Dugong and manatee critical-habitat mapping from thermal and water-quality composites
Satellite thermal, turbidity, and bathymetric data from Sentinel-3, MODIS, and ICESat-2 delineate the shallow warm-water seagrass zones that sirenians depend on, tracking seasonal habitat contraction without ever resolving an individual animal.
Sensors
- Sentinel-3 OLCI: 300 m spatial resolution, roughly daily global revisit at mid-latitudes. Delivers chlorophyll-a, total suspended matter, and water-leaving reflectance products via the Ocean Colour products suite. Primary source for turbidity and phytoplankton-bloom tracking in sirenian range states.
- MODIS Terra / Aqua: Sea-surface temperature at 1 km (daytime and night-time bands 31/32), chlorophyll-a at 1 km, daily revisit per platform. The 20-plus-year archive is the backbone for detecting interannual thermal anomalies such as cold-water events that have historically caused manatee cold-stress mortality in Florida.
- ICESat-2 ATL03 / ATL08: Photon-counting green lidar (532 nm) capable of bathymetric retrieval to roughly 20–30 m in clear water, with a 91-day repeat cycle and approximately 14 m along-track ground-track spacing. Provides shallow-water depth estimates that, combined with thermal and turbidity layers, constrain the physically accessible foraging zone to depths sirenians can reach (typically less than 10 m).
- Landsat 8 / 9 OLI: 30 m multispectral resolution, 16-day revisit per satellite (8-day combined). Coastal aerosol (Band 1, 443 nm) and blue/green bands support empirical turbidity retrieval in optically shallow water. Useful for mapping turbidity gradients at finer spatial scale than Sentinel-3, particularly around river plumes and dredging zones.
- MODIS / Aqua MODIS SST (night-time): Night-time SST composites at 1 km reduce solar-glint contamination and provide the clearest thermal signal for identifying sub-threshold temperature zones (below roughly 20°C for West Indian manatees, below 18°C for dugongs in some range studies). Compositing over 8-day or monthly windows fills cloud gaps at the cost of temporal precision.
What the water column reveals about a habitat, not an animal
Dugongs and manatees are not detectable as individual objects from any operational civilian satellite. The largest adults reach about three metres in length, well below the resolution floor of every thermal or water-colour sensor listed here. The analytical objective is therefore different from wildlife census work: it is to define, at seasonal and interannual timescales, which areas of shallow coastal water are physically capable of supporting a sirenian population.
Three observable proxies do most of the work. Sea-surface temperature sets a hard lower bound: dugongs in Australian and Indo-Pacific waters show strong avoidance of waters below roughly 18°C, and West Indian manatees suffer cold-stress syndrome and mortality below approximately 20°C, a threshold documented during Florida cold snaps. Turbidity limits light penetration and therefore seagrass photosynthesis; without seagrass, the forage base collapses. Bathymetry constrains the zone to depths where sirenians can surface to breathe comfortably, in practice shallower than about 10 m for most feeding behaviour. Combine all three and you have a habitat-suitability surface rather than a species distribution map.
Cold snaps, turbidity pulses, and the geometry of contraction
The most operationally useful product is not a static suitability map but a time series that captures contraction events. MODIS night-time SST composites, archived continuously since 2000, allow analysts to reconstruct every winter season in which surface temperatures dropped below species-specific thresholds across a defined coastal polygon. For Florida manatees this kind of retrospective analysis has been used by researchers to correlate cold-event extent with stranding records, giving managers a spatial picture of which power-plant warm-water refugia are load-bearing for the population.
Turbidity events are shorter in duration but can be equally severe. A cyclone, a dredging programme, or a flood plume can raise total suspended matter enough to extinguish seagrass beds within weeks. Sentinel-3 OLCI total-suspended-matter products at 300 m, available with roughly one-day latency through the Copernicus Marine Service, can track a turbidity pulse as it spreads and dissipates. The honest caveat: cloud cover is the dominant gap in tropical and subtropical coastal monitoring. A Category 4 cyclone, precisely the event most likely to generate a destructive turbidity pulse, also produces days of cloud that blind optical sensors entirely. Compositing over 8 to 16 days recovers spatial coverage but loses the peak signal.
ICESat-2 and the bathymetric constraint
Thermal and turbidity layers describe surface conditions. Bathymetry tells you whether the seafloor is actually within reach. ICESat-2's green photon-counting lidar penetrates the water column in clear conditions and has demonstrated bathymetric retrieval to 20 to 30 m depth in peer-reviewed studies using ATL03 photon-cloud data. The 91-day repeat cycle means it is not a monitoring tool in the sense that MODIS is; it is a structural dataset used to build the depth layer of a suitability model.
The practical workflow is to clip the thermal and turbidity suitability surfaces to the ICESat-2-derived shallow-water mask, typically the zero-to-ten-metre isobath in areas where sirenian foraging has been documented. This removes deep-water areas that would otherwise score as thermally suitable but are ecologically irrelevant. Where ICESat-2 coverage is sparse, existing nautical chart data or Landsat-derived bathymetry products (using the Stumpf ratio method on blue and green bands) can fill gaps, though with lower accuracy in turbid or optically complex water.
Resolution limits and what the output actually is
It is worth being direct about what this analysis cannot do. At 300 m to 1 km resolution, the minimum mapping unit is far larger than an individual animal, a feeding trail, or even a small seagrass patch. The output is a grid of habitat-suitability scores, typically at 300 m to 1 km pixels, classified into categories such as thermally suitable year-round, seasonally marginal, or unsuitable. Ground-truthing with acoustic telemetry data or aerial survey transects is necessary before any management decision rests on the satellite layer alone.
