Model behavior: Computers saw blizzard coming
They all nailed it this time.
Most of the computer models meteorologists use to predict weather picked up on the Blizzard of 2016 several days in advance.
"I do not remember seeing four or five modeling systems having this much consistency" up to a week before the storm, National Weather Service director Louis Uccellini said at a news conference Thursday.
Models rarely agree so consistently on a large storm that's several days away, said Weather Underground meteorologist Robert Henson.
Epic blizzard with heavy snow begins to move up East Coast
"The major global models have been in general agreement all week on a major storm in the mid-Atlantic," Henson said, referring to the American Global Forecast System (GFS) and the European Centre for Medium-Range Weather Forecasts, two of the world's top weather models.
He said that the GFS, in particular, insisted for days that snowfall would be heavy in the D.C. area.
The two models have made big news in recent years. The European model correctly forecast the track of Hurricane Sandy several days in advance in 2012, while the GFS did not. But the opposite was true for a blizzard roughly one year ago, when the GFS correctly predicted the worst of the storm would miss New York City.
Other commonly used models include the Canadian Model, the North American Mesoscale Model (NAM) and the UKMET (The United Kingdom Met Office model).
Fliers face chaos as blizzard slams East Coast
Debate continues in the weather world as to which is the better model, the GFS or the European, though the head of the National Oceanic and Atmospheric Administration Kathryn Sullivan recently said it's a foolish game to compare the two: both are excellent tools forecasters use to make predictions.
But even with the excellent large-scale predictions from the models, the actual details of the forecasts remain in the hands of the human meteorologists, WeatherBell meteorologist Ryan Maue said.
Those details such as how heavy the snow will be, precipitation type — snow, rain or sleet — and rain/snow lines depend on smaller scale features that may not be picked up by the models but could drastically change the storm's impact.
"The humans need to earn their keep and provide value-added expertise and experience-based guidance," Maue said.