Peatland drainage and subsidence monitoring
Sentinel-1 InSAR time-series can detect millimetre-scale surface subsidence over drained peatlands, while optical imagery maps the canal networks driving that collapse. Together they underpin credible REDD+ and voluntary carbon accounting.
Sensors
- Sentinel-1 SAR (C-band, InSAR): 5 m × 20 m IW-mode resolution; 6-day repeat at mid-latitudes, 12-day at equatorial sites. Phase differencing across repeat passes yields line-of-sight displacement measurements with theoretical precision of a few millimetres per epoch, though coherence over vegetated peat is frequently poor.
- ALOS-2 PALSAR-2 (L-band SAR): L-band (23.6 cm wavelength) penetrates vegetation canopy better than C-band, sustaining coherence over scrub and secondary regrowth that defeats Sentinel-1 InSAR. Stripmap mode delivers 3 m × 3 m resolution; 14-day revisit. Better suited to forested peat domes where C-band decorrelates rapidly.
- Sentinel-2 MSI: 10 m visible and near-infrared bands, 20 m red-edge and SWIR bands, 5-day revisit with two satellites. SWIR band 11 (1610 nm) and band 12 (2190 nm) are sensitive to surface moisture and bare peat exposure. Used to map canal extent, drainage-induced vegetation stress and land-cover transition.
- Planet SuperDove: 3 m resolution, daily revisit, 8 spectral bands including red-edge. Resolves individual canal widths down to roughly 6 m and detects early vegetation stress in the weeks after a new cut is made, well before Sentinel-2 captures a cloud-free pass in humid tropical regions.
Why peat subsides and why that matters for carbon accounting
Peatlands store carbon accumulated over millennia. In undrained condition the water table sits at or near the surface, suppressing aerobic decomposition. Drain the peat and two processes begin: consolidation, which is mechanical compression as buoyant support is removed, and oxidation, the microbial breakdown of organic matter that releases CO2 directly to the atmosphere. Both cause the surface to drop.
The subsidence rate is not trivial. Published field studies from Sumatra and Borneo record losses of 2 to 5 cm per year in the first decade after drainage, slowing as the most labile carbon is consumed but continuing for decades. At those rates, a peat dome drained for plantation agriculture can lose a metre of surface elevation within 20 years. Each centimetre of oxidised peat corresponds to a calculable CO2 flux, which is why subsidence measurement is increasingly used as a proxy for emissions in REDD+ methodologies and voluntary carbon market verification frameworks.
What a drainage canal gives away in radar and optical data
Drainage canals are the proximate cause of peatland degradation, and they are detectable from orbit. In Sentinel-2 and Planet SuperDove imagery, canals appear as linear dark features in SWIR and NIR bands, where open water absorbs strongly. Canal widths in active concessions typically range from 3 to 10 m. Planet's 3 m pixels resolve the narrower cuts; Sentinel-2 at 10 m will detect canals reliably only above roughly 15 m width unless spectral unmixing is applied.
SAR imagery adds a complementary signal. Open water in a canal produces specular reflection that returns very little energy to the satellite, creating a dark linear signature in Sentinel-1 backscatter. Vegetation alongside freshly drained canals shows measurable changes in backscatter as soil moisture drops and plant stress develops over weeks to months. Change detection between pre- and post-drainage SAR scenes can flag new incursion even under persistent cloud cover, which is the dominant obstacle to optical monitoring in equatorial peat regions.
InSAR over peat: what it measures and where it fails
Interferometric SAR compares the phase of radar returns across two or more passes to derive surface displacement with millimetre-scale sensitivity. Over bare or sparsely vegetated peat, this works well. Multi-temporal InSAR stacking methods such as SBAS or PS-InSAR, applied to Sentinel-1 time-series, have been used in published studies to map subsidence bowls around drainage canals in Southeast Asian peatlands, resolving differential settlement at rates of a few millimetres per year.
The hard limit is temporal decorrelation. Dense tropical vegetation scatters radar energy from a constantly changing canopy. Between two Sentinel-1 passes separated by 12 days, the phase relationship over closed-canopy forest is largely random noise rather than a coherent signal. Coherence values below 0.3 are common over vegetated peat, and below that threshold displacement estimates are unreliable. L-band PALSAR-2 decorrelates more slowly because longer wavelengths interact with larger structural elements in the canopy, but it is not immune. Practically, InSAR is most useful over degraded, partially cleared or bare peat surfaces, and at the edges of drainage networks where vegetation is stressed and thinning. Expecting millimetre precision across intact forested peat domes is unrealistic with current spaceborne systems.
