Mangrove tidal inundation frequency and hydroperiod mapping
Inundation frequency and hydroperiod duration control mangrove species zonation and canopy vigour more than any other variable. Multi-temporal Sentinel-1 SAR stacks, cross-validated with Sentinel-2 reflectance, resolve the tidal gradient within the fringe at scales relevant to management decisions.
Sensors
- Sentinel-1 SAR (C-band, ESA): 10 m spatial resolution in Interferometric Wide Swath mode, 6-day repeat at the equator with both satellites active. C-band backscatter penetrates the canopy partially and responds strongly to the dielectric contrast between standing water and dry substrate beneath the canopy, making it the primary tool for counting inundated observations across tidal cycles.
- Sentinel-2 MSI (ESA): 10 m (visible and near-infrared) to 20 m (red-edge, SWIR) resolution, 5-day revisit with both satellites. Provides canopy-level NDVI, red-edge chlorophyll indices and NDWI for vigour assessment and open-water delineation. Cloud cover over tropical coasts is a persistent constraint; cloud-free compositing over weeks to months is typically required.
- Landsat 8/9 OLI (USGS/NASA): 30 m resolution, 8-day combined revisit. Lower spatial detail than Sentinel, but the archive extends to 1972 across the full Landsat family, enabling multi-decadal hydroperiod trend analysis and change attribution that Sentinel's 2014-onwards record cannot support alone.
- ALOS-2 PALSAR-2 (JAXA): L-band SAR at 3–25 m depending on mode. L-band wavelength (23 cm) penetrates closed mangrove canopy more deeply than C-band, improving detection of inundation beneath dense upper-canopy cover where Sentinel-1 backscatter saturates or becomes ambiguous. Revisit is 14 days, and commercial tasking is available for priority sites.
Why hydroperiod, not just wet or dry
Most satellite inundation products answer a binary question: is this pixel flooded right now? For mangroves, that is the wrong question. Species such as Rhizophora and Avicennia occupy distinct positions along the tidal gradient because each tolerates a specific range of inundation frequency and duration. Rhizophora typically establishes at mid-intertidal elevations where inundation occurs on most tidal cycles; Avicennia and Sonneratia can persist higher in the frame, with inundation frequencies as low as 10 to 20 percent of tidal events. Confounding the two produces maps that look precise but carry no ecological meaning.
Hydroperiod mapping answers the continuous question: across a defined period, what fraction of observations show a pixel as inundated? A stack of 50 to 200 Sentinel-1 acquisitions spanning several months captures many tidal cycles and converts backscatter time series into an inundation frequency raster. That raster is a proxy for the tidal frame position of each mangrove zone, which in turn predicts species composition, root-zone anoxia stress, and vulnerability to sea-level rise. It is a fundamentally different product from a single-date flood map.
What a SAR backscatter stack actually measures beneath the canopy
C-band radar at Sentinel-1's frequency (5.405 GHz) interacts with mangrove canopy through a combination of direct canopy scattering, double-bounce between the trunk base and the water surface, and volume scattering within the canopy layer. When the substrate beneath the canopy is inundated, the double-bounce return is strongly enhanced because the specular water surface acts as a near-perfect reflector at low incidence angles. This produces a characteristic backscatter increase of several decibels relative to the dry-substrate state, and it is detectable even when the water surface is not directly visible from above.
The method is not without limits. Dense, tall mangrove canopy attenuates the C-band signal before it reaches the ground, reducing the double-bounce contrast and potentially causing inundation to go undetected beneath closed-canopy stands. Published studies using Sentinel-1 over mangroves in Southeast Asia and West Africa report detection accuracy in the range of 70 to 90 percent depending on canopy density and incidence angle, with lower accuracy in tall, closed-canopy Rhizophora forests. ALOS-2 PALSAR-2 in L-band partially resolves this by penetrating deeper into the canopy, though its less frequent revisit limits the number of tidal-cycle observations in a given period. Analysts should treat C-band hydroperiod maps as reliable for the seaward fringe and open-channel zones, and interpret closed-canopy interior results with more caution.
Building the frequency raster: stack size, tidal phasing and thresholds
The core workflow takes a time-ordered Sentinel-1 stack, applies a backscatter threshold (typically derived from a bimodal distribution analysis or change-detection algorithm such as the Otsu method) to classify each acquisition as inundated or not, then sums the inundated observations and divides by the total valid observations. The result is a per-pixel inundation frequency between zero and one.
Stack size matters considerably. A 30-image stack spanning three months may capture only neap-to-spring tidal variation inadequately if the 6-day revisit happens to alias the tidal cycle. A 100-image stack spanning 12 months is more reliable, and the Sentinel-1 archive from 2014 onwards provides sufficient depth for most tropical sites. Acquisition geometry should be consistent: mixing ascending and descending passes changes the incidence angle and therefore the backscatter response, introducing artefacts. For sites where tidal gauge data are available, acquisitions can be tagged with predicted tidal height at the time of overpass, enabling hydroperiod to be expressed as a function of tidal stage rather than just observation count, which is the more physically interpretable output.
