Here's a wonderfully complicated essay, with magnificently incomprehensible charts, from OmniNerd that analyzes the accuracy of weather forecasts found on the web. If you understand it don't tell me - I don't want to know. But from reading the conclusions way down at the bottom of the page I gathered the Weather Channel forecasts of high and low temps seem to be more accurate than those of the service I've been using, Accuweather. So I looked a little more closely at the Weather Channel and sure enough, their forecasts for the next 10 days were about 3 degrees warmer than Accuweather's. I have switched widgets. I feel warmer already.
The weather in this particular spot, where I live, seems very difficult to predict and none of the services are right very often. In fact, accuracy seems to vary from year to year - some years one service is more likely to be right, some years another. I don't know why that should be but I imagine it has something to do with the forecasting models they use. So it remains to be seen if the Weather Channel will be any more accurate than Accuweather this year, but in the meantime just looking at the warmer widget cheers me up.
A bazillion years ago - well, half a century, to be exact - I had an algebra teacher who was a former Navy meteorologist and who used to get off onto a weather riff from time to time. What he said was, you can beat almost any weather forecaster over time by simply forecasting today's weather for tomorrow every day and keeping score. I've done that informally ever since, and I'm a pretty good predictor overall. One of these days maybe I'll get around to a more careful test.
No comments:
Post a Comment