Is the Atlantic overturning conveyor slowing?

—

—

—

What the Atlantic overturning conveyor actually does

Warm surface water flows north, releases its heat to the air over the North Atlantic, becomes cold and salty enough to sink, and returns south at depth. It is one limb of a global loop that touches every ocean basin.

World map showing the global thermohaline circulation as a loop of red warm-surface currents and blue cold-deep currents. The map labels three regions where surface water sinks to depth: the subpolar North Atlantic south of Greenland, the Norwegian and Labrador Seas, and the Southern Ocean around Antarctica.
The global conveyor. Red ribbons are warm surface currents; blue ribbons are cold deep currents. The labelled “deep water formation” sites — including the subpolar North Atlantic south of Greenland — are where the conveyor sinks.
Three-dimensional view of the North Atlantic showing named currents that make up the conveyor: the North Atlantic Current and Norwegian Atlantic Current carrying warm surface water north (red and orange arrows), the East Greenland Current and overflows through the Denmark Strait carrying cold deep water south (blue and yellow arrows), and the Labrador, Irminger and Iceland basins labelled as the basins where sinking happens.
Zoomed into the North Atlantic. The same loop, up close. Warm water (red) reaches the seas around Iceland and Norway, cools, sinks, and returns south as the cold East Greenland Current and the overflows through the Denmark Strait (yellow). When the overturning weakens, less heat reaches the surface here — which is exactly the pattern in the “cold blob” map below.

Global conveyor: “Thermohaline Circulation” diagram via Wikimedia Commons. North Atlantic close-up: ocean-circulation diagram of the North Atlantic and Nordic Seas via Wikimedia Commons.

Rest of the world is warming. This patch is different.

Sea-surface temperature trend · not a direct flow measurement

Temperatures in 2015 against 1951–1980 — and the one spot that stayed cool

Robinson-projection world map of global temperature anomalies for 2015 compared to a 1951-1980 baseline. The entire globe is warmer than baseline (orange to deep red) with one striking exception: a deep-blue patch in the subpolar North Atlantic south of Greenland, known as the cold blob, and a smaller cool ring around Antarctica.
The whole world is warmer. One spot isn’t. This is an illustration from one year, 2015, not current data. The blue lobe south of Greenland is the “cold blob” — the visual signature climate scientists most often associate with a slowing Atlantic overturning conveyor. The colour bar at the top-left of the map shows the anomaly scale in degrees Fahrenheit (−3.6°F to +3.6°F vs the 1951–1980 average).

Image: NASA Scientific Visualization Studio (2015 record-warm global temperatures), via Wikimedia Commons. Public domain.
Conveyor strength, last five years
—
—
How fast it's slowing
—
Sv per decade (RAPID annual means)
Cold blob, last 12 months
—
—
Second-opinion reading
—
—

How much water the conveyor moved, year by year

Each point is one year's average. Higher = more water moving north (a stronger conveyor). The dashed line is the early-2000s average. Gold markers flag landmark research papers.

    Also in the literature

      The longer record: North Atlantic sea temperature since 1856

      Sea-surface temperature index · not a direct flow measurement

      How warm the North Atlantic surface was each year, after removing its long-run straight-line warming. Up to 2022 the record is NOAA’s index from the Kaplan temperature analysis, which stopped in January 2023. From 2023 on, NOAA’s ERSSTv5 analysis, worked out the same way, carries the record forward.

      How weak could the Atlantic overturning conveyor get this century?

      Strength of the conveyor as a share of its first five measured years (April 2004 to March 2009), with two projections for 2100 under a middle emissions scenario. A schematic, not a forecast dial.

      The first marker is the RAPID array’s last five years against its first five. The two 2100 markers come from Portmann et al. 2026 (Science Advances): the average of the CMIP6 climate models is a 32% decline by 2100 under the middle scenario SSP2-4.5 (their 90% range is ± 37 points, so it spans most of the scale), and their best observation-constrained estimate is a 51 ± 8% decline. The IPCC’s 2021 assessment says the conveyor is very likely to weaken this century under every emissions scenario, with only low confidence in how much, and medium confidence that there will be no abrupt collapse before 2100.

      What a weaker Atlantic overturning conveyor means near you

      Pick a country to see how a weaker Atlantic overturning conveyor would reach it. The page never looks up where you are.

      Choose a country above to see the impact profile.

      What drives this

      • Ocean heat →Warming changes density contrasts and the heat carried north.
      • Polar ice →Freshwater from ice melt can cap dense-water formation in the North Atlantic.

      What this drives

      How we know this

      • Sources: Monthly overturning at 26°N from the RAPID-MOCHA-WBTS observing array, release v2024-2 of October 2026, published by the British Oceanographic Data Centre under the UK Open Government Licence; daily sea-surface temperature from NOAA (OISST) for the cold blob; the North Atlantic temperature index from NOAA’s Physical Sciences Laboratory (Kaplan analysis, to January 2023) and, from February 2023, from NOAA’s National Centers for Environmental Information (ERSSTv5); the published Overturning in the Subpolar North Atlantic Program mean; recent-paper metadata from OpenAlex.
      • Update cadence: The cold-blob temperatures and the recent-papers list refresh every night; NOAA adds a month to the ERSSTv5 index early each month, and we check it every night. The Kaplan index no longer changes. The RAPID series changes only with a new release, about once a year and one to two years behind the calendar; we check BODC’s catalogue every night and say on the page when a newer release exists.
      • Time coverage: Direct overturning at 26°N from April 2004 to November 2024; cold-blob temperatures from November 2023; the North Atlantic temperature index from 1856, Kaplan to January 2023 and ERSSTv5 after; landmark papers are annotated on the chart.
      • Methods: The headline compares the last five years of the RAPID record with its first five (April 2004 to March 2009); the trend is a straight line through the monthly values. RAPID revised its calculation in this release, so the values differ slightly from earlier ones. The cold blob is the sea-surface temperature of a box south of Greenland minus the ocean as a whole, in 12-month averages. The North Atlantic index is the basin average minus a straight-line trend fitted from 1856 to January 2023, NOAA’s recipe for the Kaplan index, which we apply to ERSSTv5 too. Over 2000–2022, when both exist, the two analyses’ yearly averages correlate at 0.92, with ERSSTv5 0.13°C warmer on average; the chart note gives the current figures.
      • Uncertainty: Twenty years of RAPID measurements cannot prove a collapse; the IPCC’s 2021 assessment has medium confidence that there will be no abrupt collapse before 2100. Papers disagree on tipping risk. The temperature fingerprint depends on the forcing and may be unreliable as a proxy on its own (Mackay et al. 2024; Terhaar et al. 2025).
      • Limitations: The Atlantic overturning conveyor is related to, but not the same as, the Gulf Stream. The "near you" panel shows the country you pick; the page never detects your location. Pre-2004 trends require paleo-climate proxies with wider error bars.
      • Primary data: RAPID-MOCHA-WBTS observing array · BODC release v2024-2 · NOAA OISST (CoastWatch ERDDAP) · NOAA PSL Kaplan-based index (ended January 2023) · NOAA NCEI ERSSTv5 North Atlantic index ·Intergovernmental Panel on Climate Change Sixth Assessment Report — Working Group I, Chapter 9 · How this page is built