DT Wxrisk
7 min readJan 12, 2022


Late Monday night, as many of you know, I made a post concerning the historic and stunning 0z regular or operational GFS model which showed an enormous snowstorm for all of Tennessee, North Carolina, Virginia, Maryland, Delaware, with a historic amount of snowfall. The model actually had kicked out areas of 12 to 28 inches of snow including areas such as Richmond and Greensboro into Virginia Piedmont and western North Carolina. It also had significant snow in DC/ Philly, NJ/ NYC and New England of 8 to 18 inches.

But as I also pointed out such an extreme solution did NOT match the GFS ensemble. You will hear me use that word ENSEMBLE a lot over the next two days and it is REALLY important even if you are just a regular weather person to understand what this means with respect to forecasting weather events Beyond 72 hours.

In the weather business when a meteorologist talks about a weather model ENSEMBLE, we are referring to a particular type of technique or process where a weather model is run 20 to 30 times from the starting point or initialization. Individual or deterministic/ operational weather models are sometimes very good and sometimes they are bad -especially beyond 4 days. That is why it is more important to focus on Trends and the model ensembles as opposed to the operational or deterministic runs. You get around that problem by looking at the ensemble mean — mean or an average of the 20 or 30 possible solutions. The idea is that you get a better consensus of what is most likely to happen and you are not thrown off by having a bad model run or a bad Model cycle.

CLEAR? Ya got that?

The average of 20 or 30 models taken as a mean is far more likely to be correct or close to being correct then one or two models Solutions.


Starting early this morning at 6 am/ the 6Z operational GFS took the main LOW pressure area inland passing it across central North Carolina, across Richmond, then up along the big cities in bigger I-95. That continued on the operational/ Regular 12z GFS and with the new 18z GFS. This received a lot of play on the regular TV stations and various websites and of course it affected the forecast on your smartphone weather app.

BUT … in every one of these instances the operational run of the GFS model did not match its ensemble. This is a big warning flag that all experienced meteorologists know. The ensemble mean and the operational run should show the same thing when they are in agreement, it tells the meteorologist that the models have a good handle on what is going on in the forecast can be made with more confidence.

In this instance the discrepancy is massive. This is also the case with the midday 12z operational or regular European model as it also brings the LOW
inland across eastern North Carolina, eastern Virginia, the Delmarva, into New Jersey and across NYC and into eastern New England. This track would bring the snow changing to sleet then rain in the big cities of I-95 from Richmond to Boston.

And this was also the case for the12z /midday Canadian model

An all 3 models… the GFS, the Canadian, and European model solutions would have the heaviest snow in the western half of North Carolina, all of southwest Virginia, the Shenandoah Valley, the Virginia Piedmont, western and central Maryland, eastern Pennsylvania, northwest New Jersey, interior southeast New York, and central & northern New England. The big cities of I-95 from Boston to Richmond would see a changeover from snow to ice to rain *** IF THESE MODELS ARE CORRECT***.

But again in every instance, NONE he operational models with the Inland track match the ensembles which have a track of the LOW further east which is colder and keeps everybody all snow.

This image shows the track of the12Z midday Tuesday European model. Notice that the LOW pressure area is on the coast or Inland which is why there is a changeover from snow to rain in the big cities of Interstate 95.

But here we can compare that track to the European ensemble. Notice it is significantly different. All those little red L on the map show possible positions of the LOW pressure area on Sunday and Sunday night January 16. As w e can see, the vast majority of them are significantly further to the EAST. This means more snow, much less rain, and heavy possible accumulations from western North Carolina into central Virginia, and all the way to Boston.

So the next question is why? WHY don’t the regular Canadian, GFS, and European models match their own and sandals.? What’s causing this huge discrepancy?

This image explains exactly what is going on. Please take a look at this map.

Now here is the 18z operational GFS model. Again, this run of the GFS model takes the main LOW well inland so everybody goes over to rain from Central North Carolina to Boston.

But once more the 18z GFS ensemble is significantly further east and much colder. It keeps the entire I-95 Corridor as all snow with several inches and it even has snow into the Raleigh metro area.

Finally, here is the 18 Z European ensemble. Notice that the LOW pressure area is significantly further to the east it’s colder and it supports


It is of course quite possible that todays trend showing on the regular operational models — where the LOW comes INLAND and forcing the rain/snow line into or just past Interstate 95 and into Interstate 85 in North Carolina, could be 100% correct. That is entirely possible.

But the vast majority of the data does not show that. It still shows a major or significant snowstorm, depending on your location, for a good portion of the Mid-Atlantic and New England regions.

The KEY remains that piece of energy coming down from Manitoba Canada into the Upper Midwest. As I stated before, the midday models are taking that energy in a different direction from what it was showing yesterday. That change in how some of these 12z models handle that energy in the jet stream in the Upper Midwest is why the models today changed and have the snow changing to rain on the coast in the Mid-Atlantic and New England regions.

I am not at all convinced that the operational models of the European the GFS the Canadian model depiction of this energy coming down from Manitoba into Missouri in Illinois is correct. It is quite possible that this change is bogus and that further models in the future tonight or on Wednesday could change the track of its energy coming down from Minnesota into the Midwest

Interior portions of the Mid-Atlantic and New England should expect to see a major snowstorm. The probability is high of 6+ inches of snow WEST of Interstate 85 in North Carolina, and in southwest Virginia, all of the Shenandoah Valley, the eastern half of West Virginia, western and central Maryland, central and eastern Pennsylvania, interior southeast New York, and interior portions of New England. In fact, some of these areas have a pretty good chance of receiving 12 inches of snow.

In the critical I-95 corridor region, there are still a wide number of solutions on the table. The atmosphere is definitely cold enough to support all snow at the start of the event in all of these areas even into the Raleigh-Durham metro and interior southeast Virginia. area. We could see something like a 3–6 inch snowfall band in the I-95 Corridor before it goes over to sleet and rain or it could stay all snow or mostly snow was just a short interval of sleet.

IF he model depiction of this event being a long duration is correct and IF…IF it takes a more easily tracked like the GFS European ensembles are I95 that will see a foot of snow.

And yes there is plenty more activity coming up after this event



DT Wxrisk

Meteorologist ... Atheist.. Dyslexic ..Baseball.. Fat tail distributions ..Good Judgement Projection… Black Swans/ Taleb …Choas / non Linear Dynamics… ENTP