Can the North Atlantic Tell Us What European Summers Will Do Years Ahead?

There is a frustrating gap in what climate science can and cannot tell you. Ask whether Europe's summers will keep warming over the coming decades, and the models answer with confidence. Ask what the summer three or five years from now will actually be like (wetter or drier, closer to average or another scorcher), and the same models mostly shrug. That middle distance, the years-to-a-decade range, has been one of the hardest windows to forecast.

A study published on 22 July 2026 in Earth System Dynamics, by Williams, Müller and Pinto, makes the case that the North Atlantic Ocean holds part of the missing signal. Their focus is Atlantic Multidecadal Variability, or AMV: a slow, decades-long rise and fall in sea-surface temperatures across the North Atlantic. Because the ocean stores and releases heat far more sluggishly than the atmosphere, a pattern like AMV changes on a timescale that, in principle, a model might see coming.

The team tested that idea with initialised decadal hindcasts of the MPI-ESM climate model: runs that are started from observed ocean conditions and then asked to predict forward, so their skill can be checked against what actually happened. Two results stand out. First, AMV itself proved predictable roughly one to seven years ahead in the model. Second, and more surprising, AMV leaves a fingerprint on European summer weather (a cyclonic circulation pattern) that is essentially invisible in ordinary historical simulations but emerges once the model is initialised from real ocean states.

That contrast speaks to a genuine puzzle in the field known as the "signal-to-noise paradox." In several settings, forecast models seem to predict the real world's swings better than they predict their own, as if the true climate signal is being drowned out by too much internal noise inside the models. The European summer imprint here behaves like a case of exactly that: present in reality and in the initialised runs, muffled in the free-running ones. The paper does not claim to resolve the paradox, but it adds a concrete instance of it, and a physical mechanism to hang it on.

This is a mechanism-level finding drawn from one climate model's hindcasts. It says there is a real, physically grounded source of multi-year predictability in the North Atlantic, and that current models can partly capture it. It does not hand forecasters a finished tool, and it is emphatically not a forecast for any particular summer. Whether the effect holds up across other models, and how much usable skill it delivers in practice, are open questions the authors leave for further work.

Still, the direction is encouraging for anyone who cares about planning years, not just months, ahead: water managers, farmers, energy systems that must brace for hot, dry stretches. If a slow ocean rhythm carries even a faint, readable signal about the summers to come, that is a thread worth pulling. This study pulls it, and finds something on the other end.