EVIDENCE FILE 11 / ATLANTIC OVERTURNING
Is the AMOC Weakening? How Scientists Measure the Atlantic
An ocean can change its circulation without revealing the change at the surface. What do the instruments measure, and how far can a trend take us?
All Evidence FilesTHE SHORT ANSWER
RAPID combines several measurements across the Atlantic near 26°N to estimate overturning transport. A 2025 paper found weakening over 2004–2023, while other analyses emphasize pauses, corrections, and sensitivity to the chosen interval. The observations establish a changing circulation during the measured record. Attributing that change and forecasting an abrupt collapse require further evidence and models.
THE EVIDENCE MAP / ATLANTIC OVERTURNING
From a transport measurement to a climate question
The observed record, the explanation for its trend, and the forecast of a tipping point require different tests.
- 01Measurement
RAPID estimates Atlantic transport
Measurements near 26°N combine boundary-current, interior-ocean, and wind-driven contributions into a record of overturning transport since 2004.
- 02Interpretation
The fitted interval shows weakening
A 2025 study reports a declining transport trend for 2004–2023. That is an estimate for a stated period, with a stated uncertainty, within a variable circulation.
- 03Competing explanation
Natural variability and corrections
Different intervals, component corrections, and natural variability alter the estimated trend. Separating those effects from externally forced change requires more than its slope.
- 04Open test
Extend and compare the ocean records
Continue RAPID and compare independent subpolar observations such as OSNAP with model responses. Coherent changes across records can sharpen attribution without inventing a shutdown date.
OBSERVATION / INFERENCE / LIMIT
Read the signal.
Across the basin
RAPID has sustained measurements near 26°N since 2004; it estimates transport through an Atlantic section.
1.0 Sv per decade
McCarthy et al. (2025) report weakening of 1.0 [0.4–1.6] Sv per decade for 2004–2023.
No shutdown date
A fitted trend over an observing interval does not identify a future tipping point or its timing.
INSIDE THE EVIDENCE
The chain of reasoning.
Each step has a different level of certainty. Follow the measured result into the explanation, then see where the trail ends.
The instrument is an array, not one speedometer
A photograph of the sea cannot measure the total flow through it. The RAPID system combines the Florida Current transport, the wind-driven surface contribution, and transport across the ocean interior. Moorings supply information about conditions through the water column; the combined estimate tracks northward upper-water flow and its deep return. This is a basin-section measurement assembled from components. Reading it correctly requires asking which component changed, how it was estimated, and whether the same calculation was used throughout the record.
RAPID observing system and public dataPut the weakening estimate beside its interval
McCarthy and colleagues reported a weakening of 1.0 Sverdrup per decade, with a stated range of 0.4–1.6, over 2004–2023. One Sverdrup is a million cubic metres of water per second. Their analysis examines how a persistent signal emerges from substantial variability and identifies useful information in deep-water and thermocline transports. The result belongs to that record and that method. Extending the fitted line to zero would introduce a future behavior the instruments have never observed, rather than measure a countdown to collapse.
McCarthy et al. (2025): Signal and NoiseA pause can sit inside a longer decline
A separate study examined the absence of a clear trend during 2011–2021 and how a revised Florida Current record reduced an earlier weakening estimate. It combined observations with models to investigate natural variability and external forcing. The useful comparison is methodological: which years, corrections, and components entered each calculation? A shorter flat interval does not erase every longer trend, while a longer downward fit does not prove that the circulation weakens steadily every year. Those questions must be tested against the same corrected observations.
Research: a pause since the early 2010sModels ask a different question from moorings
The IPCC's Sixth Assessment projects AMOC weakening during this century across the considered emissions scenarios, with medium confidence that an abrupt collapse will not occur before 2100. That confidence statement leaves uncertainty; it is not a guarantee. A 2025 experiment across 34 climate models found that Southern Ocean upwelling sustained a weakened Atlantic overturning under extreme forcing. Such experiments explore mechanisms and model responses. Their assumptions, representation of ocean processes, and definition of collapse must be examined before treating them as a forecast for a particular year.
Consequences begin before the flow reaches zero
The AMOC redistributes heat, salt, carbon, and nutrients. A change in its strength can therefore affect regional temperature and rainfall, sea level, and marine ecosystems. Those impacts have their own regional uncertainties and cannot be read straight from a single transport value. Europe also exchanges heat through the atmosphere, so an AMOC story requires more than an image of one warm current stopping. The practical question is how a changing circulation alters conditions where people and ecosystems live, even while substantial flow continues.
NOAA: AMOC State of the Science fact sheetThe open test is a longer, wider record
In its September 2025 update, the National Oceanography Centre reported a trend of −0.9 ± 0.7 Sv per decade over twenty years and emphasized sustained observation to distinguish natural variability from human influence. OSNAP adds a separate view of overturning, heat, and freshwater transport in the subpolar Atlantic. These records can test whether changes appear coherently across regions rather than only in a selected interval at one latitude. Keeping instruments in the ocean is part of the answer: the next observations can challenge both confident reassurance and confident alarm.
NOC: September 2025 RAPID updateTHE EASY MISREADING
AMOC weakening is not interchangeable with the Gulf Stream disappearing, and a trend measured across twenty years is not a reliable calendar date for collapse.
PRIMARY RESEARCH & MISSION RECORDS
Check the source.
These are the studies and mission records behind this explainer. The illustrated article and Field Note continue the investigation.
- RAPID: observing system and transport data
- McCarthy et al. (2025): Signal and Noise at 26°N
- Research: a pause in weakening since the early 2010s
- NOC: twenty years of measurements, September 2025
- IPCC AR6: Summary for Policymakers, AMOC assessment
- Baker et al. (2025): overturning under climate extremes
- NOAA: State of the Science fact sheet
- OSNAP: the subpolar Atlantic observing program
EXPLORE IN 3D / ATLAS H-01
Explore the ocean beneath the surface
Descend through sunlit water and deep habitats in the Ocean Atlas, then return to the Atlantic circulation record. The illustrated descent shows environments and energy sources; its water movement does not simulate the AMOC.
Explore Ocean Descent in 3DFOLLOW THE THREAD
Stay with the question.
The Atlantic carries warmth into colder latitudes and returns water through its depths. Reading the warning in that circulation means following the instruments, the competing analyses, and the regional consequences of a slowdown.
Watch the AMOC documentaryMORE EVIDENCE FILES
Keep looking.
EVIDENCE FILE 01 / EUROPA
How Do We Know Europa Has an Ocean?
Galileo's magnetic measurements point to a conducting layer beneath Europa's ice. A salty ocean is the leading explanation. Its chemistry and seafloor activity remain open questions.
Read the evidence
EVIDENCE FILE 02 / TITAN & PLUTO
What Is the 5.11-Micrometer Mark on Titan and Pluto?
The 5.11-micrometer absorption is measured on both worlds, but the Pluto feature is broader. A shared wavelength is a clue to investigate, not a chemical identification.
Read the evidence