Sensors
- ALOS-2 PALSAR-2: L-band (1.27 GHz) SAR with 3–10 m resolution in spotlight and stripmap modes. Long wavelength penetrates the mangrove canopy and double-bounce scatters off trunks and prop roots, making it sensitive to woody biomass structure. Structural collapse or trunk snap reduces HV backscatter measurably even when foliage persists. Revisit at a given site is roughly 14 days for single-pass; shorter with right/left-look combinations.
- Sentinel-1 C-band SAR: C-band (5.405 GHz) in IW mode at 10 m ground range resolution, 6-day revisit over most tropical latitudes with both satellites active. Less canopy-penetrating than L-band, but captures surface roughness change, flooding under canopy, and wind-row damage patterns. Freely available from Copernicus; rapid post-event acquisitions are routinely tasked.
- Sentinel-2 MSI: 10 m visible and near-infrared bands, 20 m red-edge bands (B5, B6, B7), 5-day revisit with both satellites. NDVI and red-edge chlorophyll indices detect canopy stress and defoliation once cloud clears. Cloud cover over post-cyclone scenes can delay usable optical acquisition by days to weeks, which is the primary operational limit.
- Landsat 8/9 OLI: 30 m multispectral, 16-day revisit per satellite (8-day combined). Longer archive than Sentinel-2 (Landsat 5 data from 1984 onward) supports pre-storm baseline construction over multiple seasons, reducing the risk of conflating cyclone damage with seasonal phenology.
Why mangroves fail quietly and flood exposure rises sharply
Mangrove forests are the first line of coastal defence against storm surge. Their pneumatophores, prop roots and dense stem networks dissipate wave energy and reduce surge height. Published field studies have documented wave height reductions of 50–70% across 500 m of intact mangrove fringe, though the exact figure depends heavily on tree height, density and inundation depth. When a cyclone snaps or uproots that fringe, the hydraulic resistance drops and the next surge event reaches further inland at greater velocity.
The mapping problem is that optical satellites often see green. A fallen or snapped mangrove trunk can retain foliage for two to four weeks post-storm, so NDVI remains elevated and the canopy looks intact from space. Structural damage is hidden inside a green signal. This is precisely the gap that L-band SAR fills.
What a floating roof gives away: the physics of L-band backscatter in mangroves
At 1.27 GHz, PALSAR-2's wavelength is roughly 23 cm, long enough to pass through leaf layers and interact with woody stems and flooded soil. In an intact mangrove, the dominant scattering mechanism in HV cross-polarisation is volume scatter from branches and trunks. In HH, double-bounce between trunk and waterlogged soil is strong. When a storm snaps trunks or flattens the canopy layer onto the ground, both mechanisms change: HV backscatter drops as the volume scatter source is removed, and HH double-bounce collapses because the vertical trunk element is gone.
The practical detection threshold in published PALSAR studies is roughly a 2–3 dB change in HV backscatter for moderate structural damage, though this varies with moisture conditions at acquisition time. Wet soil after a cyclone raises background backscatter and can partially mask the signal, so multi-temporal averaging of pre-storm scenes reduces noise in the baseline. Sentinel-1 C-band adds a complementary view: it is less sensitive to woody structure but picks up surface flooding under the canopy as a specular return drop, which helps distinguish wind damage from inundation-driven dieback.
Combining SAR structure and optical colour to classify damage severity
The most defensible change-detection workflow uses SAR as the primary structural indicator and optical indices as a severity classifier once cloud permits acquisition. The sequence runs as follows. A pre-storm baseline is built from at least two to three PALSAR-2 scenes across different seasons to capture phenological variation. Post-storm HV backscatter is differenced against that baseline. Pixels exceeding the 2–3 dB threshold enter a candidate damage mask.
Sentinel-2 red-edge bands (B5 at 705 nm, B7 at 783 nm) are then used to compute the red-edge chlorophyll index, which is more sensitive to canopy stress than broadband NDVI and less saturated over dense tropical vegetation. Where optical data is available within two to three weeks of the storm, a joint classification assigns each pixel to one of three severity classes: structural collapse with defoliation, structural damage with foliage retained, or surface flooding without canopy loss. The third class is important because inundated-but-intact mangrove is recoverable; structurally collapsed forest is not, on any relevant planning timescale.
Honest limits: cloud cover over post-cyclone tropical coastlines routinely delays the first usable Sentinel-2 scene by one to three weeks. In that window, SAR is the only source. PALSAR-2's 14-day revisit means the post-storm acquisition may not arrive for up to two weeks unless emergency tasking is requested through JAXA. Sentinel-1, with its 6-day revisit, often provides the first post-storm SAR pass and can anchor the timeline even if its structural sensitivity is lower.
