eduggs

Not great wave spacing on the 18z GFS. The best baroclinicity is offshore. That's not a great look. But the trof axis is nice as things stand, so it wouldn't take much of a sharpening of the trof to kickstart a more tucked surface system. Inter-model ensembles are still supportive of something. Unfortunately everyone is eager for a slam dunk biggie.

Maybe over 12"? Should be lots of 1 foot+ totals in SNJ.

Can you squeak out another .1" to get to 9"?

Interesting post. By their very nature, ensemble means will be more expansive at the edges than reality. One or two outliers can blur and mask the true sharpness of the gradient. That's probably part of the issue. But I wouldn't expect that to be so noticable at 12hrs. I think the handling of mid-level dry air may be an issue on the EC as well as the GFS (and maybe others?).

If the beaches and eastern sections are going to accumulate a coating to an inch, it starts now. The bands have finally aligned SW to NE.

Agreed. Except the irony is the GFS was probably the best model with this from 5 days out. At least the best for all parameters except the only one that anybody looks at, QPF. I hope that minor issue gets fixed. It's unfortunate that so many people will learn the wrong lesson from this. People have to stop looking at QPF maps and especially 3rd party snow maps. Anybody who looked at the GFS 700mb charts was not surprised by the northward extent of the snow.

If that area gets nothing than it's a slight bust. Forecast there calls for 1-3." Yesterday's CMC, RGEM, EC etc gave that area a few inches. But the modeled cutoff has been very sharp and plenty of individual runs over the past few days (CMC, ICON, NAM, HRRR) gave nothing.

I expect flakes in the air in places like Sandy Hook, NJ and along the south shore of LI etc over the next few hours. Whether they accumulate is still unclear.

The issue is not surface dry air. It's 850mb - 700mb.

The SREFs were way too far NW yesterday, which they often are with coastal events. I don't recall the EPS showing an 8" mean and PHL. Maybe for one run with an extremely sharp cutoff? The EC did end up a little too far NW with precipitation... but I don't have access to the soundings to determine if the placement of features was too far NW or just the surface QPF.

Only the QPF parameter was off on the GFS. And that is the least accurate model parameter. If you simply used the 50% humidity at 700mb to mark the precipitation shield, it lined up very well with the EC and CMC/RGEM. In fact the EC might have been even further NW than the GFS. My appreciation for the GFS model has increased after this event, with the major caveat that its QPF parameter along the boundary with a cold, dry airmass cannot be trusted.

Doesn't really seem like an overachiever relative to modeling. Guidance was locked and loaded for that area (once it caught on) almost 2 days ago. But I guess it seems like an overachiever relative to expectations considering recent warmth and recent history.

That's why we have meteorologists to interpret model output. Otherwise we wouldn't need people to forecast the weather. The GFS was actually showing something very similar to other guidance once you took the dry air into consideration.

The GFS might be a great model with a very minor glitch in the algorithm used to model surface precipitation.

The obvious answer is the near saturated lower levels. But that seems too simple to be right. I don't recall the GFS having this problem to this degree in previous years. And this is not the first time it's been observed this winter.

That's not the GFS. That's a 3rd party interpretation of the GFS output. Plus the GFS is showing -38C at 750mb at the peak of the "precipitation" near NYC. I think we should look more at the model forecast soundings and mid-level charts and less at 3rd party snow maps.

The GFS was right a few days ago when it started strengthening and sharpening the southern stream wave and signaling a greater separation and storm threat. The model caught this feature earlier than other guidance. But the GFS also appears to have a problem with its QPF parameter that has been observed several times this winter season, and notably today. It shows precipitation at the surface that would actually sublimate or evaporate before reaching the ground. Simulated radar reflectivity maps look pretty good, but QPF is way off. The dry lower-mid level air was very well modeled with this storm. In fact it appears that, outside of the QPF parameter, the GFS did very well.

For radar reflectivity it's not bad.

It's 45/35 now and dropping off pretty quickly, especially the dew point. Plenty of time to cool down.

We have a threat tomorrow. A small change in the height field would have delivered us a major snowstorm. Correlations with indices are meaningful but weak. Weather at the local level is primarily dependent on random chance even though many correlations exist.

I suspect the sharp gradient in 700mb humidity represents approximately how far north the accumulating snow should actually be getting on the 18z GFS.

We're not necessarily going to see storms like tomorrow in the long-range anomaly charts. Averaging also skews and masks the details. But the details are everything with local weather.

Big difference between 1 and 3 inches. My guess is you'd be happy with 3 and a little disappointed with 1. I won't be satisfied with anything less than 4, but I would still be excited by a dusting. I expect nothing.

Watching the precipitation spreading into KY... So far it looks roughly well-modeled. I'd love to see the incipient shield punch right into Ohio. And ideally into southwest PA. The further north the initial SLP pushes, the further north along the coast it is likely to shift or redevelop to. Just wishcasting... Shift the entire synoptic setup 20 miles north and all of I-95 is likely looking at a moderate snowstorm.

Gradient! The EC goes from 1"-10" in 40 miles across NJ. Ouch!