Are weather forcasts too hot?
I’ve hacked up a little program to track the accuracy of weather forecasts in an attempt to determine if the forecasting computer models are flawed.
It’s no secret that predicting the weather is hard and even the experts frequently get it wrong, but I seem to be noticing that when they get it wrong, they always tend to error on the optimistic side. For example, if the 5 day forecast predicts 20C, it actually ends up being 15C. Being off by a few degrees is understandable but if it truely was a science, then it should be wrong the other way approximately the same amount of times. So when they predict 20C, half the time we should get 25C but it never seems to work that way.
I know very little about forecasting but from what I understand, all forecasters use computer models. Usually they run a few different models then set the forecast based on the consensus. If my theory is correct, my data will show that on average the forecast is always too optimistic and that will prove that the computer models are wrong.
If that turns out to be the case then it raises some other questions.
First of all, the weather modelers should already know this. They have access to all the forecast data and can easily compare the predictions with the actual temperatures and then adjust their models accordingly to make them more accurate. Why haven’t they done that?
A wrong forecast may not seem like a big deal but does it have other implications? Do the same people that wrote the software which makes short term forecasts also write the climate models that predict global warming?
It’s going to take some time to gather enough information to be meaningful (at least a year), and maybe I’m wrong? Once I’ve gathered enough data to make some pretty graphs I’ll post a link to a site where everyone can take a look at the results.
