Peatland water-table depth estimation by InSAR surface deformation
Peatlands rise and fall with their water table. Repeat-pass InSAR translates that surface motion into water-table depth estimates, giving hydrologists a spatially continuous record where dipwells are sparse.
Sensors
- Sentinel-1 A/B (C-band, 5.6 cm wavelength): Interferometric Wide Swath mode at 5 x 20 m resolution; 6-day repeat at mid-latitudes with both satellites operational, 12-day with one. Coherence degrades rapidly over vegetated bogs due to wind-driven canopy motion between passes. Best suited to bare or short-vegetation peat surfaces and PS-InSAR processing over longer time series.
- ALOS-2 PALSAR-2 (L-band, 23.6 cm wavelength): L-band penetrates low vegetation canopy, maintaining coherence over sedge and Sphagnum surfaces where C-band fails. Stripmap mode delivers 3 m resolution; ScanSAR delivers 100 m at wider swath. Nominal 14-day repeat, though off-nadir scheduling reduces effective revisit. The longer wavelength tolerates several centimetres of vegetation sway before decorrelating.
- TanDEM-X (X-band, 3.1 cm wavelength): Primarily a DEM mission, but bistatic and repeat-pass modes have been used for short-interval deformation studies. X-band coherence over vegetated peat is poor for intervals beyond a few days. Most useful for generating a high-quality baseline DEM against which longer-term Sentinel-1 displacement time series can be referenced.
- NISAR (L- and S-band, launch expected 2025): NASA-ISRO joint mission designed explicitly for deformation monitoring. L-band at 12-day repeat with global coverage, S-band adding a second frequency for coherence discrimination. Expected to be transformative for peatland InSAR globally once the archive matures, though no operational data exist at time of writing.
What a floating bog is actually doing underfoot
Peat is not solid ground. An intact raised bog can be 90 percent water by mass, and its surface elevation tracks the water table with a near-linear relationship. When the water table drops by 10 cm, the surface subsides by roughly 1 to 5 cm depending on peat type, bulk density and degree of humification. That ratio, sometimes called the elastic storage coefficient or the surface-to-water-table displacement ratio, has been measured in field campaigns across Fennoscandia, the UK and Southeast Asian tropical peatlands, and it typically falls between 0.1 and 0.5. The relationship is not perfectly linear at depth, and irreversible compaction occurs once peat dries below a threshold, but the elastic component is large enough to be detected from orbit.
This is the physical basis for InSAR-based water-table estimation. The satellite does not measure water directly. It measures the range change between sensor and ground surface across repeat passes, resolves that into a line-of-sight displacement, and the analyst converts displacement to water-table depth change using a site-specific or literature-derived coefficient. The method is indirect, and that indirection matters for how results should be used.
Coherence: the problem that determines everything
InSAR works by comparing the phase of radar backscatter between two passes. If the scattering geometry changes between passes, phase coherence is lost and the interferogram becomes noise. Vegetation is the enemy. Sphagnum moss, sedge tussocks and shrub heath all move with wind and grow between passes, scrambling the phase signal. C-band Sentinel-1 loses coherence over dense Sphagnum bogs within 6 to 12 days under typical UK or Scandinavian summer conditions. L-band ALOS-2 fares better because the longer wavelength is less sensitive to small canopy displacements, but it is not immune.
Three practical strategies exist. First, restrict analysis to winter acquisitions when vegetation is dormant and frost can temporarily stabilise the surface. Second, apply Persistent Scatterer InSAR (PS-InSAR) or Small Baseline Subset (SBAS) processing, which identifies pixels that remain coherent across many interferograms and builds a displacement time series only from those. In vegetated bogs, PS density is low unless artificial corner reflectors are installed. Third, deploy corner reflectors at key monitoring points. Trihedral corner reflectors of 60 to 90 cm aperture produce a radar cross-section that dominates surrounding vegetation returns, giving a reliable PS anchor. Several UK peatland restoration projects have done exactly this, combining a sparse reflector network with dipwell validation.
What the published record shows, and where it stops
Published studies using Sentinel-1 over drained agricultural peatlands in the Netherlands and Germany have detected subsidence rates of 10 to 50 mm per year attributable to water-table drawdown and oxidative peat loss, with spatial patterns matching drainage ditch networks at scales of tens of metres. L-band studies over tropical peatlands in Sumatra and Borneo have mapped drawdown following canal construction, with displacements of several centimetres detectable over months-long intervals. Rewetting responses after drain blocking in UK blanket bogs have also been captured, though the signal is smaller and noisier because rewetting raises the water table by less than drainage lowers it, and the surface rebound is partly masked by vegetation regrowth.
The honest limits are these. Absolute water-table depth cannot be recovered from InSAR alone; only changes relative to a reference epoch are measurable. The displacement-to-water-table conversion coefficient must come from field data or published analogue studies, and it varies by peat type. Atmospheric phase delay, particularly from water vapour gradients, introduces errors of 5 to 20 mm per interferogram that can be partially corrected using ERA5 reanalysis or GACOS tropospheric correction services but not eliminated. And over tropical peatlands with dense forest cover, even L-band coherence can be insufficient without a long time series and careful SBAS processing.
