The Brief

Researchers at Utrecht University have identified an abrupt northward shift in the Gulf Stream — a 219-kilometre jump occurring within two years — as a potential early warning signal that appears roughly 25 years before the Atlantic Meridional Overturning Circulation collapses in high-resolution simulations. Satellite altimetry data from 1993 to 2024 show the Gulf Stream has already been drifting north at a statistically significant rate along the U.S. East Coast, though whether this reflects AMOC weakening or natural variability remains contested.

The Report

A study published in Communications Earth & Environment by René van Westen and Henk Dijkstra of Utrecht University’s Institute for Marine and Atmospheric Research has found that an abrupt northward displacement of the Gulf Stream precedes the collapse of the Atlantic Meridional Overturning Circulation by approximately 25 years in high-resolution ocean simulations — offering what the authors describe as “a very accurate early warning indicator that actually goes off.”

The AMOC is the system of ocean currents that carries warm, salty surface water from the tropics to the North Atlantic, where it cools, sinks, and returns southward at depth. A healthy AMOC transports roughly 17 to 20 Sverdrups — each representing one million cubic metres of water per second. The Gulf Stream is its most visible surface component along the U.S. East Coast.

Using a model running at 0.1-degree resolution — ten times finer than the standard climate models used by most research groups — the team slowly added freshwater to the North Atlantic to simulate ice sheet melt. As the AMOC weakened, the Deep Western Boundary Current that normally flows beneath the Gulf Stream lost strength, reducing the vorticity that tethers the Gulf Stream to the continental shelf. The result was a two-stage process: a gradual 133-kilometre northward drift over 392 years of simulated weakening, followed by an abrupt 219-kilometre jump in just two years. Twenty-five years after that jump, the AMOC collapsed entirely.

The researchers then compared their simulation against real-world observations. Satellite altimetry from 1993 to 2024 shows the Gulf Stream’s detachment point near Cape Hatteras, North Carolina, has been shifting north at 0.16 degrees latitude per decade — a statistically significant trend. Subsurface temperature records dating to 1965 confirm the pattern. The daily-averaged Gulf Stream path has already reached 38 degrees north. “That’s something we can measure,” van Westen said. “So it’s very likely that this reflects that the AMOC is indeed weakening.”

The finding is not without complication. Sang-Ki Lee, an ocean-climate researcher, cautioned that the observed northward shift in satellite data is linked to a strengthening of the current driven by the North Atlantic Oscillation — not the weakening the study’s mechanism requires. “It is too premature to say that the shifting is already happening,” he wrote. A separate 2025 study from the Woods Hole Oceanographic Institution, analysing air-sea heat fluxes rather than surface temperatures, concluded the AMOC “has not declined yet” over the past sixty years — though its lead author, Nicholas Foukal, added that this “doesn’t say anything about its future.”

The 25-year lead time between the abrupt Gulf Stream shift and full AMOC collapse is itself model-dependent. Under faster warming than the simulation’s deliberately slow forcing, that window could shrink considerably. Stefan Rahmstorf of the University of Potsdam noted that climate models “appear to be underestimating the problem, and thus potentially how soon the AMOC tipping point will be reached.” Separate statistical analysis by Peter and Susanne Ditlevsen places the most likely collapse date around 2065, with a 95 percent confidence interval spanning 2037 to 2109.

If the AMOC were to collapse, the consequences for Europe would be severe. Modelling published in Geophysical Research Letters last year projected one-in-ten-year winter cold extremes reaching minus 19 degrees Celsius in London, minus 30 in Edinburgh, and minus 48 in Oslo — even in a world 2 degrees warmer than pre-industrial levels. Peter Ditlevsen of the University of Copenhagen described the scenario as “a going-out-of-business scenario for European agriculture.” Sea levels along the North Atlantic coast could rise by an additional metre on top of existing projections, with the U.S. East Coast particularly exposed.

The study’s resolution advantage — capturing eddies and baroclinic processes invisible to standard models — is also its limitation: it remains a single-model result. Multiple models running at comparable resolution would be needed to confirm whether the Gulf Stream pathway is a reliable operational warning signal or an artefact of one simulation’s particular bathymetry and forcing profile. Direct AMOC monitoring through moored instruments has only been running since 2004, too short a record to separate signal from noise.

Van Westen framed the contribution in practical terms. “The advantage of this Gulf Stream early warning indicator is that it can be observed in near-real-time,” the authors wrote, “while other proposed indicators require much longer observations.”

The Angle

The value of the van Westen study is less in its specific numbers — the 219 kilometres, the 25-year lead time, both of which are products of one model under one forcing scenario — than in the kind of problem it is trying to solve. The AMOC is a system whose collapse would restructure agriculture, infrastructure, and habitability across the North Atlantic within decades. The primary monitoring tool for that system has been running for twenty-one years. The standard models used to project its future cannot resolve the physical processes that govern it. The question the study surfaces is not whether the Gulf Stream is the right indicator. It is why the infrastructure for answering that question does not already exist at the scale the risk demands.

A 25-year warning window sounds generous until you consider what it contains: the detection itself, the multi-model confirmation, the political absorption, the engineering response. Rahmstorf’s observation — “once we have a definite warning signal it will be too late” — is not alarmism. It is a statement about the mismatch between the timescale on which this system operates and the timescale on which the institutions responsible for responding to it make decisions. One is measured in decades and Sverdrups. The other is measured in election cycles and fiscal quarters. The instrument mismatch is more dangerous than the current itself.

The AMOC will do what the physics requires it to do. The open variable is not the ocean. It is whether the species watching it has built the apparatus — observational, computational, institutional — to act on what it sees before the window it is measuring has already closed.