Mangrove loss and recovery monitoring for coastal fisheries habitat
Mangrove loss degrades nursery habitat for penaeid shrimp, snapper and grouper before any fishing survey notices. SAR backscatter and multispectral indices tracked across Sentinel-1, ALOS-2 PALSAR-2 and Sentinel-2 time series reveal where canopy is thinning, where it has gone, and where regrowth is credible.
Sensors
- Sentinel-1 SAR (C-band, ESA): 10 m ground range detected resolution in IW mode; 6-day repeat at the equator from the two-satellite constellation. C-band backscatter responds to canopy surface roughness and flood inundation beneath the canopy, distinguishing flooded from non-flooded mangrove. Does not penetrate dense closed canopy to the soil surface.
- ALOS-2 PALSAR-2 (L-band, JAXA): L-band (1.27 GHz) penetrates closed mangrove canopy and double-bounce scatters from inundated prop-root systems, giving a direct structural signal. Stripmap mode reaches 3 m resolution; ScanSAR delivers 100 m at wider swath. Revisit is 14 days for a single satellite, limiting flood-pulse capture.
- Sentinel-2 MSI (ESA): 10 m visible and NIR bands; 20 m red-edge and SWIR bands; 5-day revisit from the two-satellite pair. NDVI and the mangrove discrimination index (MDI) use the red-edge and NIR contrast to separate mangrove from other woody coastal vegetation, though cloud cover in humid tropical zones routinely degrades monthly composites.
- Landsat 8/9 OLI (USGS/NASA): 30 m multispectral; 8-day combined revisit. The Landsat archive extends to 1972 for earlier missions, giving the longest consistent optical baseline for decadal change analysis. At 30 m, small creek-edge mangrove patches below roughly one pixel width are systematically under-counted.
Why mangrove extent is a fisheries number, not just an ecology one
Penaeid shrimp, mangrove snapper and several grouper species spend their juvenile stages inside mangrove root systems. The structural complexity of prop roots and pneumatophores reduces predation pressure; the organic leaf litter drives the detrital food web that sustains those juveniles. When that habitat shrinks, recruitment to offshore fisheries declines, typically with a lag of one to three years that makes the cause-and-effect relationship invisible to catch statistics alone.
Satellite time series make the connection visible. A consistent annual map of mangrove extent, combined with species-specific juvenile residency periods, gives fisheries managers a leading indicator of recruitment pressure years before it appears in landings data. This is not a theoretical relationship: the link between mangrove area and coastal fish and shrimp yields has been documented in peer-reviewed work across the Indo-Pacific, Caribbean and West Africa.
What a floating root system gives away to radar
Optical sensors see the top of the canopy. SAR sees more of the structure. When tidal water floods a mangrove stand, L-band radar from ALOS-2 PALSAR-2 produces a strong double-bounce return from the water surface and the vertical root or trunk, a signal that is largely absent in dry or non-inundated woody vegetation. Mapping the frequency and spatial extent of that double-bounce signal across a tidal cycle produces an inundation frequency layer, which is itself a proxy for habitat quality: mangrove species composition and root architecture vary predictably with inundation regime.
C-band Sentinel-1 is less penetrating. In dense closed canopy, the signal scatters from the upper canopy surface and saturates at moderate biomass levels, making it less reliable for biomass estimation but still useful for detecting clearance events, which produce an abrupt drop in backscatter, and for monitoring the early stages of regrowth, where canopy closure has not yet occurred. Combining C-band and L-band time series captures both the surface and structural signals.
The confusion problem at coarser resolutions
Mangrove is not the only dark, dense, salt-tolerant woody vegetation at the coast. Nipa palm, casuarina, saltmarsh scrub and secondary coastal forest can produce NDVI values that overlap with mangrove in the visible and NIR bands used by Landsat. At 30 m resolution, a pixel containing a mixed stand of mangrove and nipa palm cannot be reliably disaggregated by NDVI alone.
The mangrove discrimination index, which uses the red-edge band available on Sentinel-2 at 20 m, exploits the distinct chlorophyll absorption and cell-structure reflectance of mangrove leaves relative to other coastal species. This reduces but does not eliminate confusion, particularly at stand edges and in early regrowth phases where canopy is open and understory reflectance contaminates the signal. The honest position is that area estimates from optical sensors carry a commission error from non-mangrove vegetation that should be quantified per site rather than assumed away. SAR structural signals provide an independent check.
Reading regrowth honestly
Regrowth is harder to confirm than loss. A cleared mangrove site colonised by herbaceous vegetation will show rising NDVI within months. Genuine mangrove propagule establishment takes longer, and canopy closure at a density that restores nursery function takes longer still, typically several years under natural regeneration and dependent on hydrological connectivity and sediment regime. A time series that reports only NDVI recovery will overstate habitat restoration.
A more defensible approach combines NDVI trajectory with L-band double-bounce signal recovery, which only develops once woody stem and root structure is sufficient to produce the characteristic scattering geometry. Where both signals are recovering in phase, the probability of genuine mangrove re-establishment is higher. Where NDVI recovers but L-band double-bounce does not, the colonising vegetation is likely herbaceous or low scrub. This distinction matters for any restoration credit or blue-carbon accounting that a government or enterprise might attach to the monitoring programme.
