The blizzard of 2017, a.k.a. “Stella” according to The Weather Channel, was never a sure bet for New York City and Philadelphia. However, up until Monday night, the odds favored a period of near white-out conditions on Tuesday morning, making travel impossible for one of the most highly populated and economically productive corridors in the country.
Yet millions of people woke up to find transportation networks shut down for just a few inches of sloppy snow, with the “ping, ping, ping” of sleet hitting their windows.
So, given the issuance of a blizzard warning for all of metro New York City, people want to know: What happened to all the snow? And how did forecasters get this SO wrong?
Well, first, here’s the general explanation interspersed with a slightly more technical one.
This was always a tricky forecast for the Interstate-95 corridor, as computer models were showing that the snow/mix/rain line would fall right across the five boroughs of New York City, southwestward to Philadelphia.
The concern to forecasters was that there would be a battle between warmer air flooding in at low to mid-levels of the atmosphere ahead of the rapidly developing storm, and tremendous vertical motion that would generate heavy precipitation.
Such vertical motion can cause a column of air to cool through a process known as dynamic cooling, turning a mix of precipitation to snow.
In other words, some signs were pointing to a change from snow to sleet and rain, and others were showing an outright blizzard. What was a forecaster to do? Well, communicate the uncertainty while indicating the danger of a paralyzing blizzard.
Nevertheless, millions of people were outraged this morning that their much-advertised blizzard was not to be. This was despite the fact that the forecast was always highly uncertain for New York City, Philadelphia, coastal Connecticut and western Long Island.
Part of the problem may have been that blizzard warnings were issued, and National Weather Service (NWS) maps showing the “most likely” amount of snow were plastered across social media, and these consistently showed at least a foot of snow or more for New York.
These probabilistic maps are designed to show the range of possibilities and actually indicate forecast uncertainty, but people may tend to focus only on the upper end of a snowfall range anyway.
I’ve lived in NYC 23 years. It snows. I’ve also been through real blizzards. Now, every approaching snowstorm is hyped like Armageddon…
— Charles M. Blow (@CharlesMBlow) March 14, 2017
The combination of the word “blizzard” and “foot or more of snow” sent thousands flocking to jam-packed grocery stores on Monday and prompted Governor Cuomo to shut down portions of the city’s subway system in advance of the storm. These steps were prudent in light of the blizzard warning, but they later looked foolish as warnings were downgraded significantly.
To weather forecasters though, the eggs on their faces (or trolls on their social media accounts) on Tuesday were a bit surprising.
“Warnings signs were written on the wall and I feel most meteorologists communicated the fine line that we’d be dancing along I-95,” said Eric Fisher, chief meteorologist for CBS4 Boston.
Fisher said the track of the mid-level low, which is one component of the complex storm system, was “always a big red flag for big snow.”
“That’s been well modeled for days now. Allows for some warmer and drier air to invade, and all your snow becomes warm-air advection dependent which can be scary from a forecast standpoint,” he said via a Twitter message.
Warm-air advection snow occurs out ahead of the storm, as comparatively warm and moist air floods in, barreling into a cold, dry airmass already in place, leading to snow. Except this time, the warm air advection overpowered the cold air mass, bringing every winter storm lover’s worst nightmare: sleet.
Typically, the heaviest snows occur through a different process, rather than the warm air advection phase of a snowstorm.
Bill Karins, a meteorologist for NBC in New York, said a weather balloon launched from New York’s Weather Service forecast office on Tuesday showed a “pretty narrow zone above 32 [degrees Fahrenheit] that the models missed” at lower levels of the atmosphere. This caused snow that was falling to melt once it hit this layer, before refreezing into ice pellets, also known as sleet. Karins also used a more technical way to describe the warm layer, calling it a “warm nose” at 850 to 900 millibars, which translates to about 5,000 feet above ground level.
Gary Szatkowski, a former meteorologist-in-charge of the Weather Service’s office in Philadelphia who gained fame for his prescient warnings of Hurricane Sandy, said meteorologists have a tendency to over-analyze the bad calls and under-analyze the good ones.
“I think we always look for what went wrong in situations like this because we feel that is part of being a good meteorologist (and it is),” he said. Szatkowski, now retired but still an active meteorologist on social media, said the Weather Service’s probabilistic snowfall forecast maps have been successful in many ways.
“I’m a big fan of the probabilistic approach, including the low, likely, most NWS snowfall maps,” he said via Twitter message. “I saw how much the emergency manager community liked to know both the ‘best’ forecast and the ‘how bad it could get’ forecast. And they tend to be popular on my Twitter feed with the general public. The amount they are misused has been relatively small.”
Many of the computer models, though, hinted that the surface low pressure area would track along the coast, rather than offshore, which had a higher likelihood of bringing above freezing air into New York and Philadelphia, among other areas.
“On the topic of providing a range of numbers, that was clearly going on in the Philadelphia TV market among the broadcast meteorologists,” Szatkowski said.
To some extent, weather consumers may have a tendency to go forecast shopping to find the one that best aligns with their hope for the storm. In other words, snow-lovers latch on to the upper-limits, while snow-haters look at the bottom figures.
“Bottom-line is the broadcast meteorologists in a particular city don’t move in lockstep with each other, or with the NWS. So even if the NWS only provided one number, people will still get a multitude of numbers,” Szatkowski said.
“And I don’t view that as a bad thing. If we’re serious about communicating forecast uncertainty, people hearing somewhat different numbers can send that message. And people remember past events. They know that on a few occasions we absolutely nail it, on many occasions we are quite close, and on some occasions we bust.”
“And people judge by what they experience locally. There will be some to the northwest of the major metropolitan areas that will chalk this up as a major forecast success. Obviously many won’t, nor would I. This event will be looked at, talked about and studied,” he said.
“And like the cornerback in football who gets burned for a touchdown, you don’t forget, but you don’t let it haunt you. They have to be ready for the next play. We have to be ready for the next storm.”