Why Philly storm forecasts have been conflicting and constantly changing this winter

By slicing worldwide aircraft traffic in half and thus depriving computer models of valuable weather reconnaissance, the pandemic might be having at least a subtle effect on the quality of forecasts, in the view of some of the nation’s meteorologists.

But the coronavirus clearly has had no effect on the storm traffic across the United States, and forecasters and other experts agree that upper-air rush hour is at the core of the models’ erratic behavior.

If it seems storm threats are popping up every few days, they are. If it appears that computer models are seeing disruptive storms, losing them, then bringing them back, they are.

“Models are always ‘jumpy’ in the longer lead times,” say, a week or more in the future, said Steven Decker, an atmospheric scientist at Rutgers University. “What seems to be the case here is that the jumpiness is persisting into the shorter lead times.”

This weekend is providing yet another case study. What the models earlier in the week had seen as a potentially significant snowfall devolved to a dangerous Valentine’s Day ice storm threat, and by Friday, to something troublesome but less threatening — perhaps some unromantic sleet and light freezing rain.

» READ MORE: Icy conditions are possible Saturday night into Sunday in Philly area

The humans who get paid to interpret the models and take the heat for forecast misfires say this has been an especially frustrating period.

“You’d like to see some consistency,” said Valerie Meola, a meteorologist at the National Weather Service Office in Mount Holly.

That’s looking unlikely. Another adventurous weather week looms — virtually, and perhaps in reality, with wintry mess possible Tuesday and again later in the week.

Not that the public cares much about computer food fights, but the guidance preceding Thursday’s snowfall of 2 to 4 inches in most places varied from nuisance to disruption.

The American forecast model run on Tuesday morning, which suggested 2 or so inches of snow in Philadelphia, was superior to the 4 to 6 inches that showed up in the output Wednesday evening, right before the snow started. In the analysis assembled by former weather service meteorologist Tony Gigi, the vaunted European model runs also were flakey. The Monday evening forecast was better than Wednesday evening’s.

And take the case of last Sunday’s assault of the supersized flakes. Five days earlier, the models were seeing the kind of coastal storm notorious for generating major snowfalls. The threat evaporated Wednesday.

At lunchtime last Thursday, the weather service said, “Sunday looks to be another quiet day.” Three hours later it reported, “The model guidance has generally made an abrupt shift.”

The forecast amounts went up, down, then up again, and the improbable outcome was all of the above, depending on where you spent Sunday: Less than 2 inches fell officially in Philadelphia, but 10 in areas to the north and west.

The problem, said Decker, essentially is speed. Things are moving so fast, “disturbances over the Pacific get to New Jersey faster, so there is less time for observations over the continent to have an impact on the model forecast.”

In this pattern, he added, the data from aircraft would be especially helpful in spying on storms before they enter North America. “A lack of trans-Pacific flights could certainly be having an especially large impact,” he said.

Studies have suggested that flight reductions would have impacts. One three-year analysis published in 2017 found that flight observations contributed to a 15% to 30% improvement in short-term forecasts.

The National Weather Service, however, says the missing data are all but a nonissue, that while the models are getting only about half the pre-COVID-19 aircraft observations, they are receiving them from the passenger flights still operating, cargo planes, and other sources. ”In fact, 85% of the observations that feed our global weather forecast model comes from polar-orbiting satellites,” said spokesperson Lauren Gaches.

Cliff Mass, an atmospheric scientist at the University of Washington, said that any effects from missing aircraft data would be “very small.”

But meteorologists such as Meola and John Feerick say the missing plane data could be having impacts. Gigi said that what’s missing could be mightily important. Gaps can have huge consequences to forecast outcomes.

Global models attempt to generate three-dimensional forecasts over the entire 200 million or so square miles of the Earth’s surface, but the margins of error in the Washington-to-Boston corridor are tight — 50 million people inhabit less than 0.05% of the Earth’s surface.

Philadelphia, with one foot on the coastal plain and the other on the elevated “fall line,” often finds itself on the winter-storm tightrope, where a few miles’ deviation in a storm’s path can make all the difference in snow amounts.

» READ MORE: If you think the weather at your home is colder and snowier than Philly’s official measurements, you’re probably right

Gaches said that whatever problems computer models may be having, the public does have an important backstop, something called the meteorologist.

“You can’t just look at the models,” said Paul Walker, a senior meteorologist at AccuWeather Inc. And once upon a time, there were no models to look at.

“Back in the old days you were looking at the upper-air charts, the surface charts,” he said. Today, the computers do the drudge work, and the forecasters, or anyone else, can look at them on their computer screens with a couple of clicks.

The worldwide observation networks attempt to supply models with the “initial condition” of the atmosphere, then, using the laws of physics, they calculate how it might change. Even tiny errors propagate and worsen with time, and it’s never wise to bet against chaos.

The United States runs a variety of short-term and mid-range forecast models, and forecasters consult other models run by the European Community, the United Kingdom, Canada, and elsewhere. In fact, meteorologists sometimes complain of “over-modeling.”

» READ MORE: So much has changed, but the magic and mystery of snow endure | Book excerpt

The uncertainty appears to be more stubborn than the snow cover, and this week is shaping up to be a clinic in computer disputation as the region is situated between a mass of very cold air and warm air to the south that aims to fast-forward the seasons.

Mass, the atmospheric scientist, observed that while the global temperature last year was about a degree centigrade above 20th-century averages, “winter is not going away.”