Several days ago, European forecast models, and then the U.S. models, foresaw the development of a powerful nor'easter this week that would have widespread impacts in one of the nation's most populous regions.
Predictably, and with good reason, a storm of anxiety -- mixed with hype -- developed and intensified.
As it turned out, the models definitely were on to something. Overall, this has been one impressive storm, at least meteorologically.
Once again, however, key details have been off the mark, and that has made a huge difference in terms of impacts.
For example, the models, and thus the forecasts, were too warm with the upper-air temperatures. That put the damper on Tuesday's heavy-snow forecasts for the Poconos and other elevated areas.
The models overstated precipitation amounts for inland areas in the Midatlantic, including Philly and Washington, noted Bob Oravec, senior forecaster at the U.S. Weather Prediction Center.
Officially, only 0.66 inches of rain landed at Philadelphia International Airport; so much for the flood watches.
And of fortuitous significance to the Shore, the storm moved faster than expected, and that mitigated coastal-flooding impacts.
Winds circulate counterclockwise around centers of low pressure, thus when storms are coming up the coast, areas to the north experience onshore winds from the east.
As the storm passes, the winds swing around and blow from the north. On Tuesday, the winds shifted before high tide at the Shore, and that probably cut down on the flooding.
Models often have a hard time getting a handle on the speed of storms traveling northeast, said Oravec.
"If there was general bias in the numerical models, they are slow to move systems," he said.
What is the problem? It's nothing personal.
"The atmosphere never has a double about what it's going to do," said Oravec. "The atmosphere just reacts to the laws of physics."
Oravec says the essence of the forecast problem – common to all models -- is the quality of worldwide observations that are fed into computers to capture the "initial condition" of the atmosphere.
Models then used mathematical equations to project what the atmosphere will look like six hours in advance, then six hours later, on out to days in advance.
Any errors tend to grow with time, and given the imprecision of observations, errors are inevitable. Vast expanses of the surface and upper atmosphere are poorly monitored.
Of course, all weather ultimately is local, and if a short-term forecast is perceived as "wrong," the public could care less whether a computer model was able to foresee a storm before it was a concept.
Looking ahead, our own computer models show that the problem of nailing details will reoccur multiple times before the winter is over.