Michael McCarthy: Don't hang the Met Office out to dry

Analysis

They say in French that to understand everything is to forgive everything, and very few people in Britain are in a forgiving mood right now with the UK Met Office, after its scientists promised us a "barbecue summer" – when we have just had one of the wettest Julys on record.

Since admitting it was wrong, earlier this week, the Met Office has been subject to a torrent of criticism, which reached a peak yesterday with a derisive piece by the commentator Simon Jenkins, who said that as the forecasters were now predicting that August would continue to be wet, it would be a good idea to stock up on sun cream.

The chaps down at Exeter (Met Office HQ) are feeling "very bruised", one said yesterday, and while there is indeed at the moment no public inclination to forgive, perhaps it does no harm to try to understand. What exactly did they say and why did they say it?

When the Chief Meteorologist, Ewen McCallum, stood up a press conference in London's Science Media Centre on 30 April, he said it was "odds-on for a barbecue summer". We know all about the last two words. Let's focus on the first two. The odds he was speaking of were precise: they were 65:35. That was because his computer programmers had run 50 different simulations of the weather over the coming summer, and 65 per cent of these had indicated it would be warmer and drier than average, while 35 per cent had indicated the opposite.

Modern weather prediction involves assembling millions of pieces of data from around the world – wind speed, air temperature, air pressure, humidity – and working out how these phenomena will act on each other, simply according to the laws of physics. To get a perfect forecast you would need an infinite amount of data, but with the few million data points we now have we can get a good picture of the next five or six days.

However, accurately predicting longer than that – to make a seasonal rather than a weekly forecast – is very much harder, as a tiny difference in the data inputted at the beginning of such a program can make, over time, an enormous difference in the outcome. (This is the meaning of the often-quoted "butterfly effect" – the microscopic atmospheric perturbation caused by a butterfly flapping its wings might eventually, in theory, result in a hurricane).

So for seasonal forecasting a large numbers of model runs are carried out, each with slightly different inputs. Mr McCallum was merely reporting the result of this particular "ensemble" of forecasts: a 65 per cent chance of a good summer, a 35 per cent chance of a poor one.

This is known as a "probabilistic" forecast, but the public are not used to it, so this year the Met Office tried to "put some flesh on the bones" of its dry percentages, as a source said

The word "barbecue" was the mistake. We saw the barbecue; we overlooked the 35 per cent chance of rain. But it was there. They said that too. They didn't simply get it wrong, so give them a break. They merely ignited hope with a dream of patios and charcoal aromas, and hope is one of the most dangerous commodities of all.

Comments