Questions are now being asked about the MO’s accuracy rate-they claim around 85% but some experts put it as low as 38%-and the wisdom of having long range forecasts at all. And this is all despite the MO using a £30M super computer. It’s been pointed out before on this blog that anything other than short term forecasts have proved very iffy indeed. Now why is that?
Well if, like me, you’re interested in the answer, you could do worse than read some of the works of Professor Jeffrey S.Rosenthal. Jeff’s something of a clever clogs-he had his PHD by age 25 and is now one of the world’s leading experts on probability. Annoyingly he’s also a pretty good writer. His 2003 Probability and Statistics: The Science of Uncertainty; now a standard text across the pond, is a good place to start.
Basically, detailed weather forecasts are good for 2-3 days only. A week’s forecast is playing the long odds and anything beyond that is pure happenchance. There are simply too many variables that increase exponentially with the length of time. These are simply impossible to predict. Think the Butterfly Effect. You may as well flip a coin. And if you get it wrong, you can always blame the jet stream.