Combining subsidence and canal maps for emissions estimation
The analytical workflow links three layers. First, canal mapping from optical change detection establishes where drainage has occurred and when. Second, InSAR subsidence measurements over coherent pixels provide a localised rate of surface lowering. Third, published peat bulk density and carbon content values (typically 0.08 to 0.12 g/cm³ and 50 to 60 per cent carbon by dry mass for tropical peats) convert a subsidence rate into an approximate CO2 emission factor.
This is not a direct flux measurement. Subsidence integrates both mechanical consolidation and oxidative loss; separating the two requires ground-truth data that satellites cannot supply. Consolidation is largely a one-time event in the first year or two after drainage, while oxidation continues indefinitely. A monitoring programme that begins several years after initial drainage will see a signal dominated by oxidation, making the carbon conversion more defensible. Programmes that begin at the moment of canal construction need to account for the consolidation component explicitly, or they will overestimate CO2 emissions in early years.
Practical design choices for a monitoring programme
A credible peatland monitoring programme needs archive depth as well as ongoing acquisition. Sentinel-1 data runs back to 2014 for most tropical regions, and ALOS PALSAR (the predecessor to PALSAR-2) extends the L-band record to 2006. Starting an InSAR time-series from archive data rather than the current date gives a programme immediate historical context, which matters for baseline establishment under REDD+ accounting rules.
Cloud cover is the central constraint on optical monitoring. Equatorial peat regions in Indonesia and Malaysia can have fewer than 20 cloud-free Sentinel-2 observations per year at any given pixel. Planet's higher revisit rate increases the probability of a usable observation but does not eliminate the problem. A practical design pairs SAR-based canal detection (cloud-independent) with optical imagery used opportunistically for spectral characterisation and land-cover classification. Satellize applies this combined approach on open constellations and can add commercial tasking where archive gaps exist, as it does in its crop-estimation work for the Kingdom of Tonga.
Reporting cadence should match the carbon market's verification cycle. Annual subsidence maps with quarterly canal-change alerts are a reasonable baseline. Sub-annual InSAR stacks improve the ability to separate seasonal water-table fluctuation from permanent structural subsidence, which is a known source of ambiguity in single-epoch comparisons.
Typical figures
| InSAR spatial resolution (Sentinel-1 IW) | 5 m × 20 m (range × azimuth); typically multi-looked to 20–40 m for phase stability |
| InSAR spatial resolution (ALOS-2 PALSAR-2 Stripmap) | 3 m × 3 m; multi-looked to 10–20 m for time-series stacking |
| Subsidence detection sensitivity (coherent surfaces) | ~2–5 mm per epoch under good coherence; degraded or unreliable below coherence ~0.3 |
| Optical canal mapping resolution | ~15 m minimum canal width (Sentinel-2); ~6 m (Planet SuperDove at 3 m pixel) |
| Revisit (Sentinel-1, equatorial) | 12 days single-satellite; 6 days with both Sentinel-1A and 1B operational |
| Revisit (Planet SuperDove) | Daily (cloud-limited in tropical regions) |
| SAR archive depth | Sentinel-1: from 2014; ALOS PALSAR (L-band predecessor): from 2006 |
| Coherence threshold for usable InSAR | Typically >0.3 required; vegetated tropical peat frequently falls below this |
| Delivery formats | GeoTIFF displacement maps, GeoPackage canal vectors, CSV subsidence time-series, PDF verification report |
Analytics Satellize can run
| Canal incursion map | Supervised change detection on Sentinel-2 SWIR and Planet SuperDove NIR-red-edge composites; SAR backscatter change for cloud-affected periods | Polygon GIS layer showing canal network extent with date of first detection per segment |
| InSAR subsidence time-series | SBAS or PS-InSAR processing of Sentinel-1 IW SLC stack; L-band PALSAR-2 where C-band coherence is insufficient | Raster time-series of line-of-sight displacement with coherence mask; annual mean subsidence rate map |
| Coherence quality assessment | Per-pixel coherence calculation across all interferometric pairs; flagging of unreliable measurement zones over vegetated peat | Coherence map overlaid on subsidence output; written caveat layer in verification report |
| Approximate CO2 emission proxy | Subsidence rate converted to oxidative loss using published tropical peat bulk density and carbon fraction ranges; consolidation component flagged separately for early-drainage sites | Tabular emission estimate with uncertainty range; methodology note suitable for REDD+ or VCS submission |
| Drainage-induced vegetation stress index | Red-edge chlorophyll index and SWIR moisture index from Sentinel-2 and SuperDove; anomaly detection relative to pre-drainage baseline | Quarterly raster anomaly map; alert feed for new stress zones exceeding threshold |
| Multi-year land-cover transition matrix | Annual optical classification of peat surface condition (intact, degraded, bare, converted); change matrix compiled from archive | GIS layers per epoch; transition statistics table for baseline period and monitoring period |
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.