Ocean current and mesoscale eddy tracking from altimetry
Radar altimetry measures sea-surface height to within centimetres, revealing geostrophic currents and the mesoscale eddies that redistribute heat, carbon and marine life across ocean basins. Multi-mission merged products turn sparse along-track swaths into actionable gridded current fields.
Sensors
- Sentinel-6 Michael Freilich: Primary reference altimeter since 2020. Ku- and C-band dual-frequency radar eliminates ionospheric delay. Along-track sampling every ~300 m; cross-track repeat of 9.9156 days at 66° inclination. Sea-surface height accuracy better than 2.5 cm (1 Hz, open ocean).
- Jason-3: Operational since 2016, same ground track as TOPEX/Poseidon heritage. 9.9156-day exact repeat, Ku/C-band. Continues the 30-year climate-quality sea-level record and is merged with Sentinel-6 to increase effective spatial sampling.
- SARAL/AltiKa: Ka-band (35.75 GHz) altimeter, joint CNES/ISRO mission. Ka-band gives a narrower pulse footprint (~8 km vs ~18 km for Ku) and lower ionospheric noise, improving performance in coastal zones and for smaller eddies. 35-day repeat.
- SWOT KaRIn: Ka-band Radar Interferometer, launched December 2022. Produces swath altimetry 120 km wide (two 50 km swaths with a 20 km nadir gap), resolving features down to roughly 15–30 km wavelength, well below the ~100 km floor of conventional along-track products. 21-day repeat. Science data publicly available from 2023.
What a floating roof gives away
Sea-surface height is not flat. Warm-core anticyclonic eddies pile water into domes that can stand 30–80 cm above the surrounding ocean. Cold-core cyclonic eddies pull the surface down by a similar amount. These deformations are small enough to be invisible to the eye and large enough, at 50–500 km diameter, to dominate the kinetic energy of the upper ocean. A radar altimeter measures the two-way travel time of a microwave pulse to within nanoseconds, converts that to a range, and subtracts a precise orbit solution and a geoid model to isolate the dynamic topography signal.
From that sea-surface height anomaly (SSHA), geostrophic current velocity follows directly. On a rotating Earth, the Coriolis force balances the pressure gradient created by the height difference, so the surface current flows along contours of constant SSHA rather than down the slope. The calculation is well-established: u = -(g/f)(∂η/∂y) and v = (g/f)(∂η/∂x), where η is SSHA, g is gravitational acceleration and f is the Coriolis parameter. The physics is clean. The difficulty is spatial sampling.
Why one satellite is never enough
A single altimeter traces a one-dimensional curtain of measurements along its ground track. Adjacent tracks are separated by hundreds of kilometres at mid-latitudes on a 10-day repeat orbit. An eddy 150 km across may sit entirely between two tracks and go undetected. The solution, developed through the AVISO programme and now maintained by the Copernicus Marine Service (CMEMS), is to merge simultaneous missions onto a common grid.
The standard CMEMS merged product (SEALEVEL_GLO_PHY_L4_MY) currently combines Sentinel-6, Jason-3, SARAL/AltiKa and, where available, additional missions. The output is a daily, globally gridded SSHA field at 0.25° resolution (roughly 25–28 km at mid-latitudes). That grid spacing is adequate for eddies above about 100 km diameter but misses the submesoscale features that matter in coastal and polar margins. SWOT's wide-swath interferometry is designed to push the detectable scale down to 15–30 km, though its 21-day repeat means temporal gaps remain a real constraint for fast-moving features.
One limit worth stating plainly: the 0.25° merged grid cannot resolve eddies smaller than roughly 50–70 km, and even at that size the amplitude is underestimated. Users tracking features in the Agulhas Current retroflection or the Gulf Stream's frontal instabilities, where eddies are numerous and closely packed, should treat the gridded product as a statistical summary rather than an exact map of individual vortices.
Tracking a vortex from birth to decay
Mesoscale eddies are not momentary anomalies. The global eddy census published by Chelton et al. (2011) using 16 years of merged altimetry found that eddies with amplitudes above 1 cm and lifetimes above 4 weeks number in the hundreds of thousands annually, with median lifetimes of 32 days and some persisting beyond two years. The longest-lived travel thousands of kilometres westward under beta-drift before dissipating.
Automated detection and tracking algorithms, broadly divided into physical-parameter methods (closed SSHA contours, local extrema) and geometric methods (Okubo-Weiss parameter, winding-angle), are applied to each daily gridded field in sequence. A tracked eddy record carries its centre position, radius, amplitude, rotational speed and polarity through time. The CMEMS eddy-trajectory atlas (META3.2) provides this as a public dataset derived from the merged altimetry archive back to 1993, giving three decades of eddy statistics for any ocean basin.
Where the physics gets complicated
Geostrophic balance breaks down near the equator, where f approaches zero. Within roughly ±5° of latitude, the geostrophic approximation fails and altimetry-derived currents become unreliable. Coastal zones present a second problem: the altimeter footprint (~8–18 km depending on band) overlaps land, contaminating the return waveform. Dedicated retracking algorithms recover some signal within 5–20 km of the coast, but users should apply wider exclusion buffers for high-stakes applications.
The geoid is the third source of uncertainty. Dynamic topography is computed by subtracting a geoid model (typically EGM2008 or the GOCO series) from the measured surface. Geoid errors at short spatial scales can alias into apparent current structures. For relative anomaly work, this matters less because the mean sea surface is removed; for absolute velocity estimates it remains a limiting factor, particularly in data-sparse polar regions.
