How satellites derive sea surface temperature from infrared and microwave radiometers, why clouds and fog mask kelp coasts, and how scientists validate SST.
Every marine heatwave map you've ever seen — the angry red blobs crawling across ocean basins — is built from satellite sea surface temperature (SST). Every kelp forest's heat stress forecast leans on it. And yet no satellite has ever touched the ocean. SST from space is an act of remote inference: measuring the light the sea gives off, correcting for everything in between, and blending the result with ships, buoys and models. Understanding how it's made — and exactly where it fails — is essential for anyone reading heatwave bulletins or managing a coast.
Every object above absolute zero radiates. The ocean at typical surface temperatures emits its peak radiation in the thermal infrared, around 10–12 micrometres — invisible to eyes, but squarely in the band of satellite radiometers. The radiometer's job is to measure the precise intensity of that emission in a few narrow wavelength "windows," then invert Planck's physics to a temperature.
The workhorse instruments: the AVHRR series that has flown on polar orbiters since the late 1970s, providing the four-decade climate record, and newer sensors like MODIS, VIIRS and the Sea and Land Surface Temperature Radiometer on Europe's Sentinels. Each scans the Earth in swaths, building global SST maps daily to every few days.
Between the sea surface and the radiometer sits 100 km of atmosphere, and it is not transparent — it is selective:
The classic fix is the split-window technique: measure the same patch of ocean at two neighbouring infrared wavelengths. Water vapour absorbs the two bands differently, so the difference between the two measured temperatures encodes how much vapour correction is needed. Modern retrievals use multiple bands and sophisticated algorithms, but the split-window idea is the lineage's spine.
The infrared signal escapes only from the ocean's top ten micrometres or so — the "skin." Below that skin, the "bulk" temperature a buoy measures at half a metre can differ by several tenths of a degree. At night and in calm conditions the skin is cooler (evaporation and radiation chill the surface film); on calm sunny days a warm surface layer can form a degree or more above the bulk.
This is not error so much as definition — but it matters at the coast. A kelp forest experiences the bulk temperature of the water washing its fronds, not the skin. Blended products exist precisely to reconcile the two: satellite skin SST adjusted and merged with in-water buoy measurements and models.
There is a second way to take the ocean's temperature: passive microwave radiometry, at frequencies where clouds are largely transparent. Microwave sensors see SST through most cloud cover — a superpower for persistently cloudy coasts. The costs: coarser spatial resolution (footprints of tens of kilometres versus one kilometre for infrared), interference near land and sea ice, and sensitivity to surface roughness from wind.
For kelp scientists the trade-off is stark: microwave SST smears out the coastline itself — the exact zone where kelp lives. Infrared wins on resolution; microwave wins on persistence. Operational SST products blend both.
Kelp forests cluster on coasts notorious for marine fog and stratus — California in summer, Namibia, Peru, the cool-temperate upwelling coasts generally. These are exactly the coasts where infrared SST maps go blank for days at a time, and exactly where local temperature anomalies decide whether a forest survives a heatwave or not — the dynamics behind thermal refugia, the cool pockets where forests persist.
Worse, coastal SST is the most spatially variable on Earth: upwelling plumes, headland eddies and river plumes change temperature by degrees over a few hundred metres — finer than any satellite pixel. A 1-km infrared pixel averages all of it. Satellite SST is superb for basin-scale heatwaves — the events described in marine heatwaves 101 — and genuinely limited for forecasting the temperature on a particular reef.
Satellite SST is continuously validated against drifting buoys — thousands of them measuring bulk temperature in the upper metre — along with ship radiometers and the Argo fleet's surface values. These comparisons quantify biases (often a few tenths of a degree, varying by sensor and region) and feed bias corrections into the products.
The buoy network matters doubly because it's also our direct ocean thermometer: our guide to reading ocean buoy data shows how to pull live coastal temperatures yourself. For a specific kelp project, a handful of moored loggers at reef depth will always beat any satellite product — satellites set the context; loggers write the local truth.
For climate work, the question is not today's map but the forty-year trend. Ensuring consistency across a dozen different sensors on successive satellites — each with its own calibration quirks — is the quiet achievement of climate-data-record teams: careful inter-calibration, reference to stable buoy records, and reprocesses that propagate improvements back through time. The ensemble SST datasets used to define marine heatwave statistics trace their credibility to this unglamorous calibration labour.
Practical rules for reading SST where kelp grows:
Forecast skill is improving fast, as explored in our look at predicting marine heatwaves — many of those forecasts chain directly off satellite SST assimilation.
Against drifting buoys, modern infrared SST products are accurate to roughly a few tenths of a degree Celsius on average in clear conditions, with regional variation. Microwave products are comparably accurate but spatially coarser. Absolute accuracy is less important than consistency — the calibration record prioritises stable trends for climate use.
Those are clouds, especially the marine fog and low stratus that characterise upwelling coastlines. Infrared sensors cannot see through them. Gap-filled (blended) products interpolate using microwave data, models and neighbouring days — better for analysis, but the fill is inferred, not measured.
Not reliably alone. Nearshore waters vary over hundreds of metres — upwelling plumes, headland effects, river outflows — finer than satellite pixels and sometimes offshore-representative only. Satellite products set the regional picture; a pair of inexpensive temperature loggers at reef depth gives you the local record that matters.
The skin is the ocean's top micrometre-scale film, which satellites sense and which is often a few tenths of a degree cooler (at night) or occasionally warmer (calm sunny days) than the bulk water below. Buoys measure bulk. Blended products reconcile the two, and kelp lives in the bulk.
Continuous infrared coverage dates to the early 1980s with the AVHRR series — over four decades — extended and inter-calibrated by later sensors. Before that, SST came from ships, and the merged ship-plus-satellite records underpin heatwave climatologies stretching back further.
Marine heatwaves are weeks of anomalously hot ocean that can deforest entire coastlines. Here's the science behind them and what restoration can do.
Dr. Lena Fischer ·
From Sentinel-2 pixels to drone orthomosaics, here's how scientists detect, map and track kelp forest canopies from space — and where the limits are.
Dr. Mara Ellison ·
Seasonal ocean forecasts can now flag elevated marine heatwave risk months ahead — sometimes. Here's what skill exists, where it fails, and who uses it.
Dr. Lena Fischer ·