eduggs

The hi-res Canadian is too sensitive to elevation in terms of generating QPF.

Another step down for the RDPS. The 18z is slightly SE and noticeably drier than previous wet runs. But the axis of snow is favorable. We seem to be converging on a model consensus for between 0.2 and 0.4 frozen QPF between approx. 06z and 15z Wed. from NW to E.

Yes the delayed onset of the "anafrontal" precip. compared to a guidance a day or two ago allows the atmosphere to cool on the NAM. That has eliminated the sleet threat for now (layers above 850 are warm prior to 12z) and also allows the surface to cool to 34F or below almost everywhere. Moderate snow will stick almost everywhere at 33F.

The snow maps might look a little better on the 18z NAM than the 12z for the immediate NYC area (due to a late burst) and possibly parts of NE, but overall the NAM was less robust in developing precip. on the NW side of the cold front and cut back QPF throughout the SE and MA. I was hoping to see a more expansive precip. shield on the cold side - even an incipient baroclinic leaf - but the NAM is moving away from that scenario. Sure a more robust solution might be warmer initially near the coast, but it would also likely increase the odds of a plowable event region wide.

It looks like the 18z NAM has a burst of snow along the coastal plane Wed. morning.

The 18z NAM continues the drier and SE trend.

This looks like an old map. Litchfield County looks suspiciously dry. I'm pretty sure the latest version from ALB shows 4-5" for that area.

850s are cold but the surface is still relatively warm (outside of elevated areas) and there is even an elevated warm layer. Soundings also show a low level unstable layer. There could also be a period of sleet. I really hope future model runs stop decreasing QPF. Heavier intensity precipitation will really help mix down the colder temps aloft. This could be a classic short duration, high intensity "surprise" snowfall. The low expectation events are sometimes the most fun when they break right.

There seems to be a slight trend to lessen QPF and possibly shift the axis a little east among the majority of guidance in recent runs. But it's hard to tell if it's just noise or not.

The 06 NAM backed off QPF and the 06z ICON has the band of precip. far SE with only minor accumulations for the coastal areas. But the 06z UK and RDPS were still very wet with (likely) moderate snow accumulations. The 12z NAM was again light on QPF. Typical slight but locally meaningful model disagreement. The radar and obs early Wed. morning off to our SSW will likely tell the tale.

The 0z RDPS is wet and cold - even better than 18z. It suggests 2-6" regionwide with some sleet to start. Gotta hedge lower at this point but confidence in a snow event is increasing.

Some of the biggest back then are respected red taggers now. Which is weird. Some have grown and learned, others are still . It seemed more fun back then with fewer models, fewer off-hour runs, lower model resolution, and no twitter.

The 18z RDPS (regional Canadian) shows between 0.3 and 0.6 frozen across most of the area. This is a similar depiction but maybe a notch wetter than the NAM and GFS. The GDPS is slightly drier and further east with the boundary. The 18z ICON is SE and dry (not enthused) and the 12z UK was pretty wet (QPF). I think the best chance for accumulating snow will be near or just after daybreak as surface temperatures finally approach freezing. This looks like a general coating to 3" or so, depending on location. But I also think someone has a shot at 4"+. It could be a surprisingly snowy Wednesday morning. Not a blockbuster but 2-3" in a few hours can create a very wintry appeal.

NE guys are spoiled and in pack-protecting mode. So naturally they are not particularly excited about a modest threat book ended by likely rainers. But for this forum, especially this time of year, this is a legitimately trackable threat. For many many years in this area, 4" was considered a solid, plowable snowfall. And there is a chance, albeit low, to hit that threshold somewhere in the region early Wednesday morning.

I think these setups along semi-stalled steep thermal gradients can be sneaky productive. They can yield bands of moderate snow training over the same area. More often than not they disappear beyond 48 hours out. But inside two days, from memory they often spread precip further NW than modeled (at least as modeled during the post 2000 period). Looks like short to moderate duration. And right now you have to favor areas with some elevation just inland - not necessarily far NW. The southern s/w will likely pump some more moisture into the deep south than expected, but how much of that translates up the coast is unclear. The southern s/w is dampening and not really involved as a local weather producer. It would be nice if it traversed eastward a little faster and we could buckle the 500mb flow slightly and get a bit more surface reflection along the coast. Of course depending on where the boundary sets up, that could halt the cold slightly too far NW of the coastal cities.

The ETA and NGM are gone. This isn't 2001 anymore. Other countries have supercomputers, observation networks, and talented developers. Models are constantly being developed. Unfortunately most Americans have have little care or awareness for what happens in the rest of the world.

I think the ICON is a really good model. Verification/ranking compared to the ECM and GFS would probably depend on the particular geophysical parameter and also region. It has performed well in Germany in terms of low level temperatures and about on par with the GFS in terms of precipitation forecasts (although FV3 has been poor lately). I think it's better than both the UK and the Canadian models, at least in Europe. Since it's a global model, it should perform well in the US, but I have not checked closely since I have been outside the US for a while.

The long-term temperature trend in winter is clearly increasing. If you plot a relatively short duration dataset that starts with a relatively warm period, you might get a near horizontal trend. But that's a bit of a statistical fluke. Most places are warming in winter and in summer. Of course there are significant fluctuations in this warming and the changes are inhomogeneously distributed spatially. High latitudes and high elevation areas are warming faster than low elevation, mid-latitudes.

Adding about 2 deg F to long-term temperature averages for monthly guesses beats just about all other prediction methods for long-term temp forecasting at most mid latitude locations. For mountainous and urban locations, the additive factor should be higher. For some continental locations that radiate well (important for daily mins), the factor is a little lower. Most forecasters are not scientists. Climatologists can model and predict increases in global average temperatures. But meteorologists cannot predict short-term regional forecast temperatures in the same way. It is well understood that weather does not equal climate. But because warming has been so significant in most locations, adding a constant factor actually does out-predict most regular forecasts in most places (on average). It does not work every month... just over the long term (like 12 consecutive months averaged).

Any reports from Putnam County? I'm far far away for this one unfortunately.

Its seems like the storms with the strongest mid-level winds tend to have weird localized temp gradients. Weaker storms with weak mid-levels tend to have a more typical elevation and latitudinally dependent temperature gradient.

Pretty wild that central Orange county is reporting 37-38F while southern Dutchess at river level is 32 -33F. Even Peekskill on the river in Westchester is 34.