Cliff Mass of the University of Washington has a good breakdown of the “failure” of meteorologists to correctly predict snowfall during this week’s storm. “Failure” gets the scare-quote treatment because the forecast wasn’t actually far off: models showed most of the snow falling in a relatively narrow band on the west side of the storm system, which is exactly what happened. The placement of that snow band, however, was uncertain–depending on which of the several major weather models you looked at, it fell somewhere between New Jersey and Cape Cod. Small differences in the storm’s track would mean large differences in snow over NYC.
That uncertainty was understood and acknowledged by NOAA and the National Weather Service. This graph hoisted from Mass’s blog post shows snow accumulation from Monday through Thursday, as predicted by a group of models (one line per forecast). Each model was started with slightly different initial conditions, or subtly different representations for how the atmosphere works. This is what atmospheric scientists cal an ensemble, and it gives a sense of the uncertainty in NYC’s forecast: the forecast accumulation varied from about 6 inches at the low end to almost 40 inches at the top.
Predicted snow accumulation (in inches) in New York City from different models
For a variety of reasons, the reported forecasts skewed high. The NWS doesn’t do the best job it could communicating the uncertainty. TV news, all else equal, would rather have you freaked out and glued to your screen watching their parka-clad weatherman, and they report the forecast accordingly. I suspect a lot of it is just due to the fact that we aren’t that good at dealing with uncertainty.
And, of course, a lot of that “failure” can be traced to a certain attitude that mistakes a certain eight-mile island for the majority of the civilized universe. Out here in Suffolk County, the forecast was more or less right on, and we got hammered. The highest snowfalls in the state were at Orient Point, the tip of Long Island’s North Fork. There were no humans there to measure, but I suspect they were probably even higher at Great Gull Island.
