summer and somewhat longer in winter when vegetation is dormant.
- ALOS-2 PALSAR-2 (L-band SAR, 23.6 cm wavelength): Stripmap mode at 3 x 3 m resolution; ScanSAR at 60 m. L-band penetrates herbaceous peat vegetation more effectively than C-band, maintaining coherence over longer temporal baselines of 14-42 days. This makes it substantially better suited to persistent-scatterer and small-baseline InSAR on vegetated mire, though revisit is 14 days and tasking is required outside standard acquisition plans.
- RADARSAT Constellation Mission (C-band SAR): Three-satellite constellation achieving 4-day revisit at mid-latitudes in compact polarimetry mode. Medium-resolution mode at 16 m supports small-baseline subsets over peatland corridors where Sentinel-1 coherence is marginal; the shorter revisit interval reduces temporal decorrelation compared with single-satellite C-band.
- Copernicus DEM GLO-30: Global 30 m digital elevation model derived from TanDEM-X, used to remove topographic phase from interferograms and to compute slope and drainage indices along cable corridors. Vertical accuracy is approximately 4 m LE90 globally, better in low-relief peatland terrain. Not a deformation product itself, but essential reference geometry for InSAR processing.
What peat actually does to a buried duct
Peat is not soil in the engineering sense. It is partially decomposed organic matter with a water content that can exceed 90% by mass when saturated. Drain it, dry it or load it, and it compresses. Rates of 1 to 5 cm per year are common in actively drained lowland raised mires in the UK, the Netherlands and Scandinavia. Tropical peatlands in Borneo and Sumatra have recorded subsidence exceeding 5 cm per year following drainage for agriculture. These are not worst-case figures; they are published field measurements from levelling surveys and GPS benchmarks.
A buried duct or direct-buried fibre cable is a rigid linear structure. The ground beneath it is not rigid. Where subsidence rates vary along a corridor, the cable experiences differential settlement: one section sinks faster than its neighbour, imposing bending stress and, over time, joint separation or duct cracking. Fibre itself tolerates very little longitudinal strain before attenuation rises or breaks occur. Engineers designing routes across peatland need to know not just whether subsidence is happening, but where the gradient of subsidence is steepest, because that gradient is the mechanical threat.
Why InSAR is the right tool, and where it fails
Synthetic aperture radar interferometry measures the phase difference between two radar acquisitions of the same scene. If the ground has moved between acquisitions, the phase shifts in proportion to the displacement along the line of sight, with millimetre-scale sensitivity under good conditions. Persistent-scatterer InSAR (PS-InSAR) identifies pixels that maintain stable phase over a long time series, typically corner reflectors, rocks or built structures, and estimates deformation rates at those points. Small-baseline subset (SBAS) methods relax the requirement for permanent scatterers by combining interferograms with short spatial and temporal baselines, accepting distributed coherence from patches of ground rather than point targets.
Wet, vegetated peat is one of the most challenging surfaces for either method. The vegetation canopy changes between acquisitions, the dielectric constant of the surface shifts with rainfall and frost, and there are almost no natural corner reflectors. C-band coherence over active blanket bog in summer can collapse entirely within 6 days. L-band fares better because the longer wavelength is less sensitive to centimetre-scale vegetation motion, but it does not eliminate the problem. In practice, useful PS density over open peatland is low: tens of scatterers per square kilometre rather than the thousands available in urban areas. Corner reflectors installed along the cable corridor can dramatically improve this, and their use is documented in infrastructure monitoring programmes in Scandinavia and the Netherlands. Without them, the analyst must accept that deformation estimates will be sparse and interpolated rather than continuous.
Corridor processing: from raw SAR to a subsidence rate map
Processing begins with co-registration of a stack of SAR acquisitions to a common reference geometry, using the Copernicus DEM GLO-30 to remove topographic phase. For SBAS analysis, interferograms are formed from pairs with short baselines, typically under 150 m perpendicular and under 48 days temporal for L-band, shorter for C-band. After phase unwrapping, atmospheric phase screens are estimated and removed, most commonly using the temporal behaviour of the phase signal combined, where available, with ERA5 reanalysis tropospheric delay data.
The output is a velocity map in millimetres per year in the radar line-of-sight direction, which must be projected onto the vertical using the known incidence angle, typically 30 to 46 degrees for Sentinel-1 IW mode. For a cable corridor, the analyst then extracts a longitudinal profile of vertical velocity, flags sections where the velocity gradient exceeds a threshold consistent with the cable's tolerable differential settlement, and ranks those sections by severity. Seasonal displacement, the annual compression and rebound cycle driven by water table fluctuation, is separated from the secular trend by fitting a model with both a linear rate and a sinusoidal seasonal component to the time series at each scatterer.
L-band versus C-band: a practical choice for peatland
The choice between Sentinel-1 and ALOS-2 PALSAR-2 is not a matter of one being better in the abstract. It depends on the peat type, the vegetation cover and the acceptable latency. For blanket bog with dense Sphagnum or cotton grass cover, L-band is almost always preferable: published studies comparing C- and L-band coherence over Irish and Finnish peatlands consistently show L-band maintaining coherence at 28-day intervals where C-band has decorrelated. For bare or sparsely vegetated peat surfaces, such as recently harvested peat extraction sites or dry tropical peat during drought, C-band can perform adequately and offers the advantage of free, open-access data with 6-day revisit.
RADARSAT Constellation Mission occupies a middle ground. Its 4-day revisit reduces the temporal decorrelation window compared with Sentinel-1, and its compact polarimetry mode provides additional scattering information that can help separate vegetation from ground motion signals. It is not a substitute for L-band over dense vegetation, but it is a practical option where Sentinel-1 coherence is marginal and ALOS-2 tasking costs are a constraint.