Archive depth, cloud and the limits of the time series
The Landsat archive provides the longest consistent optical baseline, with usable data from Landsat 5 TM onwards (1984) and a continuous 30 m record through Landsat 7, 8 and 9. Sentinel-1 SAR data are available from 2014; Sentinel-2 from 2015; ALOS-2 PALSAR-2 from 2014. ALOS PALSAR (the predecessor) extends L-band coverage back to 2006. For sites in persistently cloudy tropical zones, annual optical composites from Landsat or Sentinel-2 may rest on only a handful of clear observations per year, and a single undetected cloud shadow can mimic a canopy-loss event. SAR is unaffected by cloud, which is why it anchors the change-detection layer in cloud-prone environments even when its spatial resolution is coarser.
Minimum detectable loss patch size is a practical constraint. At 10 m Sentinel-1 resolution, a clearing of roughly 0.01 ha is theoretically detectable, but speckle noise in single-look SAR raises the reliable detection threshold to patches of 0.05 ha or larger without multi-look averaging or spatial filtering. At Landsat 30 m, the practical threshold is closer to 0.1 ha. Sub-pixel mangrove fringing along narrow tidal creeks is systematically missed by all current open-constellation sensors.
From pixels to a fisheries management product
The analytic chain runs from annual or seasonal extent maps, through a change layer flagging loss, degradation and regrowth, to a habitat-quality index that weights area by inundation frequency and canopy continuity. That index can be spatially joined to known spawning and nursery zones for target species, producing a recruitment-habitat pressure score that updates with each new satellite pass.
Satellize runs this chain on open Sentinel and Landsat constellations, adding ALOS-2 PALSAR-2 tasking on client licence where L-band structural data are needed for a specific site. The Tonga crop-estimation programme demonstrated that the same time-series compositing and change-detection pipeline transfers across tropical island and coastal settings with limited ground-truth data. Fisheries agencies wanting to commission a baseline mangrove extent map with annual update cadence should start by defining the coastal boundary, the target species list and the acceptable spatial resolution, since those three choices determine which sensor combination is appropriate before any processing begins.
Typical figures
| Spatial resolution (SAR, operational) | 10 m (Sentinel-1 IW); 3–100 m (ALOS-2 PALSAR-2 depending on mode) |
| Spatial resolution (optical, operational) | 10–20 m (Sentinel-2 MSI); 30 m (Landsat 8/9 OLI) |
| Revisit cadence | 6 days SAR (Sentinel-1 two-satellite); 5 days optical (Sentinel-2 two-satellite); 8 days (Landsat 8+9 combined); 14 days (ALOS-2 PALSAR-2) |
| Spectral bands used | C-band 5.4 GHz (Sentinel-1); L-band 1.27 GHz (ALOS-2); Red, NIR, red-edge 740 nm, SWIR 1610 nm (Sentinel-2); Red, NIR, SWIR (Landsat OLI) |
| Minimum detectable clearance patch | ~0.05 ha (SAR after speckle filtering); ~0.1 ha (30 m optical) |
| Cloud impact | SAR unaffected; optical composites in humid tropics may have fewer than 10 clear observations per year per pixel |
| Archive depth | Optical back to 1984 (Landsat 5); SAR back to 2006 (ALOS PALSAR), 2014 (Sentinel-1, ALOS-2) |
| Delivery formats | GeoTIFF extent and change layers; vector polygons (GeoPackage or Shapefile); tabular area statistics; web map tile service |
| Update cadence (typical product) | Annual baseline with quarterly degradation alerts; seasonal composites in high-cloud environments |
Analytics Satellize can run
| Annual mangrove extent map | Supervised classification combining SAR backscatter intensity and optical spectral indices (NDVI, MDI) with reference training from global mangrove databases | GeoTIFF and vector polygon layer, area statistics by administrative zone |
| Change detection layer (loss, degradation, regrowth) | Bitemporal and time-series differencing of SAR backscatter and NDVI composites; threshold-based flagging validated against L-band double-bounce signal | Annual change GeoTIFF with loss/gain/stable classes; tabular summary report |
| Tidal inundation frequency map | Multi-temporal Sentinel-1 SAR stack analysed for flood frequency using water-surface backscatter thresholding across tidal phases | Inundation frequency raster (percentage of observations flooded); GIS layer for habitat-quality weighting |
| Regrowth verification index | Combined NDVI trajectory and L-band double-bounce signal recovery over three-to-five year post-clearance windows to distinguish genuine mangrove re-establishment from herbaceous colonisation | Site-level regrowth confidence score; time-series chart per monitoring polygon |
| Recruitment habitat pressure score | Spatial join of habitat-quality index (area weighted by inundation frequency and canopy continuity) to species-specific nursery zones; annual update | Scored nursery-zone GIS layer; annual fisheries management briefing note |
| Aquaculture encroachment alert | Optical and SAR change detection flagging conversion of mangrove to pond geometry using shape and backscatter signature of open water within former canopy extent | Quarterly alert report with coordinates and estimated area affected |
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.