Environmental context mapping for fish-aggregating device deployments
Satellite oceanography reveals whether the water around a fish-aggregating device can actually support a tuna aggregation. SST, chlorophyll, mixed-layer depth and current fields from VIIRS, Sentinel-3 and Copernicus Marine Service data explain why some FADs fish and others do not.
Sensors
- VIIRS (Suomi-NPP / NOAA-20): Provides sea-surface temperature at roughly 750 m pixel resolution in clear conditions, with daily global coverage from two satellites. The NOAA ACSPO SST product has a documented accuracy of approximately 0.3–0.5 K against drifting buoys, which is sufficient to resolve the 0.5–1 °C gradients that mark productive thermal fronts. Cloud cover is the primary gap; multi-day compositing reduces but does not eliminate it.
- Sentinel-3 OLCI: Ocean and Land Colour Instrument on Sentinel-3A and -3B delivers ocean-colour imagery at 300 m resolution with a combined revisit of roughly 1–2 days at the equator. Chlorophyll-a concentration is retrieved via the OC4Me algorithm, with a published detection floor around 0.01 mg m⁻³ in Case-1 open-ocean water. Turbid coastal water and sun glint reduce accuracy; open tropical ocean is where OLCI performs best.
- Copernicus Marine Service (CMEMS) ocean-current reanalysis and forecast: The CMEMS Global Ocean Physics Analysis and Forecast product provides daily surface and sub-surface current fields at approximately 1/12° resolution (roughly 8–9 km at the equator). Geostrophic and Ekman components are combined. The product is not a direct satellite observation but assimilates altimeter, SST and Argo float data; forecast skill degrades beyond about five days.
- Sentinel-6 Michael Freilich altimeter: Measures sea-surface height anomaly along a repeating ground track with a 10-day exact-repeat cycle and centimetric precision. Sea-level anomaly fields derived from multi-mission altimetry (Sentinel-6 combined with Jason-3 and other missions via CMEMS) indicate mesoscale eddy positions, which are among the strongest predictors of sub-surface mixed-layer depth and tuna habitat compression.
- MODIS Aqua (heritage context): Still operationally useful for chlorophyll and SST compositing where Sentinel-3 has data gaps. 1 km SST and 1 km chlorophyll at daily revisit. Aqua is past its design life and should be treated as a supplementary rather than primary source for new programmes.
Why the water column matters more than the device itself
A fish-aggregating device does not produce fish. It provides a reference point in otherwise featureless open ocean, concentrating whatever tuna are already moving through the area. The critical variable is whether that area contains the prey, temperature and dissolved-oxygen conditions that keep skipjack and yellowfin within striking range of the surface. A FAD placed in oligotrophic water far from any front or eddy boundary will aggregate very little, regardless of its construction.
The oceanographic literature is consistent on the broad picture. Tuna aggregations around FADs tend to be denser near thermal fronts, at the edges of mesoscale eddies and in areas where mixed-layer shoaling concentrates prey. Chlorophyll-a concentration is an imperfect but practical proxy for prey availability at the base of the food web. None of these variables is visible from a vessel on the water. All of them are measurable from orbit.
Reading the thermal field: what VIIRS SST actually resolves
VIIRS on Suomi-NPP and NOAA-20 generates a daily SST swath at 750 m resolution. Two satellites together provide near-daily global coverage, though a single overpass covers a swath of about 3,000 km. The NOAA ACSPO product achieves approximately 0.3–0.5 K accuracy against independent drifting buoys in clear-sky conditions. That precision is enough to map the 0.5–1 °C gradients that define productive fronts in the tropical Pacific and Indian Ocean.
Cloud is the honest problem. In the Inter-Tropical Convergence Zone and during monsoon periods, persistent cloud cover can leave multi-day gaps in clear-sky SST. Blended products that merge microwave SST from instruments such as the GPM Microwave Imager with infrared VIIRS data partially address this, at the cost of spatial resolution dropping to roughly 9–25 km. For FAD deployment planning, a 3–5 day composite usually provides enough temporal currency; for explaining a sudden catch-rate collapse, the gap matters more.
Chlorophyll as a proxy for the food chain, not the fish
Sentinel-3 OLCI retrieves chlorophyll-a at 300 m resolution across the open ocean, with the OC4Me algorithm performing well in the clear Case-1 waters where most industrial FAD fisheries operate. The 1–2 day combined revisit of the two Sentinel-3 satellites means that a weekly composite can be produced with reasonable spatial completeness even in partly cloudy conditions.
Interpreting chlorophyll correctly requires care. A chlorophyll-a value above roughly 0.1–0.2 mg m⁻³ in the tropical Pacific generally indicates elevated productivity relative to the surrounding oligotrophic background. But chlorophyll measures phytoplankton, not micronekton or forage fish. The link to tuna catch rates is statistical rather than mechanistic, and it weakens in upwelling zones where very high chlorophyll can be associated with cold, hypoxic water that tuna actively avoid. The honest use of chlorophyll data is as a screening variable, not a deterministic predictor.
Currents, eddies and the geometry of aggregation
The CMEMS Global Ocean Physics product provides daily current fields at approximately 1/12° resolution. For FAD operators, the most operationally relevant output is not the current speed itself but the derived convergence zones and eddy boundaries. Anticyclonic eddies create downwelling at their centres and convergence at their edges; cyclonic eddies do the reverse. The edges of both eddy types concentrate floating material, including naturally drifting objects that attract tuna, and a FAD placed near such a boundary will drift more slowly and remain in productive water longer.
