Atlantic Overturning Circulation (AMOC): Measured Weakening, Contested Clock

STAGE: FORMING

IPCC AR6 WG1, Ch. 9 (2021): 21st-century AMOC decline is "very likely" under all emission scenarios, with only "medium confidence" that there will be no abrupt collapse before 2100. The Nordic Council of Ministers published a commissioned AMOC-tipping risk assessment (TemaNord 2026:504) on 13 March 2026 — an intergovernmental watch formally begun.

The symptoms

	symptom	the evidence	as of
◐	Directly measured weakening at the RAPID array (26.5°N, the only continuous trans-Atlantic monitor, running since 2004)partial	Decline of 1.0 [0.4–1.6] Sv per decade over 2004–2023, statistically significant at 95%, from a mean of 17.0 Sv. The authors state this rate is consistent with model projections and "not consistent with a collapse in the mid-21st century" (a mid-century collapse would need ~5 Sv/decade). Pattern is decline–recovery–renewed weakening; the trend is not expected to become statistically "unfamiliar" (signal-to-noise > 2) until the 2040s. After the Volkov et al. 2024 sensor adjustment, others quote −0.8 ± 0.7 Sv/decade. [source]	2025-02-26
◐	Century-scale slowing in proxy reconstructions and ocean 'fingerprints' (record longer than the moorings)partial	Rahmstorf & Caesar (preprint, discussion opened 21 Apr 2026) argue "the balance of multiple lines of evidence strongly supports a past and ongoing AMOC slowing," citing ~15% weakening since the mid-20th century and a meridionally consistent deep-transport decline across four mooring arrays (2.6 ± 0.7 Sv/decade at 26°N in the deep western branch, Xing et al. 2026). Terhaar et al. (2025) and Worthington et al. (2021) counter that no significant long-term trend can be robustly identified. This disagreement is genuine and unresolved. [source]	2026-04-21
◐	Early-warning-signal studies claiming an approach to a tipping pointpartial	Ditlevsen & Ditlevsen (Nature Communications, Jul 2023) reported rising variance and autocorrelation in an SST-based proxy, estimating collapse between 2025 and 2095 (95% CI, central ~mid-century); the proxy and statistical assumptions are heavily disputed. van Westen et al. (Science Advances, Feb 2024) built a physics-based indicator (Atlantic freshwater transport minimum) and concluded the AMOC is "on tipping course" but explicitly could not date the threshold. Neither result is accepted as an operational forecast by any monitoring institution. [source]	2024-02-09
✅	Observed regional fingerprint: the subpolar North Atlantic 'cold blob' (warming hole)present	The ocean south of Greenland has cooled over more than a century while the global ocean warmed — an anomaly shown in IPCC maps since AR4 and in the AR6 Summary for Policymakers. A study published 28 May 2026 (Geophysical Research Letters) attributes the cold blob primarily to reduced ocean heat transport rather than air-sea heat loss, matching the modeled fingerprint of a weakening AMOC; a 2025 Science Advances study (ads1624) documents the reinforcing atmospheric pathway. [source]	2026-05-28
✅	Formal institutional assessment and an official watch infrastructurepresent	IPCC AR6 (2021): decline "very likely" this century under all scenarios; "medium confidence" of no abrupt collapse before 2100. On 19 Oct 2024, 44 AMOC and tipping-point scientists from 15 countries wrote to the Nordic Council of Ministers that the risk "has so far been greatly underestimated" and that medium confidence "is not reassuring." The Council responded: a commissioned risk assessment, "A Nordic Perspective on AMOC Tipping" (TemaNord 2026:504, led by the Finnish Meteorological Institute), was published 13 March 2026. [source]	2026-03-13
—	Mainstream literature assigning a likely full collapse before 2100absent	Baker et al. (Nature, Feb 2025), across 34 climate models under extreme greenhouse and freshwater forcing, found Southern Ocean winds sustain a weakened-but-running AMOC in every case — collapse this century judged unlikely. Drijfhout et al. (Environ. Res. Lett., 28 Aug 2025) found northern-AMOC shutdown by 2300 in 67% of extended CMIP6 runs under SSP5-8.5, 30% under SSP2-4.5, 21% under SSP1-2.6 — with the shutdowns unfolding after 2100, over 50–100 years, though the commitment point may come earlier. The center of gravity: substantial weakening this century, shutdown as a next-century tail risk scaled by emissions. [source]	2025-08-28

The honest read

Base rate: in the roughly 12,000 years of the stable Holocene, the AMOC has never collapsed; abrupt shutdowns are documented only in glacial climates under massive meltwater pulses, and the one prior instrumental-era collapse scare (2005, from sparse snapshot data) dissolved into natural variability. The measured 2004–2023 decline (~1 Sv/decade) is what climate models project for ordinary forced weakening — five times too slow for a mid-century collapse (McCarthy et al., 2025). What usually happens, and is happening, is gradual weakening with regional effects; full shutdown is a post-2100 tail risk that grows with emissions (21–67% of extended CMIP6 runs by 2300, Drijfhout et al., 2025). The stage moves if the RAPID trend crosses signal-to-noise 2 — not expected before the 2040s — or if IPCC AR7 withdraws the "no abrupt collapse before 2100" assessment; this file is the reference case for a slow risk, and it should be read at that speed.

The watch

A future automated watcher polls: RAPID-MOCHA-WBTS 26°N transport time series — https://rapid.ac.uk/data/data-download — calibrated releases roughly every 12–18 months (latest release Sept 2024, covering April 2004–early 2023); mirrored with commentary at the Met Office Climate Dashboard AMOC page, https://climate.metoffice.cloud/amoc.html. Secondary trigger to poll: IPCC AR7 WG1 drafting of the AMOC abrupt-change assessment.. Until then, the stage is reviewed by hand against the same sources — and says so.