Thursday, March 7, 2013

Wimpy models or wimpy DC?

The lack of snow yesterday was probably the biggest bust of winter weather forecasting this season (emphasis mine):

They shut down the schools and the government for what turned out in much of the area to be light rain and a bit of a breeze.

But when “they” — beleaguered bureaucrats, politicians and educators — made that choice in the small of the night, all the experts were telling them that a big storm with an even bigger name was bearing down on the Washington region, preparing to snarl roads, snuff out power and cause all manner of chaos.
“We made our decisions based on, unfortunately, faulty weather predictions,” said Pedro Ribeiro, spokesman for D.C. Mayor Vincent C. Gray (D). “You can’t really blame the government officials for using the data the scientists gave them.”

Ribiero said media forecasts “did tend to hype it up a little bit, but what we look at is the data. We have to take that as valid.”

Those forecasts were plain wrong. “This was the biggest bust in the history of the Capital Weather Gang,” said The Washington Post’s chief meteorologist, Jason Samenow. He said the major mistake was to accept computer models that said the amount of moisture in the storm would make up for the warmth of the air below.

Every forecaster in town overestimated the storm; at 3 a.m., the Weather Service still predicted eight to 10 inches of snow for the District.

Have we placed too much reliance on models? Or is there some hubris in the weather forecasting community now that 1-3 day forecasts have gotten more accurate than in the past?

Or perhaps there is another reason:

But Bowers, who has worked for the system for three decades, said he might have reached a different conclusion years ago. In the 1980s, Montgomery never shut down, he said. Bowers remembers a convoy of plow trucks heading up Interstate 270 to clear a path for buses to pick up sixth-graders who had been stranded at an outdoor education camp for a week.

“We’ve gotten wimpier,” he said.

No comments: