Last night, the models seemed to converging towards a snowy solution. It looked like there was a decent chance of meeting winter storm watch criteria. However, today’s runs paint a different picture as there continues to be model chaos. Even the question of precipitation type has not been completely resolved.
Why is this forecast so hard? It turns out, there are many layers of complexities involved.
The overriding unresolved question is how much snow will the storm produce and, if it does, how much will stick. Both these questions are very much up in the air as very small changes in the storm track can make huge differences in the amount of precipitation it produces.
The precipitation gradient (the difference in the amount of precipitation within the band as you move across the band) varies greatly over a relatively short distance when moving from north to south. For that reason, the different model forecasts have differed significantly. Also, with marginal temperatures, a period of light snow might not accumulate at all. That certainly would apply to this morning’s GFS which would start as rain and probably end a period of slushy wet snow.
A period of moderate to heavy snow like the NAM is predicting is needed for the storm to produce significant accumulations. Right now I think that is the least likely scenario.
Below I’ve attached the Short Range Ensemble Forecast System diagram from this morning showing its various members forecasts of precipitation amount and type. Remember the various SREF members are run by tweaking the initial condition and physics to try to get a feel for how sensitive the system is to slight mistakes in analysis or way that the physics are represented in the models.
I’ve also superimposed where the various runs of the different models fit in with the ensemble members showing that the differences in the ensemble members do a pretty good job of displaying the spread of the various operational models last night into this morning. All but two of the members show all snow solutions.
Some like the various GFS and NAM solutions would offer accumulating snows that would dwarf any we’ve seen so far this year in and around the city. Others like last night’s GGEM and European probably would yield little accumulating snow.
The SREF data for Reagan National, Baltimore-Washington and Dulles Airports are given below.
BWI: mean snow 0.25”, max 0.8”, min 0”
DCA: mean snow 0.44”, max 1.03”, min 0”
IAD: mean snow 0.4”, max 1.01”, min 0”
The mean of most of the models, not just SREF members, looks to be around 0.40” - which could be 1-3” of snow or so.
Today’s European model which is not shown on the plume diagram, suppresses the storm so far south it produces no snow at all across our area. The very tight precipitation gradient makes this storm a bust waiting to happen. Slight shifts in the track and intensity of the storm can lead to huge changes in the amount of snow the area might receive.
It’s still too early to make any definitive call on the storm though I still think it probably will not be a major storm. There are still too many things that can go wrong. More likely would be a light to moderate snow storm, a light rain to snow storm with little or no accumulation, or even a complete miss.
Earlier this winter I discussed why forecasts of snowstorms sometimes bust. Two of the factors were discussed that are now combining to make any forecast of the storm track and evolution unusually difficult for this time range. Models often have trouble handling the interactions between multiple streams of flow especially if there are also a number of embedded upper level impulses.
During La Nina years, such troubles seem to happen even more often than in other years (at least that is my perception). The various model runs are handling the interaction between the northern and southern stream quite bit differently from one run to another.
On the 500 mb maps below, I’ve annotated two blue lines where the two stream of flow meet and appear to converge towards each other. Note that this confluence zone on the GFS (left image) is much farther south than on the NAM (right image) forecast. Areas of confluence such as this are usually associated with rising rather than falling pressure. This building of pressure helps the force the approaching surface low on a more eastward than northward track especially on this morning’s European model solution. This confluence is also one of the reasons this storm is not likely to be true Nor’easter that tracks up the coast.
Again, imagine that the winds are blowing parallel to the blue lines on the maps. Note how much farther north the winds from the southwest extend on the NAM versus GFS. This difference and northward extension of the southwesterly flow allows the southern stream moisture to come farther north on the NAM than on the flatter looking GFS.
All the light precipitation model runs have a solution similar to that of the GFS with some (the Euro runs) having the southern stream moisture even more suppressed. The wetter runs look more like this morning’s NAM.
The other problem maker is how the models handle and resolve convection, especially when there is such a juicy southern stream system. Convection in the atmosphere occurs at too small of a scale to be handled explicitly (handled precisely by equations) in the models. Therefore, the convective processes need to be approximated and those approximations can lead to problems if the heat released by the approximation is located in the wrong spot compared to where it is released by the real convection in the atmosphere.
At least once last year, Steve Tracton noted that the convective scheme in the model looked like it was causing a low pressure system to track too far east. How the various models handle the approximations of the convective processes can impact where the low tracks on the various model runs.
Given there are two complex streams of flow with a juicy, potentially convective laden southern stream storm, it’s not surprising that the models are still having problems deciding on the storm track and more importantly on how much snow might fall in the metro area.
As Jason alluded to earlier today, we need to get into the heavier band of precipitation so we get into enhanced precipitation rates. The NAM model forecasts fit that scenario. It forecasts a well-defined 850 low (left hand panel in the image below) to track to our south.
In the left panel in the image below, note how far south the freezing line is (the purple line) and how close the spacing of temperature lines is on map. Such close spacing indicates there is a strong front present. At the point of the red arrow, there is a east wind and to its north the winds are more northeasterly indicating convergence is taking place and implying there is strong frontogenesis taking place. That’s a fancy way of saying that the frontal boundary is strengthening which helps enhance lifting along it.
The 700 mb (right hand panel) shows both where the air is saturated (the dark green area) and the lifting is located along the boundary. The area within the bounded line near the -15 on the map is an area where there is strong lifting. Note that it is along and just north of where the tight temperature gradient is at 850 mb. Such strong frontogenesis is also often associated with a banded structure to the precipitation. Whether one of these heavier precipitation bands ends up over your house can determine whether you get a period of heavy snow or not. If one were located over D.C., it would likely produce a period of heavy accumulating snow.
At the end of the day, the difficulty with this forecast is that there is potential for banded heavy snow over our area if the NAM is right; or for us to get fringed or even missed if the European model is right.
To me, the heavier NAM forecast looks to be the outlier and is therefore least likely scenario. More likely scenarios would be a couple of inches of snow or a complete whiff with the storm staying south. However, with such a wet system having such a tight precipitation gradient, it’s not over until it’s over. Any of the scenarios discussed earlier by Jason are still possible.