Sentinel-2 as a canopy-vigour layer, not a replacement
Sentinel-2 MSI does not see through mangrove canopy to the water beneath. What it does provide is a canopy-level health signal. NDVI and the red-edge chlorophyll index (using bands at 705 nm and 740 nm) respond to chlorophyll content and canopy stress. Mangrove stands experiencing hypersaline conditions from restricted tidal flushing, or suffering from prolonged inundation stress, typically show suppressed NDVI and elevated red-edge stress indices before any structural mortality is visible in canopy height products.
The practical combination is to use the SAR-derived hydroperiod raster as the primary structural layer and overlay Sentinel-2 vigour composites to identify zones where observed inundation frequency diverges from expected vigour. A zone with low inundation frequency but anomalously low NDVI may indicate hypersaline stress from restricted tidal exchange, a common precursor to dieback. Cloud cover over tropical coasts means that a single-date Sentinel-2 image is rarely useful; median composites over 30 to 90 days are the standard approach, accepting that phenological variation is averaged out.
Honest limits and what they mean for decision-making
Three constraints deserve direct acknowledgement. First, the 10 m Sentinel-1 pixel is large relative to the width of many mangrove tidal creeks and the narrow fringe zones at the seaward edge. Sub-pixel mixing means that pixels at zone boundaries carry ambiguous inundation signals. Very High Resolution commercial SAR (such as ICEYE or Capella at 0.5 to 1 m) can sharpen the fringe boundary but lacks the revisit frequency needed for hydroperiod counting without a long multi-year campaign.
Second, the method assumes that backscatter change is driven by inundation rather than other causes, including rainfall on the canopy, wind roughening of open water, or seasonal phenological change in canopy structure. Quality control requires masking rain-affected acquisitions and, where possible, cross-referencing with tidal predictions. Third, sea-level rise projections and future hydroperiod modelling require coupling the satellite-derived current-state map with a digital elevation model of sufficient accuracy. Shuttle Radar Topography Mission data at 30 m is often too coarse for intertidal elevation gradients of less than 0.5 m; airborne LiDAR or ICESat-2 photon-counting data improve this significantly where available.
Satellize runs hydroperiod stacking workflows on Sentinel-1 open-archive data as part of its coastal analytics service, with Sentinel-2 vigour composites delivered as companion GIS layers. The Tonga crop-estimation programme demonstrated the organisation's approach to combining multi-sensor open-data stacks with site-specific validation, a methodology that transfers directly to mangrove monitoring contexts.
Typical figures
| Primary SAR resolution (Sentinel-1 IW mode) | 10 m range × 10 m azimuth (multi-looked product) |
| Sentinel-1 revisit at equator | 6 days (both satellites); 12 days with one satellite |
| Sentinel-2 optical resolution | 10 m (B2, B3, B4, B8); 20 m (B5, B6, B7, B8A, B11, B12) |
| Sentinel-2 revisit | 5 days combined; cloud cover typically requires 30–90 day compositing in tropics |
| ALOS-2 PALSAR-2 resolution | 3 m (spotlight) to 25 m (ScanSAR); L-band 1.27 GHz |
| Minimum detectable inundation frequency increment | Approximately 5–10% frequency steps, depending on stack size and canopy density |
| Sentinel-1 archive depth | 2014 to present (Sentinel-1A); Sentinel-1B data 2016–2021 |
| Landsat archive depth | 1972 to present across Landsat 1–9 family; 30 m resolution |
| Typical stack size for reliable hydroperiod | 50–200 acquisitions over 6–18 months recommended |
| Deliverable formats | GeoTIFF raster (inundation frequency 0–1), cloud-optimised GeoTIFF, GIS polygon zones by frequency class, PDF report |
Analytics Satellize can run
| Tidal hydroperiod frequency raster | Multi-temporal Sentinel-1 SAR backscatter thresholding (Otsu or bimodal classification) across 50–200 acquisition stack | GeoTIFF raster, 10 m, inundation frequency 0–1 per pixel, with uncertainty band layer |
| Mangrove species-zonation proxy map | Frequency class segmentation correlated with published species-specific inundation tolerance ranges from peer-reviewed mangrove ecology literature | GIS polygon layer with frequency-class labels and confidence flags; PDF interpretation note |
| Canopy vigour composite | Sentinel-2 median composite NDVI and red-edge chlorophyll index (B5/B8A ratio) over 30–90 day cloud-free window | GeoTIFF vigour raster and anomaly map relative to site-specific baseline period |
| Hydroperiod trend analysis | Annual hydroperiod rasters computed from Sentinel-1 stack per calendar year, change detection using Mann-Kendall trend test on per-pixel frequency time series | Multi-year trend GeoTIFF showing increasing, stable or decreasing inundation frequency per pixel; tabular summary by management zone |
| Tidal-stage-tagged inundation probability curve | Sentinel-1 acquisitions co-registered with tidal gauge or modelled tidal height at overpass time; logistic regression of inundation state against tidal stage per pixel | Per-pixel inundation probability as a function of tidal height; CSV and GIS layer for hydraulic model input |
| Stress-anomaly alert | Cross-comparison of SAR-derived hydroperiod with Sentinel-2 vigour index; flagging of pixels where vigour is suppressed relative to inundation-class expectation | Quarterly alert report with mapped stress zones and ranked priority areas for field verification |
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.