From canopy loss to coastal protection deficit: the translation step
A damage map answers where the forest was destroyed. A protection-deficit map answers which communities now face elevated surge exposure because of that destruction. The translation requires two additional inputs: a coastal digital elevation model and a population or infrastructure layer.
Published surge attenuation models parameterise mangrove drag coefficients by forest density and height, both of which can be estimated from pre-storm PALSAR-2 backscatter intensity and, where available, GEDI spaceborne lidar canopy height data. Removing the damaged fringe from the model and re-running the attenuation calculation gives a first-order estimate of how far a surge of given height would now penetrate compared to the pre-storm scenario. The output is not a certified hydraulic model; it is a ranked priority list of coastal segments where protection has degraded most severely. That is enough to direct field survey teams and inform replanting or temporary protection decisions before the next storm season.
Satellize ran a comparable multi-sensor crop-estimation workflow for the Kingdom of Tonga, a Pacific island state where cloud cover, small land area and infrequent revisit create the same data-scarcity problems that complicate post-cyclone coastal assessment across the Pacific basin.
Archive depth and the baseline problem in frequently disturbed coastlines
A single pre-storm scene is a poor baseline. Mangroves experience seasonal flooding, drought stress and incremental anthropogenic clearing that all shift backscatter and NDVI independently of storm damage. The Landsat archive, which extends to 1984, and the PALSAR/PALSAR-2 archive from 2006 onward, allow analysts to construct a multi-year phenological envelope for each pixel. Damage is then defined as a departure from that envelope rather than from a single prior scene, which substantially reduces false positives from seasonal variation.
For coastlines with no prior SAR coverage, the first post-storm acquisition effectively becomes the baseline for future events. Starting that archive now, before the next cyclone season, is the only way to have a defensible pre-storm reference when it is needed. This is not a theoretical concern: the western Pacific, Bay of Bengal and Caribbean all have mangrove-fringed coastlines where PALSAR-2 coverage is sparse outside of emergency acquisition windows.
Typical figures
| Primary SAR spatial resolution | ALOS-2 PALSAR-2: 3 m (spotlight) to 10 m (stripmap); Sentinel-1 IW: 10 m ground range |
| SAR revisit | PALSAR-2: ~14 days single-pass, shorter with emergency tasking; Sentinel-1: 6 days (dual satellite) |
| Optical spatial resolution | Sentinel-2: 10 m (VIS/NIR), 20 m (red-edge); Landsat 8/9: 30 m |
| Optical revisit | Sentinel-2: 5 days; Landsat 8/9 combined: ~8 days. Cloud cover is the operative constraint over post-cyclone tropics. |
| SAR frequency and polarisation | PALSAR-2: L-band 1.27 GHz, HH+HV; Sentinel-1: C-band 5.405 GHz, VV+VH |
| Minimum detectable structural damage | ~2–3 dB HV backscatter change (published PALSAR studies); finer thresholds require site-specific calibration |
| Optical spectral bands used | Sentinel-2 B4 (665 nm), B5 (705 nm), B7 (783 nm), B8 (842 nm); Landsat OLI B4–B5 |
| Archive depth | PALSAR/PALSAR-2: 2006–present; Sentinel-1: 2014–present; Landsat: 1984–present |
| Typical post-event product latency | SAR change map: 24–72 hours after acquisition. Optical severity layer: days to weeks depending on cloud clearance. |
| Delivery formats | GeoTIFF damage extent rasters, GeoJSON severity polygons, PDF assessment report, optional WMS feed |
Analytics Satellize can run
| SAR structural damage mask | Multi-temporal HV backscatter differencing against phenological baseline (PALSAR-2 and/or Sentinel-1) | GeoTIFF binary mask and continuous change-magnitude raster; delivered within 72 hours of post-event SAR acquisition |
| Canopy loss severity classification | Joint SAR/optical classification using red-edge chlorophyll index and backscatter change; three-class scheme (collapse, structural damage, inundation only) | GeoJSON polygon layer with severity attribute; area statistics by severity class in accompanying PDF report |
| Coastal protection deficit ranking | Damaged fringe removal from parameterised surge attenuation model; ranked coastal segments by change in estimated surge penetration distance | Prioritised coastal segment shapefile with attenuation-loss attribute; tabular summary for field survey planning |
| Pre-storm phenological baseline | Multi-year PALSAR-2 and Landsat time-series envelope construction; pixel-wise mean and standard deviation of backscatter and NDVI by month | Archive raster stack and baseline statistics layer; reusable for future event detection |
| Mangrove extent pre/post comparison | Object-based image analysis on Sentinel-2 composites cross-validated against SAR mask; comparison against Global Mangrove Watch reference layer | Area-change summary table and before/after extent polygons in GeoPackage format |
| Population exposure uplift estimate | Overlay of protection-deficit segments against WorldPop or national census gridded population; summation of population within newly exposed surge zones | Exposure table by administrative unit; input layer for humanitarian prioritisation briefings |
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.