The weather forecast that shaped the D-Day landings still looks hard to beat. New research from the University of Reading shows that today's best weather computers, including modern machine-learning models, would have struggled to call the June 1944 weather window any better than Captain James Stagg and his team did with sparse observations and no satellites.
The study, published in the journal Weather ahead of the release of the D-Day film Pressure, used AI-based weather models to produce 50 different forecasts of the storm that forced General Dwight Eisenhower to delay the invasion by 24 hours.
A forecast made with almost nothing
In 1944, forecasters had no satellites, no radar networks over the Atlantic, and only a handful of weather ships. Stagg, chief meteorologist to the Allied forces, advised Eisenhower to move the invasion from June 5 to June 6, when a narrow window of calmer weather was expected over the English Channel.
Lead author Professor Andrew Charlton-Perez said: "Captain James Stagg had one of the hardest jobs in history. Get the forecast wrong, the invasion could have failed and the Second World War could have been lost."
Machines add data, not certainty
The researchers fed modern models with the best available reconstructions of the 1944 weather. Even so, the forecasts disagreed about whether the June 6 window would hold. The decisive weather features were small and short-lived, the kind that still slip through the gaps between observations.
The authors say the D-Day case shows the limits of computing power alone. The human ability to weigh conflicting evidence and make a judgment under pressure remains central to forecasting, even in an age of machine learning.
Why it matters now
Weather models improve every year, but the D-Day test suggests that small, fast-moving storms remain their weak spot. The findings point to the value of denser observation networks and to the care needed when handing forecast decisions to AI. The same storm that tested Stagg now tests the machines built to replace him.