Ian Kessler-Gowell walks up a hill in the snow on Nov. 15, 2018, in Alexandria. This was the first snow of the season. (Matt McClain/The Washington Post)

The Washington region sits right on the dividing line of a potential winter storm coming from the south Sunday. It could witness a little snow, a lot of snow or no snow Sunday into Monday, depending on the exact track of the storm.

Computer models continue to present a wide range of possibilities for the storm, as they did Monday. They have pushed back the arrival time of any flakes, from Saturday night to most likely sometime during the day Sunday.

The storm will start to take shape in southern Texas on Friday and will then gain strength as it tracks across the South through Saturday. By Saturday night, precipitation will start spreading into the Carolinas and perhaps southwest Virginia. On Sunday, it should spread farther northeast through the Mid-Atlantic, but the question is just how far north.

Washington remains near the edge of where significant snow starts and stops on many of the models. This is a storm in which areas south and southwest of the Washington region have increasing chances for snowfall in significant amounts.

Roanoke, Blacksburg, Lynchburg and, possibly, Charlottesville and Richmond may witness a heavy snowfall. But, even in these areas, it is too soon to discuss specific amounts, given uncertainties in forecasts five days before the storm is set to arrive. Small model shifts could substantially reduce or increase snow potential even in these areas.

While the average of all models places the D.C. region right on the edge of an accumulating snow event, some models predict little or no snow for the region, and some predict quite a lot.

Here is what the current suite of models project:

  • Operational American (GFS) model: Light snow and flurries from the District south during the day Sunday. Moderate snow in southwest Virginia and moderate snow in central Virginia.

Operational American model forecast of snow mainly south of Washington on Sunday afternoon and evening.
  • Experimental American (GFS-FV3) model: Heavy snow for the entire region Sunday into early Monday. Major accumulations.

Experimental version of the American model (GFS-FV3) simulates heavy snowfall in the Washington region Sunday afternoon and evening. (TropicalTidBits.com)
  • Canadian model: Light to moderate snow falls from Washington south Sunday into Monday with some accumulation. Only light snow brushes northern areas with little or no accumulation. Moderate to heavy snow affects southwest and south central Virginia.

Canadian model shows some light snow in the Washington region Sunday afternoon and evening.
  • European model: Light snow mainly stays south of the Washington region with no accumulation. Substantial snowfall mostly confined to southwest Virginia.

European model shows snow well south of the Washington region Sunday night. (WeatherBell.com)

The models summarized above are considered the main or operational simulations in larger modeling systems that contain dozens of projections that also present a wide range of snowfall possibilities.

Even though the operational American model predicts little snow for the Washington region, 11 out of its 20 simulations in its larger modeling system indicate snow, and seven of the 20 predict a substantial snowstorm (of at least a few inches). A few project double-digit totals, but nine of 20 predict little or no snow.

Snowfall simulations through Monday from American (GFS) modeling system. (WeatherBell.com)

The European modeling system also expresses significant snow as a possibility. Twenty percent of its 50 simulations project at least six inches of snow in Washington.

Two related problems make this storm a difficult one for the models to predict.

1. They are having difficulty predicting how quickly flow from the northwest across the northeast will relax and shift east. The models that relax this flow give the southern storm enough room to move north and drop substantial snow. Those that keep the northwest flow locked in place shove the storm out to sea — and we just get skirted or missed.

2. These type of systems often have a sharp cutoff between where accumulating snow is heavy enough to cause problems and where the snow starts and stops altogether. Allow the northern edge of this cutoff zone to migrate 100 miles north of the predicted position and more substantial snow could fall in zones where models predict little.

In the past, we’ve seen quite a few cases in which model forecasts creep the dividing line between substantial snow and little snow northward with time. So this is something we will carefully watch.

This remains a very uncertain forecast for the Washington region, and we may not have a solid sense for another one to three days as to where the cutoff line will set up.