Finally, altimetry sees only the surface expression of a three-dimensional structure. A warm-core eddy may extend 500–1000 m deep, but the altimeter integrates steric effects across the water column without resolving the vertical profile. Combining SSHA with Argo float profiles or glider data is the standard approach for depth-resolved current estimates.
Operational uses beyond oceanography
Eddy tracking has direct commercial and strategic value. Fishing fleets target the productive frontal zones at eddy edges, where upwelling concentrates nutrients. Offshore operators routing supply vessels or planning cable and pipeline surveys need current forecasts accurate enough to affect fuel budgets and schedule risk. Defence and intelligence users care about mesoscale current structure because it affects acoustic propagation and submarine detection ranges.
Numerical ocean forecast models (NEMO, HYCOM, MOM6) assimilate the merged SSHA fields in near-real time to constrain their eddy-resolving runs. The quality of that assimilation determines how far ahead a useful current forecast extends, typically 5–10 days at mesoscale resolution before eddy position errors grow beyond a radius. Altimetry is the primary observational backbone for that assimilation cycle.
Satellize runs analytics on the CMEMS merged altimetry archive and near-real-time streams, producing eddy-trajectory overlays and geostrophic current grids calibrated to client-specified ocean regions. The workflow is the same one that underpins our Tonga crop-estimation programme in a different domain: open constellation data, rigorous physical processing, and outputs formatted for the client's planning cycle rather than a research journal.
What SWOT changes, and what it does not
SWOT's KaRIn interferometer is the most significant change to ocean altimetry since the TOPEX/Poseidon launch in 1992. Its 120 km swath produces two-dimensional SSHA maps at kilometric posting, making submesoscale features visible for the first time from orbit. Early science results confirm detection of features at 15–30 km scales that the conventional along-track network cannot see.
The constraint is revisit. At a 21-day repeat, SWOT cannot track a fast-moving eddy through its lifecycle the way the merged multi-mission product does on a daily cadence. The expected operational mode is to use SWOT snapshots to calibrate and correct the coarser daily merged fields, not to replace them. For users who need both spatial detail and temporal continuity, the two data streams are complementary, not interchangeable. That trade-off will persist until a constellation of SWOT-class instruments is in orbit, which no agency has yet funded.
Typical figures
| Spatial resolution (merged gridded product) | 0.25° (~25 km at mid-latitudes); minimum detectable eddy diameter ~50–70 km |
| Spatial resolution (SWOT KaRIn) | ~2 km posting; effective resolved wavelength ~15–30 km |
| Temporal resolution (merged L4 product) | Daily global grid; individual mission repeat 10–35 days depending on satellite |
| SWOT revisit | 21-day exact repeat |
| Sea-surface height accuracy (open ocean, 1 Hz) | ~2–3 cm RMS for Sentinel-6 and Jason-3; ~2 cm for SARAL/AltiKa (Ka-band advantage) |
| Radar frequency | Ku-band (~13.6 GHz) and C-band (~5.3 GHz) dual-frequency (Sentinel-6, Jason-3); Ka-band (35.75 GHz) for SARAL/AltiKa and SWOT |
| Latency (near-real-time product) | ~3–5 hours for NRT along-track; ~1 day for merged gridded NRT (CMEMS) |
| Archive depth | Merged altimetry record from 1993 (TOPEX/Poseidon era); CMEMS eddy-trajectory atlas (META3.2) covers 1993–present |
| Coastal exclusion zone | Standard products unreliable within ~20 km of coast; dedicated coastal retracking recovers ~5–20 km with reduced accuracy |
| Equatorial validity limit | Geostrophic approximation unreliable within approximately ±5° latitude |
Analytics Satellize can run
| Daily geostrophic current field | Geostrophic balance applied to CMEMS merged SSHA L4 product; standard u/v derivation from SSH gradients | NetCDF or GeoTIFF current-vector grid, client-defined bounding box, daily cadence |
| Eddy detection and classification | Closed-contour detection on SSHA field (physical-parameter method per Chelton et al. approach); polarity, amplitude and radius extracted per feature | GeoJSON point and polygon layer per daily snapshot, attributed with eddy type, amplitude and estimated radius |
| Eddy trajectory and lifetime report | Frame-to-frame nearest-neighbour tracking on detected eddy centroids across daily SSHA sequence; cross-referenced against CMEMS META3.2 atlas | Trajectory shapefile with lifecycle metadata (birth date, peak amplitude, displacement distance, decay date) and PDF summary report |
| Frontal zone delineation | SSHA gradient magnitude thresholding to identify sharp SSH boundaries associated with current fronts (e.g. Gulf Stream north wall, Agulhas front) | Weekly GIS polyline layer of principal frontal positions with historical envelope for anomaly assessment |
| Vessel routing current advisory | Geostrophic current vectors combined with eddy-centre positions to compute favourable and adverse current corridors along user-specified route segments | Route-specific current profile (speed and direction at waypoints), updated daily, delivered as JSON feed or PDF briefing |
| SWOT-enhanced submesoscale snapshot | SWOT KaRIn L3 SSHA ingestion and geostrophic derivation at ~2 km posting; merged with daily background field to extend spatial dynamic range | High-resolution SSHA and current GeoTIFF for target region, timestamped to SWOT overpass, with coverage mask indicating swath gaps |
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.