Officials announced Thursday morning that the famous groundhog Punxsutawney Phil saw his shadow, which according to tradition means we will suffer through six more weeks of cold weather.
While it doesn’t take a genius to say depending on a rodent for a bona fide weather forecast isn’t the greatest idea, official studies on how accurate Phil is have been conducted.
The answer? He’s terrible.
Phil has been predicting whether we’ll have six more weeks of winter since 1887, though only 121 of those times have been recorded, including this year, according to Stormfax Almanac. It’s unclear why nine years are missing from the record.
He has predicted six more weeks of winter 108 times, and an early spring only 18 times. According to Stormfax, Phil has only been right 39 percent of the time.
“If Punxsutawney Phil is right 39 percent of the time, that’s much, much worse than a climatological prediction,” Tim Roche, a meteorologist at Weather Underground, told Live Science. “Even if you flip a coin, you’ll still be right close to half of the time, that’s a 50 percent accuracy rate. So you’ll be better off flipping a coin than going by the groundhog’s predictions.”
Phil’s accuracy is even worse if you look at his predictions after 1969, when the accuracy of weather records is less in question, Roche added. Looking at just those years, accuracy drops to 36 percent.
But Phil shouldn’t feel too bad. Long-range weather forecasts tend to be inaccurate even by experienced meteorologists, according to the Washington Post. One-day weather forecasts are typically accurate within a few degrees, but even five-day or 10-day forecasts get more unpredictable.
“We have never seen an independent, peer-reviewed analysis supporting the ability to provide a skillful forecast at such long ranges,” wrote Jason Samenow and Angela Fritz, weather editor and deputy weather editor. “By skillful, we mean that the forecast offers an improvement over simply looking at the average weather for a given location at a given time.”