I’ve long wanted to keep records of weather forecasts and compare them to actual results. But, I’m too lazy.

Fortunately, Forecast Advisor has started to do that for us! And, it turns out that my suspicions were correct.

At least for Boston, the weather forecasts are less than 80% accurate. Now, consider that these data are looking at just the 1-3 days ahead forecasts, and in fact are collapsing the data. The day-ahead forecasts are usually more accurate than the 3-days ahead forecasts. The company sells a data feed that I’m sure drills down to this level and beyond.

But consider: for March 2011, the National Weather Service predicted the high temperature within 3° F of the actual high temperature only 40% of the time! They accurately predicted the low temperature +/- 3° 53% of the time in March. They did slightly better with their precipitation forecasts – if they called for rain or snow, they were right about 83% of the time.

Over the past year, they did better on temperature predictions (63% right on the highs, 70% right on the lows) but worse on the precipitation calls (75% accurate).

What can we conclude? National Weather Service forecasts are better than chance… but not by a lot. Turns out, you don’t need a Weatherman to know which way the wind blows…