Technical Discussion: What Could Some Models Be Missing?

I published my official forecast of 8-16 inches Monday, earlier than I usually issue one and earlier than most news stations released forecasted snow accumulations.  A lot has changed since then, but the forecast has generally remained the same, with the range being narrowed now to 8-14 inches since the low pressure is not going to be a classic deepening low pressure that is stacked through all levels of the atmosphere.  That doesn’t mean the storm won’t be strong.  In fact, I am still expecting the low pressure to be slightly closer to the coast and stronger than many weather models show.  I’m going to explain why here, and a lot of this goes back to my original ideas from my very lengthy blog post Monday when I explained why I expected models to trend towards the solution that they now currently have.  I stated that the ECMWF model was a little overboard but there would likely be a low pressure center of decent strength a little southeast of the ECMWF but more northwest and stronger than almost all other weather models had.  Most of this is materializing, though models continue to jump around, which isn’t really unusual.  The general trend ticking west has been noticed, though the GFS has remained steady which is worrisome.  This is why the 0z model runs are going to be crucial.

What do I expect with the 0z runs?  I expect another tick westward generally, and if not a significant tick westward then slightly more moisture than some models have.  In fact, my model of choice right now appears to be the RGEM, which is normally one of my favorite models.  I’ll go into that in a second.  First, I want to show my reasoning for why the GFS SHOULD trend towards a little more moisture with the storm.  The 18z run of the GEFS is here, with the top left frame being the operational GFS.  All these frames show is how much precipitation is supposed to fall in the storm.  Each ensemble is run under slightly different conditions, but if you look closely you see that most of the frames that are not the top left one have more precipitation than the top left one, at least specifically for Southwestern Connecticut.  Where the operational model seems to give SWCT around .45 inches of QPF, more members want to give it at least .5, some have as much as .7-.8.  Only one member of the 11 has noticeably less QPF for SWCT than the operational GFS and it is not by much at all, while multiple have significantly more QPF.  This is a signal that the operational GFS is likely too dry with the storm and is not printing out as much snow as it should be.  As one of the drier weather models only showing .45 inches of snow (ending up as 6-7 inches of snow in the cold, high ratio snow) this model is in the low end of predicted snowfall for the region, and you can see that its ensembles are signaling that it may be a little too dry.

Perhaps more impressive, though, is the consistency of the SREFs, the Short Range Ensemble Forecast members.  These are a number of high definition, short range models run together and then averaged.  I find they are really only useful inside of 48 hours, and we have finally reached that point.  They forecast the percent chance of SWCT reaching certain thresholds for snow, which can be seen here.  As can be seen, there’s a 100% chance of over 4 inches of snow, 65-70% chance of over 8 inches of snow, and around a 40% chance of over a foot of snow.  This fits in well with the going forecast.  The most interesting part is they actually trended a little drier on the last run (which makes sense as they typically lag behind NAM trends, meaning they may trend wetter at 21z).  As I recently tweeted, for Bridgeport they have a spread of 5-21 inches with a mean near 11, right in the heart of my forecast, and most are clustered from 7/8 inches to 14/15 inches with just a few outliers on either side.  It will be interesting to analyze the 21z run when it comes in just before 9 PM.

But why do I expect the low pressure to be slightly stronger/a hair west?  It has to do with the themes and trends of coastal storms the past few winters.  A common trend in the last 36 hours before they form is to deepen more rapidly than models form.  The reason why models seem to not have as good a handle on this storm is because of the complexity of upper level features interacting.  There will be a double-barreled low, but models have trouble specifying exactly where there will be the greatest low pressure falls.  This can be seen here, tomorrow at 7 PM on the 18z NAM.  In the top middle frame you see 4 red Ls, or pressure falls, inside what looks like a very broad low pressure center.  There is nothing really defined here yet.  The model struggles to then consolidate energy around a single low pressure center in the following frame, as seen here.  More than likely, with the 500mb trough that you can see on the top right tilting negative (bottom points slightly east) there will be enough energy for a single low pressure center to form quicker than some models want it to.  What this does is allow a slightly stronger gradient to set up and it can enhance precipitation across the area quicker.  We are talking a few hours difference, but it could make the difference of a few inches of snow, which is why I am sticking with my 8-14 inches of snow that is slightly higher than the National Weather Service.  0z data will be very telling, and if we see the weather models focus in on a single low pressure center slightly earlier due to the digging trough turning neutral then expect to see some other forecasts rise to the range that mine is at.  If not, the focus turns to “now-casting” and seeing what current trends are showing.  There remains a chance that I will have to lower my forecast slightly, maybe by a inch or two, but I feel quite confident where I am given the ratios and a rapidly strengthening low pressure center moving near the 40/70 benchmark.

So that is a quick summary of why I expect to see a slightly wetter model trend with the 0z guidance, and if not then we need to focus tomorrow on when the actual low pressure forms.  I expected this a few days ago to happen, and here we are today not much changing.  Again, there is certainly a bust potential with a storm as complex as this one, but I would say confidence is continually increasing in the forecast that I have put out here.  I will have another more general outlook in about an hour or hour and a half regarding school closures for Thursday and Friday, and I will be breaking down additional short range models with regards to snowfall timing.  Stay tuned.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>