Snow bust: Lessons

After a week of drum-beating yielded but a trace of snow, those who were counting on a decorative snowfall and a day or two off from school this week (or at least a late opening) are entitled to more than a trace of disappointment.

As a matter of policy, we oppose finding comfort in others' miseries, but the Beltway snow-lovers suffered a far deeper hurt because of higher expectations, and some of them aren't taking it well.

You can read about it on this American Weather Forums link.

That no measureable snow fell in Philadelphia yesterday qualified as a major surprise, but perhaps more shocking was the lack of precipitation of any kind.

What we learned anew is that the computer models and the people who rely on them are fallible.

Although the storm was foreseen a week in advance, this was not a crowning moment for computer science, particularly for the North American Mescoscale Model, or NAM, used for short-term forecasts.

The NAM was calling for about 1.3 inches of precipitation for Philadelphia, according to weather service meteorologist Ray Martin, who was keeping score. Had that much fallen, and the bulk of it as snow, that could have meant several slushy inches.

However, only 0.13 inches of liquid was measured at Philadelphia International Airport, with only a "trace" of snow.

As Peter Mucha's online story points out, heavy precipitation was supposed to chill the air column and lead to a changeover to all snow.

The NAM amounts never materialized, and temperatures hovered close to 40 during the storm, and most of what fell was liquid.

Part of the error had to do with the fact that the precipitation shield didn't get as far northwest as expected, said Geoff Manikin, at the goverment's Environmental Modeling Center.

But part of it evidently had to do with the model's inherent wet bias.

Jon Nese, the former Franklin Institute meteorologist now at Penn State, said he typically trims the NAM estimates by a third.

Tony Gigi, meteorologist at the National Weather Service in Mount Holly, recommends going further, cutting by half. "This model has just been way too wet lately," he said.

"It really does make these east coast storms wet,” said Manikin.

“As to why that is, we don’t have a great answer," he said. "It is something we’re looking at.”

Overall the U.S. models generally are inferior to the European, in the view of Cliff Mass, a University of Washington professor of atmospheric sciences, and need serious upgrading.

But in this particular case, he says the forecasters deserve some slack.

After all, this is March, a volatile period for the capricious atmosphere, and the career of this storm as a snow-maker in the I-95 corridor was never clearly resolved by the models.

"This was a difficult forecast," he said."There clearly was a lot of uncertainty."

If anything, he said, he would fault the weathermen for not making that clearer. "The biggest failure was the lack of communication of the uncertainty,” he said.

In the discussions posted publicly by the Mount Holly weather-service office, the reservations about the forecasts were evident.

How much of that came through to the public is another matter; generally we've found that in matters of weather,  the public appetite for uncertainty is less than voracious - especially in matters of snow.

As Gigi pointed out, for whatever reasons snow is an "emotional" issue. "There aren't many people in the middle of the road," he said. "You either love it or you hate it."

What everyone does want is an accurate forecast, including the weathermen - to the nth power.

The atmosphere often has other ideas.