Sea-level anomaly mapping for tuna habitat depth prediction

Satellite altimetry measures sea-surface height anomalies that proxy thermocline depth, concentrating or dispersing skipjack, yellowfin and bigeye tuna across tropical oceans. This page explains the physics, the sensors, and the honest limits of the technique.

Sensors

Jason-3: Radar altimeter on a 10-day exact-repeat orbit at 1,336 km altitude. Along-track sea-surface height precision of approximately 2.5 cm. The primary operational reference for sea-level anomaly time series since 2016, continuing the TOPEX/Poseidon–Jason lineage.
Sentinel-6 Michael Freilich: ESA/EUMETSAT altimeter launched November 2020, flying 30 seconds behind Jason-3 on the same ground track to enable cross-calibration. Carries a Ku/C dual-frequency synthetic-aperture radar altimeter (SRAL) achieving along-track SSH precision better than 2 cm. Provides the continuity record now central to operational oceanography.
SARAL/AltiKa: CNES/ISRO Ka-band altimeter (35.75 GHz). The higher frequency reduces ionospheric noise and gives finer along-track resolution than Ku-band instruments, useful for resolving mesoscale eddies at scales of 50–100 km. 35-day repeat cycle. Drifting orbit phase since 2016 improves spatial sampling at the cost of temporal regularity.
CNES-CMEMS DUACS merged product: Not a single satellite but a multi-mission gridded product produced by CNES and distributed via Copernicus Marine Service. Combines Jason-3, Sentinel-6, SARAL and others onto a 0.25-degree global grid at daily to weekly resolution. This is the operationally practical input for fisheries applications: the merged anomaly field is more spatially complete than any single track.

What a sea-level anomaly actually tells a fisheries analyst

Sea-surface height (SSH) measured by a radar altimeter encodes the three-dimensional structure of the water column below. Where warm water accumulates, it expands and raises the surface. A positive sea-level anomaly (SLA) of, say, +15 cm typically indicates a warm-core structure in which the 20 °C isotherm, a standard proxy for the thermocline base, sits 30–60 m deeper than the regional mean. That extra depth gives tuna, particularly yellowfin and bigeye, more vertical room to forage without thermal stress.

Negative anomalies work the opposite way. A cold eddy or upwelling dome depresses the 20 °C isotherm toward the surface, sometimes to within 40–60 m of it. Skipjack, which tolerate cooler water less well than bigeye, are effectively squeezed into a thin surface layer and become easier to locate. This habitat compression is the mechanism behind the well-documented correlation between negative SLA and high skipjack catch rates in the western and central Pacific, a relationship formalised in published work using the TOPEX/Poseidon record and confirmed with Jason-series data.

The physics is solid; the calibration is regional

The SLA-to-thermocline relationship rests on the reduced-gravity approximation: in a two-layer ocean, the depth of the interface scales linearly with surface height anomaly, modified by the density contrast between layers. In the equatorial Pacific, where the thermocline is sharp and the density contrast is large, empirical regression coefficients are well constrained and the relationship is reliable at mesoscale (50–500 km) resolution.

Outside the central and western Pacific, caution is warranted. In the Indian Ocean, barrier layers, salinity-driven stratification that decouples the halocline from the thermocline, can break the simple two-layer assumption. In the Atlantic, the thermocline is generally shallower and more variable, and SLA explains less of the observed depth variance. Any operational system should be regionally calibrated against in situ Argo float profiles or historical CTD data before being presented to fishing fleets as actionable guidance. Presenting uncalibrated Pacific coefficients in the Indian Ocean would overstate thermocline depth by tens of metres in some conditions.

From altimeter track to usable fishing layer

Raw altimeter data arrives as along-track SSH measurements spaced roughly 5–7 km apart. The DUACS merging process interpolates these onto a regular 0.25-degree (approximately 28 km at the equator) daily grid using optimal interpolation. That grid spacing is adequate for resolving the large mesoscale eddies (100–500 km diameter) that dominate tuna habitat structure in tropical oceans, but it smooths out sub-mesoscale features at scales below about 50–70 km.

Latency matters operationally. Near-real-time DUACS products are typically available within 2–3 days of observation. The delayed-mode (higher-accuracy) product takes 6 weeks. For fleet guidance, the near-real-time version is the practical choice, accepting that SSH precision is slightly lower (around 3–4 cm RMS versus 2 cm in delayed mode). A 10-day Jason-3 repeat cycle means the merged product is updated continuously as successive tracks fill in, but any given 0.25-degree cell may not have a direct overpass for several days.

Derived thermocline depth estimates carry uncertainty of roughly ±10–20 m in well-calibrated Pacific regions, growing to ±30 m or more in poorly constrained basins. Analysts should communicate this uncertainty to skippers rather than presenting a single deterministic depth.

Combining SLA with complementary fields

SLA alone identifies habitat structure but not prey availability. The most useful operational products combine SLA-derived thermocline depth with sea-surface temperature (SST) and surface chlorophyll from ocean-colour sensors. The convergence zones around eddy edges, where SLA gradients are steep, often co-locate with fronts that aggregate prey. A positive SLA core may indicate deep habitat, but the highest catch probability tends to occur at the eddy periphery where upwelling nutrients support forage fish.

