Why #BlizzardOf2015 Was a Bust

Despite forecasts for a “potentially historic” storm, it seems the only history the #blizzardof2015 made in New York City was that the subway was shut down for a snowstorm for the first time. Storm totals in Manhattan so far are around 8", falling far short of the 20-30" forecasted by the National Weather Service’s (NWS) New York City office. Totals increase moving east—areas of Queens received about a foot of snow, while Long Island faced the brunt of the storm for New York, with widespread reports of 20"+.

Dr. Louis Uccellini, Director of the National Weather Service, urged reporters Tuesday afternoon to turn attention to Long Island and eastern New England, where snow was still falling Tuesday afternoon. “Everyone’s focused on where it didn’t snow,” Uccellini said. But the amounts in eastern New England, which are still increasing, “will likely be historic.” Those historic amounts? Heavy snow dumped across eastern Massachusetts Connecticut, and Rhode Island with totals topping two feet and reaching close to three feet Tuesday afternoon. Massachusetts also saw coastal flooding and the highest wind gusts of the storm, with Nantucket recording a 76 mph gust at 5:30 Tuesday morning.

Although impending storms are often hyped in media, in the case of this storm meteorologists conveyed what the computer models were forecasting, which was a big snowfall for NYC. So why did NYC get less than half of the forecasted amount?

The main reason is that the storm tracked about 50 miles east of where the models predicted it to track. Meteorologists make a forecast by looking at a suite of models, each with its own unique storm track and intensity, and they use experience to predict which models are going to perform better in a particular situation. If the models all agree with each other, meteorologists feel more confident in their forecasts. For this storm, by Monday morning most models indicated a high probability for heavy snowfall over New York City. The only exception to this, and the one that appears to have been fairly accurate, was the American global model called GFS, which has just been updated with higher resolution and more data. According to Uccellini, the updated GFS was just implemented January 14, and it takes awhile for forecasters to calibrate the models for how they work in different situations. Uccellini said there was still likely a tendency for forecasters to use the models with which they are more familiar.

If the model errors had occurred in a less densely populated area, this forecast bust may have gone relatively unnoticed. But given that the 50-mile error affected NYC, it looks like an even worse forecast. This is compounded by the high societal impact of snow. These kinds of errors happen in rainstorms too, but most people don’t notice the difference between getting 0.5 inches and 2 inches of rain unless they live in a flood plain.

Another factor that led to only moderate NYC snowfall has to do with what’s called banding. In most cases, precipitation happens because air near the ground is warmer than the air above. The warm air rises because it is less dense than the cold air above it. As the warm air rises, it cools and the water vapor contained in the air condenses to liquid or frozen water. As the snow moved westward over Long Island, the air ahead of the precipitation (over the city) was sinking. Sinking air is the opposite of rising air in that it does not allow for much precipitation to form. In this case, the sinking air also fueled the heavy bands of snow in the areas where air was rising. Because of the way this storm was formed, forecasters knew this phenomenon would likely occur and that there would be a sharp drop off in the snowfall amounts to the west of the heavy snow bands. The forecasters just thought that line would be further west.

“But we are getting better at predicting these things," Meteorologist David Nicosia of the Binghamton, NY NWS office said this morning. “Thirty years ago we may not have even known a storm off the coast was going to develop. The blizzard of ‘78 really took people by surprise, and now it’s a question of being off 30-40 miles. With the amount of information generally available today, society’s expectations have risen faster than our ability to provide perfectly accurate forecasts.”

The New York City NWS office posted similar remarks on their Facebook page, “The science of forecasting storms, while continually improving, still can be subject to error, especially if we’re on the edge of the heavy precipitation shield. Efforts, including research, are already underway to more easily communicate that forecast uncertainty.”

NWS Director Uccellini also emphasized the need to better communicate forecast uncertainty. Part of that uncertainty resides in the fact that predicting weather is inherently probabilistic. It’s easy to look at the 80 percent chance of at least a foot of snow and assume that there will be a foot of snow. But 1 in 5 odds of not getting that much snow is not an insignificant chance.

Officials acted on the most probable outcome based on the data available, and they maintain that the right decisions were made. In a press conference this morning, Mayor de Blasio defended the city’s choices, repeating the adage, “Prepare for the worst, hope for the best.” MTA Chairman Tom Prendergast added that because systems went offline last night, “We’re well in the position to get service up back faster.”