The ocean current system that keeps winters in London milder than winters in Montreal may be headed for a slowdown far worse than scientists previously projected. A peer-reviewed study published in Science Advances in May 2026 finds that the Atlantic Meridional Overturning Circulation could weaken by 42 to 58 percent before the end of this century, with a central estimate near 51 percent. That projection is roughly 60 percent more severe than the benchmarks used in the most recent international climate assessments, and it is grounded not in model speculation but in nearly two decades of direct ocean measurements.
For the roughly 400 million people living along Atlantic coastlines in Europe, North America, and West Africa, the implications are tangible: accelerated sea-level rise along the U.S. East Coast, disrupted rainfall patterns across agricultural regions in the Sahel, and the possibility that Northern Europe loses a significant share of the oceanic heat that moderates its climate.
What the new research shows
The study, led by Robin Portmann and colleagues, takes a different approach from earlier modeling exercises. Instead of letting climate models run unchecked, the team constrained them against real-world data from the RAPID monitoring array, a line of instruments moored across the Atlantic at 26 degrees North latitude that has tracked AMOC strength continuously since 2004. Using averaged readings from 2005 through 2023, the researchers filtered a large ensemble of climate models, keeping only those whose historical simulations matched what the ocean instruments actually recorded.
The models that survived that filter consistently projected steeper declines than the broader, unconstrained set. The resulting multimodel mean landed at roughly 51 percent weakening by 2100 under a high-emissions scenario, with the full range spanning 42 to 58 percent. By comparison, projections from the Coupled Model Intercomparison Project Phase 6 (CMIP6), the standard toolkit behind reports from the Intergovernmental Panel on Climate Change, had pointed to a notably smaller slowdown.
Portmann’s team also published its data and Python code through a Zenodo repository, an unusual step that allows any researcher to reproduce the analysis figure by figure. That transparency strengthens the paper’s standing, though it does not by itself resolve disagreements with other groups working on the same question.
A competing estimate complicates the picture
The Science Advances numbers do not stand alone. A separate study published in Nature Geoscience applied its own observational constraints to climate models and arrived at a weaker projected decline of roughly 18 to 43 percent by 2100. The gap is striking: the low end of Portmann’s range sits near the high end of the Nature Geoscience estimate.
The divergence comes down to methodological choices. Each team selected, weighted, and applied observational data differently when deciding which models to trust. No independent reconciliation of the two approaches has been published, and neither the Portmann group nor the Nature Geoscience authors have issued public statements explaining how their results should be read side by side.
What both studies agree on is the direction of travel: the AMOC is weakening, and the decline is likely to be larger than the previous generation of climate models suggested. The debate is over how much larger.
Slowdown versus shutdown
A related but distinct fear is that the AMOC could cross a tipping point and collapse entirely, not just slow down but effectively stop. Some researchers have identified statistical early-warning signals in observational records that hint the system may be approaching a critical threshold. However, separate modeling work suggests that a full shutdown is unlikely within this century, though it leaves the door open for a collapse after 2100 if emissions remain high.
The distinction matters enormously for planning. A 51 percent slowdown would stress infrastructure and ecosystems gradually, giving governments a narrowing but real window to adapt. A full collapse would be a different category of event, potentially shifting weather patterns across the Northern Hemisphere within decades. Current evidence points more strongly toward the slowdown scenario, but the tipping-point research means a collapse cannot be ruled out.
What a weaker AMOC would actually mean
The AMOC works like a planetary heat pump. Warm surface water flows northward from the tropics, releases heat into the atmosphere over the North Atlantic, then cools, sinks, and returns southward at depth. Weaken that pump by half, and the consequences ripple outward.
Northern Europe would lose a portion of the warmth the current delivers. That does not necessarily mean a new ice age, but it could mean colder winters, shorter growing seasons, and shifts in the jet stream that alter storm tracks. For the U.S. East Coast, a slower AMOC would reduce the gravitational pull that the current exerts on ocean water, allowing sea levels to rise faster than the global average. Coastal cities from Boston to Miami are already contending with accelerating tidal flooding; a significant AMOC slowdown would compound that trend. In West Africa, changes to Atlantic sea-surface temperatures could shift the West African monsoon, with direct consequences for food production across the Sahel.
The 2018 study by Caesar and colleagues, published in Nature, established that the AMOC has already weakened by an estimated 15 percent since the mid-twentieth century. The new projections suggest that decline is set to accelerate sharply, not plateau.
What to watch through mid-2026 and beyond
Three developments will determine whether the higher projections gain or lose ground. First, updated data from the RAPID array covering the period after 2023 could shift the observational baseline both studies depend on. If recent measurements show the AMOC weakening faster than the 2005 to 2023 average, the Portmann estimate would gain support; if the trend has stabilized, the Nature Geoscience range may prove more accurate.
Second, independent research groups may attempt to reconcile the two constraint methodologies. A head-to-head comparison using identical model ensembles and observational windows would clarify how much of the disagreement is methodological and how much reflects genuine scientific uncertainty.
Third, the next round of international climate assessments will have to decide which projections to incorporate. If the IPCC and national climate agencies adopt the higher weakening estimates, adaptation planning for coastal infrastructure, agriculture, and fisheries management would need to accelerate on timelines that many governments have not yet contemplated.
What is no longer in serious dispute is the direction. The Atlantic’s great conveyor belt is slowing, and the best available science now suggests it will slow considerably more than the models used to guide policy have assumed. The remaining question is whether “considerably more” means a manageable challenge or a generational crisis, and the answer depends on data that is still being collected from instruments anchored in the deep Atlantic.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