Satellite-derived mixed-layer depth products, available from CMEMS using Argo-calibrated climatologies, add another layer of context. When the mixed layer is shallow relative to thermocline depth, the foraging zone for surface-feeding tunas is well defined. When mixed-layer depth approaches thermocline depth, the habitat signal becomes ambiguous. These interactions are worth flagging explicitly in any advisory product rather than collapsing them into a single habitat-suitability score.

Operational delivery and honest limits for fleet managers

A practical SLA-based tuna habitat advisory covers a defined fishing zone, typically a national EEZ or a regional fishery management area, and is updated at least weekly. The output is a gridded map of estimated thermocline depth, an anomaly field relative to the climatological mean, and a qualitative habitat-compression index for each target species. Uncertainty bounds should accompany every depth estimate.

What the product cannot do is locate fish. It identifies where the physical environment is permissive, not where tuna are present on a given day. Skipjack in particular are highly mobile and respond to prey patches that no current satellite sensor resolves at the relevant scale. Fleet operators who understand this distinction use the product to prioritise search areas, not to replace sonar or logbook intelligence. Satellize builds this kind of advisory on the open DUACS and CMEMS data streams, applying regional calibration where Argo profile density supports it. The same analytical infrastructure underpins the environmental context work done for the Kingdom of Tonga crop-estimation programme, adapted here for oceanographic rather than terrestrial variables.

Data access is free at source via the Copernicus Marine Service. The cost is in calibration, integration and interpretation, not in the raw observations.

Archive depth and what it enables for stock assessment

The continuous satellite altimetry record runs from TOPEX/Poseidon in 1992 through to Sentinel-6 today: more than three decades of SSH observations on a consistent reference frame. That archive supports retrospective analysis of how thermocline depth variability, driven by ENSO cycles and the Pacific Decadal Oscillation, has shifted tuna distribution over multi-year periods. Years with strong positive SLA anomalies across the central Pacific (El Niño conditions) are associated with eastward displacement of skipjack biomass and reduced catch rates in western Pacific EEZs. The archive makes these patterns quantifiable rather than anecdotal.

For stock assessment bodies such as WCPFC or IOTC, the SLA time series provides a physically grounded environmental covariate that can be incorporated into habitat-based stock models. It is more mechanistically interpretable than SST alone because it speaks directly to the three-dimensional habitat volume rather than a surface condition. The limitation is that the 0.25-degree grid and the smoothing inherent in optimal interpolation mean fine-scale habitat heterogeneity is underrepresented. Models that treat the gridded product as ground truth will underestimate the patchiness that drives actual catch variability.

Typical figures

Along-track SSH precision (Jason-3 / Sentinel-6)	~2.5 cm / <2 cm (delayed mode); ~3–4 cm (near-real-time)
DUACS merged grid resolution	0.25° (~28 km at equator); mesoscale features ≥50 km resolved
Temporal repeat (Jason-3 / Sentinel-6)	10-day exact-repeat orbit; merged product updated daily
Near-real-time product latency	2–3 days from observation
Delayed-mode product latency	~6 weeks (higher accuracy, for retrospective analysis)
Altimeter frequency (Jason-3 / Sentinel-6)	Ku-band (13.6 GHz) primary; C-band (5.3 GHz) ionospheric correction
Altimeter frequency (SARAL/AltiKa)	Ka-band (35.75 GHz); finer along-track resolution, lower ionospheric noise
Thermocline depth uncertainty (calibrated Pacific)	±10–20 m typical; ±30 m or more in poorly constrained basins
Archive depth	1992–present (TOPEX/Poseidon through Jason series to Sentinel-6)
Coverage	Global ocean, ±66° latitude (Jason-3/Sentinel-6 inclination)

Analytics Satellize can run

Weekly thermocline depth anomaly map	Reduced-gravity two-layer regression of DUACS SLA onto thermocline depth, calibrated against regional Argo float climatology	GeoTIFF and NetCDF grid, EEZ or RFMO area, updated weekly
Habitat compression index by species	Species-specific thermal tolerance thresholds (skipjack, yellowfin, bigeye) applied to estimated thermocline depth to score vertical habitat volume	Weekly raster layer with per-cell habitat score and uncertainty band, delivered as GIS layer or PDF chart
Eddy identification and tracking	Automated detection of closed SSH contours in DUACS field using published Okubo-Weiss or winding-angle methods; eddy centre, radius, polarity and intensity tracked through time	Eddy track shapefile with daily position and intensity, 30-day rolling window
ENSO-phase habitat displacement report	Composite SLA anomaly fields stratified by ONI index phase, drawn from the 1992-to-present archive, to contextualise current conditions against historical El Niño and La Niña analogues	Quarterly PDF briefing with anomaly maps and catch-displacement context for regional fishery managers
Multi-variable habitat convergence layer	Co-registration of SLA-derived thermocline depth with SST fronts and surface chlorophyll from Sentinel-3 OLCI/SLSTR; eddy-edge convergence zones scored as high-priority search areas	Weekly GIS layer suitable for integration into vessel navigation systems or fleet coordination platforms
Retrospective habitat covariate series for stock assessment	Monthly mean thermocline depth anomaly time series extracted for defined stock assessment strata, formatted for input into habitat-based surplus production or age-structured models	CSV time series with metadata, covering the full 1993-to-present altimetry record

Sources

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.