Sea-level anomaly from Sentinel-6 and companion altimeters provides the most direct observation of mesoscale eddy structure. A positive sea-level anomaly of 10–20 cm typically marks an anticyclonic eddy centre; the productive edge is at the anomaly gradient. At 1/12° resolution, eddies of 100–200 km diameter are well resolved. Smaller sub-mesoscale features are not, and this is a genuine limit of the publicly available current products.
Mixed-layer depth: the variable that ties everything together
Mixed-layer depth (MLD) is arguably the single most important environmental variable for explaining FAD catch-rate variability, and it is the one most often absent from operational fisheries dashboards. When the mixed layer is deep (greater than 80–100 m in some tropical regions), tuna can disperse vertically and are harder to catch at the surface. When the mixed layer shoals to 20–40 m, tuna are compressed toward the surface and aggregate more tightly around any floating object.
MLD is not directly observable from passive optical or thermal sensors. It is inferred from altimeter-derived sea-level anomaly (which correlates with thermocline depth via the reduced-gravity approximation) and from CMEMS reanalysis products that assimilate Argo float profiles. The uncertainty in these inferred MLD fields is real: the CMEMS product is a model output constrained by observations, not a direct measurement. Treating it as indicative rather than precise is the right posture. Satellize incorporates CMEMS MLD fields alongside SST and chlorophyll layers in its environmental context outputs, following the same open-data methodology applied to the Tonga crop-estimation programme.
For operators planning FAD deployments weeks in advance, the forecast skill of CMEMS current and MLD products is useful out to about five days and degrades significantly beyond that. For post-season analysis of why catch rates varied across a fleet, the reanalysis product is more appropriate than the forecast.
Putting it into practice: limits and realistic expectations
Satellite environmental context does not replace skipper knowledge, onboard sonar or the GPS buoy data that tracks where a FAD actually drifts. What it adds is spatial coverage no vessel can match and a historical record that allows systematic analysis of which oceanographic configurations produced the best aggregations across seasons and years.
The practical workflow is straightforward. Before a deployment campaign, SST composites and chlorophyll maps identify candidate zones where fronts and moderate productivity coincide. Current forecasts indicate likely drift trajectories. After the season, catch logs are overlaid on the reanalysis environmental fields to build a site-specific model of the conditions that correlate with high catch rates. That model is fleet-specific and fishery-specific; it cannot be transferred wholesale from the Pacific to the Indian Ocean without recalibration. Anyone offering a universal FAD optimisation algorithm should be asked to show the validation data.
Typical figures
| SST spatial resolution (VIIRS) | 750 m (ACSPO product); blended microwave-infrared products at 9–25 km |
| SST accuracy (VIIRS ACSPO) | Approximately 0.3–0.5 K against drifting buoys in clear sky |
| SST revisit | Daily (two satellites); cloud-free composite typically requires 3–7 days in ITCZ |
| Chlorophyll-a resolution (Sentinel-3 OLCI) | 300 m; detection floor approximately 0.01 mg m⁻³ in Case-1 ocean water |
| Chlorophyll revisit (Sentinel-3A + 3B combined) | 1–2 days at equator |
| Ocean current field resolution (CMEMS 1/12°) | Approximately 8–9 km at equator; daily analysis and 5-day forecast |
| Sea-level anomaly (Sentinel-6 / multi-mission) | Along-track centimetric precision; gridded product at approximately 1/4° via CMEMS |
| Mixed-layer depth source | CMEMS reanalysis (model + Argo assimilation); indicative, not direct measurement |
| Archive depth | VIIRS SST from 2012; Sentinel-3 OLCI from 2016; CMEMS reanalysis from 1993 |
| Delivery formats | NetCDF, GeoTIFF, GIS vector layers; API-accessible via CMEMS for current/MLD data |
Analytics Satellize can run
| Pre-deployment zone suitability map | Multi-variable scoring of SST gradient magnitude, chlorophyll-a anomaly relative to seasonal baseline, and CMEMS current convergence; published approach follows Zainuddin et al. and similar habitat-index literature | Weekly GeoTIFF or web map layer showing ranked deployment zones, with confidence flags for cloud-affected SST pixels |
| FAD drift trajectory forecast overlay | CMEMS surface current forecast integrated over 5-day horizon; particle-tracking approach consistent with published Lagrangian drift methods | Daily updated vector layer showing probable drift corridors from user-specified deployment coordinates |
| Thermal front detection and classification | Single-image gradient and histogram-based front detection applied to VIIRS SST composites; method class consistent with Cayula-Cornillon algorithm family | Polygon GIS layer of detected fronts with estimated temperature differential and front length, updated with each cloud-free composite |
| Mesoscale eddy boundary delineation | Sea-level anomaly gradient analysis from CMEMS altimetry-derived product; eddy-tracking consistent with Chelton et al. published methodology | Weekly polygon layer of eddy centres and estimated productive edge zones, with anomaly magnitude in centimetres |
| Post-season catch-environment correlation report | Retrospective overlay of operator catch logs against CMEMS reanalysis SST, chlorophyll and MLD fields; Pearson and Spearman correlation with environmental variables; logistic regression for binary high/low catch classification | PDF analytical report with statistical tables and maps identifying the environmental signatures associated with above-median catch rates for the specific fleet |
| Mixed-layer depth anomaly alert | CMEMS MLD product compared against 10-year climatological baseline; anomaly threshold trigger at user-defined departure (e.g. MLD shallower than 30th percentile for season) | Email or API alert with regional MLD anomaly map attached as GeoTIFF |
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.