First, let me note that up here in Maine we are getting another storm that is supposed to dump a foot of snow on us. This one doesn’t have enough wind to qualify as a blizzard, but snow is snow. Second, I’d like to add a few notes about the previous storm, the blizzard that socked most of New England but spared New York City.
From my perspective, it’s a good thing when events cause the mainstream press to talk about uncertainty, whether the subject is climate change, CBO projections of the budget, or just tomorrow’s snowfall. The so called “Blizzard that Wasn’t” has generated scores of news articles, and in reading any of them one can pretty quickly tell if the writer “gets it”. I’ve been pleasantly surprised by the quality of the discussion. While there has been plenty of criticsm of the forecast, much of it has been not simplistic complaints about how “they blew it” but rather focused disapproval of how much (or little) uncertainty the National Weather Service (NWS) chose to communicate to the public. In defending his organization and generally sticking to his guns, Louis Uccellini, NWS Director, said “This was the right forecast decision to make,” which is as close to “good decision, bad outcome” as we’re ever going to hear.
There’s obviously a cultural component to this as well. New York is the country’s largest city, a major media hub, and a place populated by folks not known for suffering in silence. If you’re a forecaster and it’s a close call, you’re in a tough position.
Looking beyond the headlines, at the decision making process in the weathermen’s war room, it turns out there was something of a multiple experts problem with two competing weather models, the NWS built one and another developed in Europe. For an easy storm the two might agree, but when they don’t, as in this case, human forecasters have to make the call. It doesn’t matter how many supercomputer CPU hours you chew up, the probability is still in your head! The forecasters went with the European model, which had a better recent track record and had notably been spot-on with Superstorm Sandy. This time, the outcome was closer to the NWS model’s prediction.
By the way, if you were looking for a sensible assessment of the storm forecast (before the outcome was known), it would be hard to beat this article from Vox’s Brad Plumer, which includes subtitles like “How bad could the storm get?” and “Could this snowstorm end up being a bust?”. There’s nothing sexy about a “To be sure” paragraph, but its presence is often a signal that you’re getting a level headed account of the facts.