Sea-surface temperature retrieval for oceanography and fisheries
Satellite SST combines thermal infrared precision with microwave all-weather coverage to track ocean heat at daily global scales. The physics of each method sets hard limits on what you can know, and where.
Sensors
- AVHRR on Metop-A/B/C: Five spectral channels including 3.7 µm, 10.8 µm and 12.0 µm thermal bands. Nominal 1.1 km nadir resolution, global daily coverage from polar orbit. The workhorse of operational SST since 1978; EUMETSAT Metop series extends that archive into the 2030s. Cloud-contaminated pixels must be masked, leaving gaps in persistently overcast regions.
- VIIRS on Suomi-NPP and NOAA-20: Day-Night Band plus thermal infrared channels at 375 m and 750 m resolution. Improved radiometric calibration over AVHRR gives SST uncertainty approaching 0.3 K under clear-sky conditions. Twice-daily polar coverage. Cloud masking remains the binding constraint.
- SLSTR on Sentinel-3A/B: Dual-view (nadir and 55° oblique) design at 1 km resolution in thermal channels (10.85 µm and 12.0 µm). The dual view allows better atmospheric correction and aerosol screening. Revisit of roughly one day at mid-latitudes with two satellites. ESA publishes Level-2 SST products operationally through the Copernicus Marine Service.
- AMSR2 on GCOM-W1: Passive microwave radiometer operating at 6.9 GHz to 89 GHz. The 6.9 GHz channel samples sea-surface emission through most non-precipitating cloud, giving SST at roughly 40–50 km resolution. Revisit approximately once per day. Fills the Southern Ocean and persistent-stratus gaps that defeat IR sensors entirely, but cannot resolve mesoscale eddies smaller than about 100 km.
What the sensor actually measures, and why it matters
Every satellite SST figure carries a depth label, whether or not it is stated. Thermal infrared sensors operating in the 10–12 µm atmospheric window respond to emission from the top 10–20 µm of the water column. That layer, the skin, is typically 0.1–0.5 K cooler than the water a millimetre below due to evaporative cooling and molecular conduction. Passive microwave radiometers at frequencies around 6–7 GHz sample a depth closer to 1 mm, producing what is conventionally called the sub-skin temperature. Neither is the same as the bulk mixed-layer temperature that a ship's intake thermometer records at one to several metres depth.
For most oceanographic and fisheries applications, the distinction is operationally significant only under strong diurnal heating conditions, when the skin can diverge from the bulk by 1–3 K in calm, sunny weather. Algorithms that blend IR and microwave products into a foundation SST, such as the GHRSST Level-4 analyses distributed through the Copernicus Marine Service, apply bias corrections and uncertainty flags to make the depth reference explicit. Buyers of SST data should always check which depth convention a product uses before feeding it into a model.
The cloud problem, and the microwave answer
Thermal infrared is exquisitely sensitive, but optically opaque cloud blocks it completely. In the Southern Ocean, the North Pacific storm track, and the persistent marine stratus decks off California and Namibia, cloud cover can exceed 80% of observations in any given month. An IR-only SST product in those regions is not just noisy; it is systematically biased toward clear-sky conditions, which tend to coincide with specific meteorological states.
Passive microwave fills that gap. At 6.9 GHz, the atmosphere is nearly transparent to non-precipitating cloud, and AMSR2 on JAXA's GCOM-W1 satellite has provided continuous global SST since 2012. The cost is resolution: the 6.9 GHz antenna footprint on GCOM-W1 is approximately 40 km × 70 km. Mesoscale eddies of 50–100 km diameter, which are ecologically important in upwelling systems, are blurred or lost entirely. Rain contamination at higher frequencies adds further uncertainty. Microwave SST is the right tool for basin-scale heat budgets and climate monitoring; it is the wrong tool for locating a 30 km upwelling plume.
Atmospheric correction: where the uncertainty lives
Converting top-of-atmosphere radiance to SST requires removing the signal contributed by water vapour, aerosols and ozone in the intervening atmosphere. The standard approach for IR sensors is the split-window algorithm, which exploits the differential absorption between two adjacent thermal channels (typically around 10.8 µm and 12.0 µm). The technique is well-established, but its accuracy degrades in the presence of Saharan dust, volcanic aerosol or very high water-vapour columns, where residual errors of 0.5–1.0 K are plausible.
SLSTR's dual-view geometry provides an independent handle on atmospheric path length, improving aerosol correction relative to single-view instruments. VIIRS benefits from improved detector stability and on-orbit calibration compared with the legacy AVHRR design. Even so, the global target uncertainty for operational SST products is typically stated as 0.4 K (one standard deviation), and regional performance varies. Users running fisheries applications in aerosol-heavy coastal zones should apply independent validation against in-situ buoy data before treating satellite SST as ground truth.
Fisheries and oceanography: what SST actually predicts
SST is not a direct proxy for fish. It is a proxy for the physical conditions that concentrate prey. Cold upwelling water carries nutrients to the photic zone; the sharp boundary where cold upwells meet warmer surface water, the thermal front, aggregates zooplankton and the pelagic fish that feed on them. Tuna, billfish and small pelagics have well-documented thermal preference windows, and fishing fleets have used satellite SST to locate fronts since the 1980s.
The operational workflow typically involves identifying SST gradients above a threshold (often 0.5–1.0 K per 10 km), overlaying chlorophyll-a from ocean colour sensors, and applying species-specific habitat models. At 1 km resolution, VIIRS and SLSTR can resolve fronts relevant to coastal fisheries. For open-ocean tuna fleets operating across hundreds of kilometres, the coarser but cloud-free microwave product is often more useful day-to-day because it is available regardless of weather.
SST also drives coral bleaching prediction. The NOAA Coral Reef Watch programme uses a combination of VIIRS and AVHRR-derived products to compute Degree Heating Weeks, the accumulated thermal stress metric used by reef managers worldwide. A sustained anomaly of 1 °C above the climatological maximum for eight or more weeks is the canonical bleaching threshold, though local acclimatisation complicates that figure in practice.
Archive depth and the climate record
AVHRR data runs back to 1981, giving a continuous satellite SST record of more than four decades. The NOAA OISST (Optimum Interpolation SST) product blends AVHRR with in-situ observations and, more recently, VIIRS to produce a daily global 0.25° gridded analysis extending from 1981 to the present. That archive is the basis for most published studies of long-term ocean warming trends.
Intercalibrating instruments across that span is non-trivial. Sensor ageing, orbital drift changing the local equatorial crossing time, and changes in the in-situ network used for validation all introduce discontinuities. The GHRSST community maintains reanalysis efforts specifically to address these biases. Any trend analysis shorter than roughly a decade should be treated with caution unless the user has verified that the product version handles inter-sensor calibration explicitly.
What Satellize delivers from these sources
Satellize runs SST analytics on open constellation data, including Sentinel-3 SLSTR and VIIRS, as part of its broader ocean and fisheries intelligence service. The approach follows published GHRSST and Copernicus Marine Service methodologies, with client-specific masking, front-detection thresholds and species habitat overlays configured per engagement. The Tonga crop-estimation programme is the company's most public analytics reference, but the underlying pipeline architecture applies equally to ocean thermal products.
Practically, the value is not in the raw SST field, which any competent analyst can download from Copernicus. It is in the derived products: front probability maps updated daily, thermal habitat suitability indices for target species, bleaching stress accumulation alerts, and anomaly time series set against a client-defined baseline climatology. Those outputs are what fisheries regulators and fleet operators actually use.
Typical figures
| IR SST spatial resolution | 375 m–1.1 km (VIIRS 375 m, SLSTR 1 km, AVHRR 1.1 km at nadir) |
| Microwave SST spatial resolution | ~40–50 km effective (AMSR2 6.9 GHz channel footprint) |
| Revisit (IR sensors) | ~1 day global with two Sentinel-3 satellites; ~12 hours with Suomi-NPP + NOAA-20 VIIRS |
| Revisit (AMSR2) | ~1 day global |
| Key spectral channels | IR: 3.7 µm, 10.8 µm, 12.0 µm; Microwave: 6.9 GHz, 10.65 GHz, 18.7 GHz (AMSR2) |
| Skin depth sampled (IR) | 10–20 µm |
| Sub-skin depth sampled (microwave) | ~1 mm |
| Typical SST uncertainty (clear sky, IR) | 0.3–0.4 K (1 SD); degrades to 0.5–1.0 K in high-aerosol or high-humidity conditions |
| Archive depth | AVHRR from 1981; AMSR2 from 2012; Sentinel-3 SLSTR from 2016 |
| Delivery formats | NetCDF-4 (GHRSST convention), GeoTIFF, OGC WMS/WCS, JSON time-series API |
Analytics Satellize can run
| Daily SST composite (cloud-gap-filled) | Optimal interpolation blending IR and microwave observations, following GHRSST Level-4 methodology | GeoTIFF or NetCDF grid, updated daily, covering client-defined area of interest |
| Thermal front probability map | Gradient-based front detection (e.g. Cayula-Cornillon single-image edge detection) applied to VIIRS or SLSTR SST | Daily GIS layer showing front location and gradient magnitude; suitable for fleet routing decisions |
| Species thermal habitat suitability index | Lookup of SST against published species-specific preference windows (e.g. skipjack tuna 22–30 °C), combined with chlorophyll-a overlay | Weekly raster suitability map per target species, delivered as GeoTIFF with accompanying PDF summary |
| Coral bleaching stress accumulation alert | Degree Heating Week calculation following NOAA Coral Reef Watch published algorithm; threshold at 8 DHW triggers alert | Automated alert email or API push when DHW threshold is crossed at monitored reef locations |
| SST anomaly time series | Pixel-wise departure from client-defined climatological baseline (minimum 10-year reference period recommended) | Monthly report with anomaly maps and area-averaged time-series charts; CSV export for client modelling |
| Upwelling index and seasonal forecast support | SST-based upwelling detection using negative anomaly thresholds in known upwelling zones, cross-referenced with wind stress data where available | Seasonal upwelling intensity summary and zone delineation shapefile for fisheries management input |
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.