What the data cannot tell you
InSAR measures displacement in the radar line of sight. It cannot directly measure horizontal movement, though east-west and vertical components can be partially separated by combining ascending and descending pass data. It cannot see through dense forest canopy to the ground beneath, which limits its use on tropical peatland where tree cover is intact. It produces sparse point measurements over open peat, not a continuous surface, unless corner reflectors are installed. Atmospheric artefacts, particularly water vapour gradients over low-lying wet terrain, can mimic or mask real deformation signals at the centimetre level; careful atmospheric correction is not optional.
The method also has a minimum detectable rate. Over a one-year time series with monthly acquisitions, the noise floor for SBAS analysis is typically 2 to 5 mm per year in vertical displacement, depending on atmospheric conditions and scatterer density. Subsidence rates below that threshold are not detectable with confidence. For cable route integrity assessment, this is usually acceptable: the damage-relevant rates are well above the noise floor. But for early-warning of incipient movement on a newly drained corridor, the method may not give useful signal until the problem is already developing.
Turning a velocity map into a maintenance decision
A subsidence velocity map is not itself a maintenance schedule. The engineering step is to convert line-of-sight displacement rates and their spatial gradients into estimates of differential settlement over the cable burial depth and duct spacing, then compare those estimates against the manufacturer's specified bend radius and the operator's acceptable strain limits for the fibre. Sections where the annual differential settlement gradient exceeds a defined threshold become candidates for proactive inspection, re-routing or installation of flexible duct couplings.
Satellize can run corridor-specific SBAS processing on Sentinel-1 or ALOS-2 archives and deliver annual velocity profiles as GIS layers aligned to the cable route centreline, with flagged high-gradient segments and seasonal amplitude maps. The same processing chain underpins the ground-deformation analytics we apply in other contexts, including the Tonga crop-estimation programme where surface change detection over small agricultural parcels shares some of the same coherence challenges as vegetated peat. For a cable operator, the practical next step is to define the route corridor geometry and identify which peat classification zones it crosses, so that sensor selection and archive depth can be scoped before processing begins.
Typical figures
| Spatial resolution (Sentinel-1 IW) | 5 x 20 m ground range; PS/SBAS output typically gridded at 20-50 m along corridor |
| Spatial resolution (ALOS-2 PALSAR-2 Stripmap) | 3 x 3 m single-look; multi-looked to 10-30 m for InSAR processing |
| Revisit interval | 6 days (Sentinel-1, two satellites, Europe); 14 days (ALOS-2); 4 days (RADARSAT Constellation Mission) |
| Minimum detectable vertical rate | Approximately 2-5 mm/year over a 12-month time series under typical atmospheric conditions; higher uncertainty over wet peat with low scatterer density |
| Radar frequency / wavelength | C-band 5.6 cm (Sentinel-1, RCM); L-band 23.6 cm (ALOS-2 PALSAR-2) |
| Reference DEM | Copernicus DEM GLO-30, ~30 m posting, ~4 m LE90 vertical accuracy |
| SAR archive depth | Sentinel-1: from October 2014; ALOS-2: from 2014; ALOS-1 PALSAR L-band back to 2006 for historical baseline |
| Coherence limitation | C-band coherence typically lost within 6-12 days over vegetated wet peat in summer; L-band maintains coherence at 14-42 day intervals over most peat types |
| Deliverable formats | GeoTIFF velocity maps, corridor longitudinal profile CSV, flagged high-gradient segment shapefile, time-series displacement plots per scatterer |
| Processing latency | Initial corridor analysis from archive: 2-4 weeks depending on stack size and atmospheric correction method; ongoing monitoring updates within 1-2 weeks of new acquisition |
Analytics Satellize can run
| Annual vertical velocity map along cable corridor | SBAS InSAR time-series analysis on Sentinel-1 IW or ALOS-2 PALSAR-2 stack; atmospheric correction via ERA5 delay estimates | GeoTIFF raster and corridor-aligned longitudinal profile CSV, clipped to operator-defined buffer width |
| Differential settlement gradient map | Spatial gradient of vertical velocity field computed along route centreline; segments flagged where gradient exceeds configurable threshold (e.g. 5 mm/year per 100 m) | Shapefile of flagged high-risk segments with gradient magnitude attribute, ranked by severity |
| Seasonal displacement amplitude map | Sinusoidal seasonal component fitted to PS/SBAS displacement time series at each coherent scatterer; amplitude and phase extracted per point | GeoTIFF of peak-to-trough seasonal amplitude along corridor; useful for distinguishing reversible water-table-driven compression from irreversible consolidation |
| Coherence quality assessment | Mean coherence computed per interferogram pair across corridor; C-band versus L-band comparison where both datasets are available | Coherence report identifying corridor sections where PS density is too low for reliable rate estimation, with recommendation on corner reflector placement or sensor substitution |
| Historical deformation baseline | SBAS processing of ALOS-1 PALSAR archive (2006-2011) and ALOS-2 archive (2014-present) to establish pre-installation and post-installation deformation history | Multi-epoch velocity comparison report showing whether subsidence rates have changed since cable installation |
| Peat subsidence risk classification along route | Fusion of InSAR velocity with peat depth mapping (where available from national soil surveys), drainage ditch proximity from Copernicus DEM-derived flow accumulation, and land-use change detection from Sentinel-2 NDVI time series | Risk-tier GIS layer (low / moderate / high / critical) per 100 m route segment, suitable for maintenance prioritisation |
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.