If your hometown was anything like mine, there was almost always a small amount of sneer towards the weatherman on TV about how they a: knew nothing and b: would always get the forecast wrong. They were sometimes the exclusive source of blame when the traffic commute was slow because they didn’t predict rain or perhaps the cause of frustration when the family made plans for the weekend based on the outlook forecast which didn’t necessarily materialize as published.
And I was one of those people. Right up until being forced to do what my local weathercaster had to do and predict the weather. Here I learned that the profession of forecasting is probably the closest thing people can be to being a clairvoyant with requires a college degree.
Please insert 25 cents for weather forecast.
But why do they get it wrong? It is their paid job, to be right. Because even though I may have just referred to it as basically a job of predicting the future, the secret behind it is that the whole thing is one very large, educated guess. And usually it is computers doing the guessing.
When one goes to create a forecast these days, they reference a series of “models” which are essentially very large super computers doing very complicated math to guess what is going to go where in the atmosphere. On a large scale, this can work out okay – but let’s look at a city, say Chicago which has an area of 234 sq miles. That works out to an effectively 15×15 mile box. No too big right? And the good people of Chicago want to know what the weekend weather will be for the weekend on Wednesday or Thursday so they can start making plans. So the computer models take in their data and try to figure out whether or not that part of the jetstream is going to move the city center and cause less than desirable picnic weather. And then the forecast will more than likely reflect this computer’s guess. But it is still a very, calculated mathematical guess. And as the weekend gets closer, the computer has to guess less and less, until pretty much the day of where finally a number could be pegged with reasonable confidence for say, temperature. So when the forecaster gets the weekend temperature wrong by fifteen degrees several days in advance, it may not have been his fault. Put frankly, technology at its current point simply doesn’t know us to allow any better to know if that one atmospheric feature will hit that 15×15 sq mile box.
RED BALL OF DEATH, RUN CHICAGO!
This isn’t to say we haven’t made huge advances. Now armies of computers work on these guesses, but it wasn’t always that. In 1922 Lewis Richardson prepared the first six hour weather forecast which took six weeks to create and it wasn’t even very accurate (Lynch 1-21). In comparison, computers now generate two forecast models every day for the most popular weather model solutions and some even go hourly – this isn’t even including some of the in house work that universities do to modify some models to try and make them more accurate. So in short, we have come a really, really, long way since 1922.
“I’m sorry Chicago, I can’t let you picnic this weekend.”
At the end of the day, cut the weather forecaster some slack when he stands up on TV. Predicting the future sure wasn’t an exact science for the ancients Greek oracles and it still isn’t one for us today. We just stick more math and less crystal balls at the problems.
IT GON RAIN.
Lynch, Peter (2006). “Weather Prediction by Numerical Process”. The Emergence of Numerical Weather Prediction. Cambridge University Press. pp. 1–27. ISBN 978-0-521-85729-1.