How we measure ocean heat — and why it matters more than air temperature

The dashboard's data, the choices behind it, and the limits of what it can tell you.

The question this tool answers

Of all the extra heat trapped on Earth by greenhouse gases, about nine-tenths has gone into the ocean. The rest has warmed the land, the air you feel day to day, and the ice. Air temperature is the number people quote, but it captures only the tip of the warming. The ocean's heat content — the total stored thermal energy from the surface down through roughly two kilometres — is the more honest accounting. This dashboard tracks it in two depth layers and four ocean basins, using the official US climate data record, and cross-references it against the leading independent groups.

The headline finding for 2025 — published by Lijing Pan and colleagues in Advances in Atmospheric Sciences on 6 January 2026 — is that the top two kilometres of ocean absorbed about 23 zettajoules of extra heat last year (one zettajoule equals a trillion gigajoules). That was the biggest single-year jump in nine years, and made 2025 the ninth consecutive annual record. To put that in human terms, Carbon Brief's coverage in January 2026 worked out that the ocean took up roughly 39 times more energy than every car, factory, power plant and household on the planet used across all of 2024 combined.

How we know

The numbers come from NOAA's National Centers for Environmental Information (NCEI) in Asheville, North Carolina. NCEI maintains the global ocean heat content record originally compiled by Sydney Levitus and his colleagues, and now curated by a team including Tim Boyer. The series reaches back to 1955 for the 0–700 metre layer, and adds a deeper 0–2000 metre series from about 2005 onward — the point at which measurement coverage of the deep ocean became dense enough to trust.

What does "dense enough" mean? Before 2005, the deep ocean was sampled mostly by research ships dropping temperature probes over the side — slow, expensive work, and especially sparse across the Southern Hemisphere. The modern picture comes from Argo, an international program operating roughly 3,800 robotic floats. Each float drifts with the currents at about 1,000 metres depth. Every ten days it sinks to 2,000 metres, then rises to the surface, measuring temperature and salinity the whole way up. At the surface it transmits its profile by satellite, then descends again. Together the array reports a snapshot of the world's upper-ocean heat every ten days.

Argo's first floats went into the water in 2000; continuous near-global coverage arrived in 2005. That is why the 0–2000 metre series in this dashboard begins then. For earlier decades the deeper ocean is interpolated from a sparser network — the values are plausible but more uncertain, and we say so on the chart.

The choices we made

Two layers, not one. Ocean heat can be reported for many depth slices: the surface, the top 100 metres, the top 700 metres, the top 2,000 metres, all the way to the seafloor. We show two because each tells part of the story. The 0–700 metre series holds most of the trapped heat and has the longer continuous record, so we lead with its cumulative total since 1955. The 0–2000 metre series captures the deeper warming Argo can now resolve — it is the figure behind the phrase "the deep ocean is warming too" — so we use it for the year-on-year line chart, which begins in 2005 where the data becomes dependable. The chart caption flags both choices.

Hiroshima bombs per second. The headline analogy traces to work by climate scientists including Kevin Trenberth and John Abraham, who noted around 2013 that Earth's energy imbalance — the surplus heat the planet gains each year — works out to roughly four to five Hiroshima-bomb energy releases every second, if you divide the global imbalance by the energy of the 1945 bombs (about 63 terajoules each, depending on the source). The figure is approximate: different assumptions about the imbalance give counts from three to six, and the long-term average is around five. For the exceptional 2025 increment specifically, the equivalent rate works out closer to twelve bombs per second. About 90% of that energy ends up in the ocean, which is why this dashboard exists.

We treat the Hiroshima-bombs framing as one analogy among several rather than the lead. It is visceral, but it can also confuse: if millions of nuclear-bomb-equivalents have already gone into the ocean and life is still going on, a reader can be left thinking the comparison is overblown rather than understanding that ocean heat surfaces over decades as floods, storms and reef death, not over seconds as a city-flattening shockwave. The hero numbers therefore lead with energy comparisons — 39 times more energy than human activity in 2024, or five times all the energy the United States has used since 1955 — and use the bombs figure as supporting context rather than the headline.

What this tool cannot tell you

It cannot tell you the sea surface temperature at a particular beach on a particular day — that is a different dataset (NOAA OISST) and a different question. For marine heat extremes at a specific reef, see our Marine Heatwave Tracker; for the sea level rise that ocean warming drives, the Sea Level Rise Monitor; for the bleaching it pushes reefs toward, the Coral Bleaching Monitor.

Year-to-year wiggles in the heat content number are real but small next to the multi-decade trend. A year with a smaller gain than the one before is not the warming "pausing" — it is heat redistributing between depth layers, which is why the integrated 0–2000 metre number is the most stable view. Other groups publish their own estimates from independent processing of the same float and ship data: China's Institute of Atmospheric Physics (the Cheng group), Japan's Meteorological Research Institute (the Ishii series), Australia's CSIRO, the UK's Met Office Hadley Centre, and the European Union's Copernicus Marine Service. They agree on the long-term trend to within roughly 10%, and the spread between them is what we use to draw the uncertainty band on the chart. We follow NOAA NCEI for the main series because it is the official US data record and the best-suited to our daily fetch, and cite the others when journalists do.

What's coming next

The Deep Argo program — a smaller fleet of floats built to descend to 6,000 metres — is rolling out through the 2020s. As its coverage grows, the abyssal ocean below 2,000 metres becomes measurable for the first time. Current estimates put perhaps 10% of the trapped heat in that layer, but the figure is uncertain by a factor of two. The next decade will sharpen it.

Credits

The ocean heat content record exists because of three generations of oceanographers — most notably Sydney Levitus, who began assembling it in the 1980s — together with the technicians and engineers who maintain Argo's 3,800 floats and replace them as they fail (each lasts about four years before its batteries die), the research-vessel crews who deploy and recover them, and the NCEI data team in Asheville who curate and publish the canonical series. This dashboard fetches from their public files. We are downstream of their work.

← Back to the dashboard