eduggs

Agreed except we can't do an outlier test with so few data points. OK maybe you could with the HRRR. But the other models aren't run frequently enough. It's risky to disregard outliers when they very often end up being the most accurate runs of a suite.

On a microscale, no. We still can't pinpoint precipitation boundaries at the scale of individual towns and cities 3 days out. That's why I think it's best to take everything into account. Based on GFS and CMC trends several days ago, some thought good snows would get to PHL and NYC. The NAM was stubborn and resisted the trend. In the end, a blended compromise would have been right. It doesn't always work that way, but it's still almost certainly best to weigh all model outcomes at least a little.

I think it's significant that the NAM has decreased the sharpness and amplitude of the 500mb trof for several successive runs today. That's doesn't mean I can predict the future or change your mind if you disagree. It's just an opinion.

EPS didn't do so great for the Mid-Atlantic snowstorm from 48 hours out. It was 40 miles too far NW with accumulating snow. The NAM correctly kept the meaningful precipitation further south. All of the major models lead the pack for some events and on some runs.

From a forecasting perspective, discarding individual models runs - without obvious cause - will lead to forecast bias. From a hobbyist perspective, it will lead to false expectations and disappointment. We've seen it a million times. If it's an outlier run, is it unreliable or the first to sniff out a trend? Without knowing the final outcome we cannot know which model runs to consider or discard. That's why it's best to consider all the major model outcomes as plausible and deal in probabilities, not absolutes.

The surface isn't the problem. Weather isn't generated at the surface. Ls and Hs don't do anything - they have no causal effect. They form and evolve in response to what happens in the upper levels.

So now there's a bit of a discernable trend with the NAM. Wonder if the rest of 0z will confirm or deny.

It's definitely relevant. You can't just discard major model runs. Every "bad" run makes a "good" outcome a little less likely. But that doesn't mean it can't come storming back tomorrow... just that it's slightly less likely.

They had a massive push of GOM moisture out ahead of a robust and sharpening southern stream wave. The Thursday night event will not have that. We'll have to do more with less. Ratios should be much better, but we shouldn't be expecting .3" liquid hourly QPF totals this go around.

Some people are thinking whiff or bomb, but a light, plowable event is another plausible option. The initial mid-level fronto could setup fairly far NW and slide through most of the NE even if the the trof never sharpens up and the coastal misses mostly offshore.

1. Smaller events are usually shorter events and that means you might not be able to enjoy them due to work or sleep. 2. In places that don't get snow events very often, there can be a feeling like you have to cash in on every chance that you get because you can't count or another to look forward to.

That actually sounds like a pretty good handle. No discernable trend - take the average. It's not really possible for them to show exactly the same surface weather 3 days out.

Yesterday's mid-Atlantic snowstorm had a ton of GOM moisture out ahead of the trof just pumping northward into a wall of cold air. We won't really have that luxury with this one. Here we need to rely on the dynamics of a rapidly deepening SLP and moist easterly inflow. The main precipitation shield is not likely to extend very far NW unless this really wraps up.

Nope, you're not whacked. The previous run was better. Little bit of wishcasting going on.

Maybe it goes negative earlier in time, but it doesn't look like it does so earlier in geographic location. I don't care if it does negative tomorrow if it does so in Nova Scotia. I don't hate the look, I just liked previous runs much better.

We'll happily take our 2-6" with decent ratios, but that NAM run is not what most are looking for. Also no sign of the kind of trend we're looking for. Flatter, less wave spacing, later developing, and SLP further offshore.

Maybe some were fooled by the earlier arrival? The 18z looked pretty clearly less impressive aloft to me.

We get mostly light snow on the 18z NAM. Widespread couple of inches. Probably decent ratios. The trof is less sharp, late developing mid-level lows, SLP offshore.

It's unfortunate that twitter is obsessed with QPF maps. They make a lot of people who should know better seem meteorologically ignorant. And they make the general public confused and angry.

It's not you. - Nova Scotia

You don't even have to look at the surface to see that guidance has shifted everything east. Look at H5 - best PVA is SE. The surface is just a reflection of the upper level divergence/convergence. Good thing is that if this sharpens up 10 %, the SLP ends up 100 miles NW. There will be plenty more changes.

Still lots of options on the table. But it seems like the upper ceiling decreases with each model run that doesn't increase the amplitude of the LW trof. Still warning snows potential almost everywhere.

We're dealing with a little bit of a transition from the initial inland surface reflection and associated precipitation in PA to the developing coastal low. It's possible I-95 or just inland sees a bit of a precipitation minimum as the transition skips over this area and refocuses closer to the coastal. The precipitation shield as modeled looks to be shrinking and tightening as the SLP winds up.

Intense lift for a couple hours.

ORH east would get hit pretty good for 4-6 hrs on that run.