Following a snowy run of the European model last night, the Facebook and Twitter universes are ablaze with graphics of the model runs showing wintry weather in the Washington area next week. Expect models and forecasts to continue jumping around over the next several days as they try to lock into a solution in a very complex and difficult weather pattern.

If you’re not a weather geek and just want the bottom line, here it is: It’s likely to turn colder during the early to middle part of next week and there is a chance a storm affects us mid-to-late week. The storm could bring rain, snow, or a mixture, or could miss us. Right now, we’re leaning towards a cold rain, but with increasing chances of snow into the mountains north and west.

Now let’s get into the weeds.

Technical discussion

When dealing with a potential storm forecast for the day 6, 7, or 8, it is important to remember that model forecasts at that time range do not show much skill and tend to generate snowstorms way more often than they occur.

Yesterday Jason noted two features on the weather maps that will play important roles in determining our snow chances.

1) The strength and position of the strong surface high that is forecast to be located to our northwest. This time of year, an Arctic high is an important component needed to get meaningful snow. The handful of model simulations that offer us snow have the high located in such a way to funnel cold air into the D.C. area. While all the models are showing a strong high there is still uncertainty about how quickly the cold air will get east of the mountains.

2) The strength and location of the upper level low pressure system. This time of year the upper level vorticity (spin) center needs to pass just to our south in order to keep winds from the north across the region, holding in the cold air. This time of year any hint of an easterly component to the wind will kill any snow chances since ocean temperatures are still quite high.

Models of upper level low pressure (or trough/jet stream)

This morning's 144 hr GFS ensemble model run diagram valid 7 a.m. Wednesday morning illustrates the uncertainty of the model handling of the upper system. This morning’s 144 hr GFS ensemble model run
diagram valid 7 a.m. Wednesday morning
illustrates the uncertainty of the model handling of the upper

The strength and evolution of this feature differs significantly from model run to model run. The figure to the right is a spaghetti diagram from the various ensemble members of the GFS model and the operational model run from overnight. Such products attempt to help quantify the uncertainty of a forecast.

When the ensemble members are closely clustered, all the green lines would be packed together. Such packing of the lines usually signifies that pattern is more predictable than when there is a lot of spread between the lines.

The ensemble members forecast of this system show significant differences in the handling of the upper system. Note how much spread there is to the various green lines and how they differ from the thick white line that dips south of our area. That white line represents the operational run of the GFS model. Note that some ensemble members have the dip in the lines well to our west while others are flatter. The ensembles are indicating that there is little hope of getting the all important details of the forecast right this early in the game.

The bulk of the ensemble members predict a dip in the jet somewhere east of the Mississippi River but the details differ. And, in this case, the details really matter. Close the low off too early and too far north and west and any precipitation would likely be rain and perhaps slower to arrive. Shear it quickly eastward and any precipitation is likely light and might very well traverse our area before the cold air reaches us, or the wave might be so flat that the precipitation stays to our south.

Model comparisons

Last night’s European operational model was perhaps the most aggressive in predicting snow across the region. However this morning’s GFS backs off giving us only light precipitation. While a handful of European ensemble members are in same general camp as the operational model, the majority of its members are more progressive and weaker with the upper center, with less cold air and less precipitation.

Note how much flatter the 168 hr European ensemble mean trough (left) is compared to the last night’s operational European run (right).

(Left) The European ensemble mean (average) upper level trough (Right) The operational European model upper level trough

The models handling of upper trough tasked with simulating our precipitation and snow threat vary significantly. The European ensemble mean coupled with the wide variation found in the GFS ensemble members suggest that the snowy European operational model solution is probably an outlier. Today’s European run closes off the low so far west that the bulk of the precipitation with it would be rain.

The take away from the ensembles and today’s wrapped up European model is caveat emptor. Don’t trust any one model solution. Uncertainty chaos abounds.

So what are our snow probabilities?

Last night’s GFS ensemble (GEFS) based the 24 hour probabilities place the probability of .50” or greater precipitation over our area from 7 a.m. Wednesday morning to 7 a.m. Thursday has risen from 30 to 50 percent (see left panel below). That is the period when the ensembles forecast the highest probability of 0.50” or greater precipitation. During that period, the peak probability of snow is in the 20 to 40 percent range in D.C. (probably closer to 20 percent than 40) with a 40 to 60 percent probability towards the West Virginia (right panel below).

The ensemble runs are showing uncertainty even about us getting significant precipitation and are definitely not overly excited about our chances of snow. That’s not surprising given we’re still a week away from the potential event with the ensemble members showing lots of spread.

Even with the slightly more snowy look to the ensemble run from last night, the probability of meaningful snow is about 10 percent in the D.C. area. The probability of the I-95 corridor seeing some non-sticking flakes has probably risen to 20 percent with somewhat higher probabilities in the western suburbs. Snowfall probabilities in the mountains are considerably higher.

(Left) The GFS ensemble based probabilities of 0.5" or more liquid precipitation by 7 a.m. Thursday (Right) the highest snow probability for the event.
(Left) The GFS ensemble based probabilities of 0.5″ or more liquid precipitation by 7 a.m. Thursday (Right) the highest snow probability for the event.

In his extended forecast discussion, the National Weather Service’s Jim Cisco notes:

there is simply too much uncertainty to go into any meaningful detail about sensible weather impacts during the second half of the forecast, but of note is the handful of deterministic solutions and ensemble members of both the GEF and ECENS (the NCEP and European Center Ensembles) that indicate the first snow…for lowland areas including the Mid Atlantic Region

His thinking is right on, it is way to early to make any definitive statements concerning the storm. Any of the ensemble solutions remain in play.

An afternoon forecast discussion from the National Weather Service smartly adds:


A betting man would bet against any significant snow based on the uncertainty expressed by the ensemble runs and based on time of year (climatology). It’s still really early in the season. Plus having a negative PNA pattern and a positive AO and NAO is usually not a positive for getting a snowstorm even during the heart of winter.

These negatives can be overcome but only if the storm evolves and tracks perfectly as it moves towards and across our area. How often has that happened in the past three years?