Day is now longer than night in the northern hemisphere as we take our first steps into spring. But before we forget about winter, it’s worth reviewing how forecasts for the season fared. For the National Oceanic and Atmospheric Administration, which released its winter outlook Oct. 18, its temperature prediction was the worst in 15 years.

And it was one exceptional month, February, which did most of the damage to the outlook’s accuracy.

When you compare NOAA’s winter temperature forecast to what happened, it’s not a pretty picture. Its general idea favored mild conditions in the western half of the nation and near normal temperatures in the east. Reality showed something close to the opposite. Temperatures actually averaged out near to below normal in the west and mild in the east.

In a review of the outlook for Climate.gov, Tom Di Liberto of NOAA’s Climate Program Office reported the temperature prediction had “negative skill,” meaning it was worse than a random guess.

That said, a three-month evaluation of NOAA’s outlook doesn’t tell the whole story, because the predictions were essentially on the mark for the first two-thirds of the season.

As late as the end of January, the outlook, prepared by NOAA’s Climate Prediction Center, seemed on track. Many of the areas highlighted for above-normal temperatures did see above-normal temperatures for the majority of the winter.

But things changed in a hurry once the calendar flipped over to February. Brutal cold snaps, influenced by the much-advertised polar vortex, caused temperatures to plummet at the beginning of February, and they didn’t recover for more than a month.

“This reversal in temperatures (which has continued into March) was not caught by the dynamical models even two weeks prior to February, as our half month lead February outlook favored above-average temperatures out West,” the Climate Prediction Center said in an email.

Let’s take a look at Glasgow, Mont., which endured one of its coldest stretches in recent memory. Average temperatures at the airport in Glasgow came in 6.6 degrees above normal in December and 8.4 degrees above normal in January. This winter appeared to be on track to follow NOAA’s outlook and come in above normal.

But then Glasgow endured its second-coldest February since records began at the end of the 1800s. The average temperature for the month was 22.6 degrees below normal. February began in Glasgow with two days of above-freezing high temperatures before a cold front sent temperatures plummeting below freezing, where they would remain for the rest of the month. Glasgow measured 10 full days with temperatures below zero, including six low-temperature readings below minus-20 and one morning, Feb. 8, coming in at minus-32.

While not as extreme elsewhere, this sudden pattern reversal didn’t just happen in the north-central part of the country. Seattle made headlines in February for the string of snowstorms that plastered the Northwestern city with 20.2 inches of snow for the month. Seattle saw temperatures 3 degrees above normal in December and 3.6 degrees above normal in January before the cold and abnormally snowy pattern settled in for February.

Why did conditions in the Northwestern states abruptly stray so far from what NOAA predicted for the season?

“This reversal is not thought to be caused by any of the factors that typically contribute to winter time predictability and is likely the result of natural climate variability,” the Climate Prediction Center said. It added that models didn’t start to signal an extremely cold February until late in January, and that it’s looking back to see whether hindsight will show that the cold spell was predictable.

The agency’s precipitation outlook was mostly on track, although wetter-than-normal conditions expanded over a larger area than predicted. It turned out to be the wettest winter on record for the Lower 48.

Di Liberto wrote that this part of NOAA’s outlook achieved a somewhat positive skill score, meaning it was modestly better than a guess.

Long-range weather outlooks are generally a good planning tool for the season to come. While they’re not as accurate as short-term forecasts, seasonal outlooks are still fairly good indicators of the probability of above- or below-normal temperatures and precipitation despite the occasional miss.

On balance, the Climate Prediction Center notes that seasonal outlooks are fairly accurate — “40 percent better than a random forecast (averaged over a 4-year period)” — and that the outlooks are more accurate some seasons and less accurate in others because of natural variability.

This past winter was just one of those in which one exceptional month foiled an otherwise solid forecast.

“These sorts of ‘failed’ outlooks put the complexity of seasonal outlooks into perspective,” Di Liberto wrote. “All it takes is one month’s extremes to flip the script on the seasonal story. And while there has been considerable progress in our ability to forecast seasonal climate, it is still incredibly difficult to impossible to predict how certain natural climate phenomena will evolve two to three months in advance.”