Turning displacement maps into hydrological products
A displacement time series is not a water-table map. The conversion pipeline has several steps. Line-of-sight displacement is decomposed into vertical motion using the satellite incidence angle, typically 30 to 45 degrees for Sentinel-1 IW mode, with the assumption that horizontal motion is negligible for peat compression. Vertical displacement is then divided by the site-specific displacement ratio to yield water-table depth change. That change is added to a reference water-table depth from dipwells or LiDAR-derived surface topography to produce an absolute estimate.
The resulting product is most useful as a spatially continuous complement to a sparse dipwell network, not as a replacement for it. For peatland restoration monitoring, the key deliverable is a map showing which areas are responding to drain blocking and which are not, at a spatial resolution that dipwells could never achieve economically. For carbon accounting, the water-table depth map feeds into emission factor look-up tables (IPCC Wetlands Supplement 2014 provides the relevant factors), converting a hydrological variable into an estimated CO2 and CH4 flux per unit area.
Practical configuration for a monitoring programme
A credible peatland InSAR programme needs at minimum: a consistent acquisition geometry (same orbital track, same ascending or descending pass throughout), a baseline DEM of better than 1 m vertical accuracy for topographic phase removal, at least 20 to 30 interferograms for SBAS processing to suppress atmospheric noise statistically, and a ground-truth network of dipwells covering the range of peat types and drainage conditions present on the site.
Sentinel-1 is the default choice for cost and revisit. ALOS-2 is worth adding for vegetated sites where C-band coherence is demonstrably poor. Corner reflectors at 500 m to 1 km spacing, logged against dipwells, anchor the displacement-to-water-table calibration. For large peatland complexes, a processing chain that ingests Sentinel-1 SLC data, applies GACOS atmospheric correction, runs SBAS inversion and outputs monthly displacement rasters in GeoTIFF is achievable with open-source tools including SNAP, StaMPS and MintPy. Satellize runs this pipeline for clients who need the analysis without the infrastructure, drawing on the same open-archive approach used in the Tonga crop-estimation programme.
One detail that is often overlooked: peat fires alter the surface dielectric properties and can produce apparent displacement signals unrelated to water-table change. Any site with a recent burn history needs optical burn-scar masking before InSAR results are interpreted.
Typical figures
| Spatial resolution (Sentinel-1 IW) | 5 x 20 m (single-look); typically multi-looked to 20-40 m for InSAR processing |
| Spatial resolution (ALOS-2 Stripmap) | 3 m single-look; 10-30 m after multi-looking |
| Revisit (Sentinel-1, one satellite) | 12 days; 6 days with both A and B operational |
| Revisit (ALOS-2) | 14 days nominal; effective revisit longer due to off-nadir scheduling |
| Radar frequency / wavelength | C-band 5.405 GHz / 5.6 cm (Sentinel-1); L-band 1.2 GHz / 23.6 cm (ALOS-2) |
| Minimum detectable displacement (single interferogram) | ~5 mm line-of-sight after atmospheric correction; ~1-2 mm with time-series averaging |
| Atmospheric phase error (uncorrected) | 5-20 mm per interferogram; reducible to ~3-8 mm with ERA5 or GACOS correction |
| Sentinel-1 archive depth | From April 2014 (Sentinel-1A launch); global coverage variable by region |
| Coherence threshold for reliable phase | Typically >0.3 required; vegetated bogs often fall below this at C-band in summer |
| Delivery formats | GeoTIFF displacement rasters, NetCDF time-series stacks, CSV dipwell-comparison tables |
Analytics Satellize can run
| Monthly surface displacement time series | SBAS InSAR inversion of Sentinel-1 SLC stack with GACOS atmospheric correction | GeoTIFF raster stack, one layer per epoch, with uncertainty band per pixel |
| Water-table depth change map | Displacement-to-water-table conversion using site-specific or literature elastic storage coefficient | GIS layer (GeoPackage) showing ΔWT in cm relative to reference epoch, with dipwell validation overlay |
| Rewetting response assessment | Before/after SBAS comparison across drain-blocking intervention date; spatial extent of surface rebound mapped | PDF technical report with maps, time series plots and statistical comparison to dipwell records |
| Persistent scatterer density and corner reflector placement advisory | PS candidate identification from amplitude dispersion index across multi-year Sentinel-1 archive | Map of coherent pixel density with recommended corner reflector locations and aperture specification |
| Annual peat subsidence rate map | Linear trend fitting to multi-year SBAS displacement time series; drained versus intact peat comparison | GeoTIFF of mm/year subsidence rate with 95% confidence interval per pixel |
| Carbon flux proxy estimate | Water-table depth map ingested into IPCC Wetlands Supplement 2014 emission factor look-up; area-weighted flux summation | Tabular report of estimated CO2-equivalent flux by management zone, with stated uncertainty from both InSAR and emission factor sources |
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.