Chlorophyll-a concentration, one of the OLCI standard products, is a useful secondary indicator: elevated chlorophyll can signal algal blooms that compete with seagrass for light, but it can also simply reflect healthy phytoplankton in offshore water adjacent to a bay. Separating these cases requires local knowledge of the oceanographic regime. The satellite layer raises the right questions; it does not answer all of them.
Building a seasonal suitability calendar for a range-state government
The most practical deliverable for a conservation agency or a government managing a marine protected area is a monthly or quarterly suitability composite covering the species' national range. Using the MODIS archive back to 2000 and Sentinel-3 from 2016, it is possible to produce a climatological baseline showing which months and which zones are reliably suitable, which are marginal, and which are seasonally excluded by cold or turbid water. Departures from that baseline in near-real time flag anomalous conditions worth investigating.
Satellize has run comparable multi-sensor coastal compositing work in the Pacific, including the Tonga crop-estimation programme which combines Sentinel and Landsat time series for agricultural monitoring. The analytical pipeline transfers directly to coastal water-quality compositing, though the oceanographic retrieval algorithms differ from land-surface vegetation indices. A government commissioning this work should expect to supply or co-develop local validation data: stranding records, aerial survey transects, or telemetry fixes are the ground truth that calibrates the satellite suitability surface to actual animal behaviour in that specific coastal system.
Practical entry point for a conservation programme
A credible first deliverable is a 20-year thermal-anomaly frequency map for a defined coastal zone, showing how often each pixel falls below the relevant temperature threshold in each calendar month, derived from the MODIS archive. This costs nothing in data acquisition (MODIS products are freely available via NASA Earthdata) and establishes immediately which areas are climatically marginal and which are stable warm-water refugia worth prioritising for protection.
The second layer, added once the thermal baseline exists, is a turbidity climatology from Sentinel-3 OLCI, identifying areas chronically affected by river discharge, coastal development, or seasonal upwelling. Together these two layers, before any bathymetric refinement, already identify the highest-priority zones for seagrass survey and sirenian monitoring effort. ICESat-2 depth data then refines the spatial boundary. The full composite model is built incrementally, which means a programme with limited initial budget can produce useful outputs at each stage rather than waiting for a complete analysis.
Typical figures
| Thermal resolution (MODIS SST) | 1 km; night-time composites preferred to reduce solar contamination |
| Water-colour resolution (Sentinel-3 OLCI) | 300 m; full-resolution product (FR) available for coastal zones |
| Turbidity / reflectance resolution (Landsat 8/9) | 30 m; 16-day single-satellite revisit, 8-day combined |
| Bathymetric lidar resolution (ICESat-2) | ~14 m along-track ground-track spacing; 91-day repeat cycle; depth range 0–30 m in clear water |
| Revisit (MODIS Terra + Aqua combined) | Near-daily; 8-day and monthly composites standard for gap-filling |
| Archive depth | MODIS: 2000–present; Landsat: 1984–present; Sentinel-3: 2016–present; ICESat-2: 2018–present |
| Minimum mapping unit (habitat suitability) | 300 m to 1 km pixel; individual animals not detectable |
| Cloud-gap limitation | Tropical cloud cover can obscure 30–60% of monthly observations; compositing over 8–16 days standard |
| Key spectral bands | OLCI: 21 bands 400–1020 nm; MODIS bands 31/32 (10.78–12.27 µm) for SST; Landsat OLI Band 1 (443 nm) for turbidity |
| Delivery formats | GeoTIFF suitability grids, NetCDF time-series composites, GIS-ready shapefiles of suitability class boundaries |
Analytics Satellize can run
| Thermal-suitability frequency map | Pixel-wise threshold exceedance counting on MODIS 8-day SST composites over the full archive; species-specific temperature cutoffs applied | GeoTIFF raster showing percentage of months below threshold per pixel, per calendar month; 20-year climatology report |
| Turbidity climatology and anomaly alerts | Sentinel-3 OLCI total-suspended-matter product time-series compositing; z-score anomaly detection against seasonal baseline | Monthly turbidity composite layers; near-real-time alert when TSM exceeds baseline by defined threshold in a protected-area polygon |
| Shallow-water habitat mask | ICESat-2 ATL03 photon-cloud bathymetric retrieval combined with Landsat Stumpf ratio depth estimates; depth-class polygon generation | GIS polygon layer of zero-to-ten-metre isobath within species range; uncertainty band on depth estimates included |
| Integrated habitat-suitability composite | Multi-criteria overlay of thermal, turbidity, chlorophyll-a, and bathymetric layers; weighted scoring calibrated against published sirenian habitat-use studies | Quarterly suitability-class raster (suitable / marginal / unsuitable); change map relative to prior quarter |
| Cold-event impact assessment | MODIS night-time SST anomaly mapping during identified cold-snap periods; area calculation of pixels below threshold within protected-area boundaries | Event report with maps, affected-area statistics, and comparison to historical cold events in the archive |
| Seasonal suitability calendar | Monthly climatological composites from MODIS and Sentinel-3 stacked across all available years; median and interquartile range per pixel per month | Twelve-panel map series (one per calendar month) showing long-run average suitability; PDF atlas and GeoTIFF stack |
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.