Lawrence Hogben on forecasting the weather for D-Day (LRB, 26 May 1994):

At the beginning of June, forecasting became more difficult when a steady Azores anti-cyclone started to misbehave, and a series of depressions threatened to run across the North Atlantic, with associated fronts menacing a hitherto sheltered Channel. Up to Friday, 2 June, it was still peaceful, but we suspected the calm was near its end. Then began the forecasters’ and the Supreme Commander’s nightmare: a calm period ending – to be replaced when, and by what? We speculated on Friday that the big change might arrive around Tuesday, 6 June. However, by the Saturday morning, our three forecast centres all agreed that the change was upon us, while disagreeing on what it would turn out to be, or when. On Saturday evening the moment for our crunch forecast arrived. The Americans were still relying on the Azores high pressure: they were for ‘go’ on Monday the 5th; the cautious Met Office were for ‘no go’; so were the pragmatic Royal Navy. Ike considered a postponement. Then, during the night, our observations showed that a stormy front was sweeping towards the Channel and, unwillingly, Caltech yielded. At a hairy 4 a.m. meeting on Sunday morning, faced by a unanimous prediction of strong winds, low cloud and rough seas for the Monday, Ike postponed the operation for 24 hours, only two hours before the main body was due to sail. So Monday was out, and the troops in their landing-craft tossed uncomfortably at moorings.