Friday, May 07, 2010

Weather Forecasts

A colleague of mine, Antony Davies, sends along this information about Pittsburgh's upcoming weather:

Weatherbug says the prob of rain from 9:00 to midnight is 96%.
Weather.com says the prob of rain from 9:00 to midnight is 10%.

I've always thought that weather forecasts were biased towards poorer weather-- if there's to be error, you'd rather it in that direction-- but that is a remarkable spread.

Does anyone know of any systematic studies on weather forecasts and bias?

2 comments:

doclawson said...

Dan Sutter's done loads of work on the political economy of weather forecasting.

Anonymous said...

http://freakonomics.blogs.nytimes.com/2008/04/21/how-valid-are-tv-weather-forecasts/

This guy tracked the forecasts of the four local news stations' weather forecasts for 7 months and compared them to NOAA's forecast and the actual weather.