OceanStWx

I mean sometimes we mess around and create storm total grids to see what we have in the forecast as a baseline. But the automated scripts can grab them and send them without someone physically pushing the button. That's why we have a "work" grid we can create at GYX. Those don't go anywhere, and I'm not sharing the image.

Capital H hate going with a deterministic snowfall forecast at this range. We know it's going to change, we've shown we have low skill getting amounts right this far out, and we just don't have a lot of data to build that forecast with. I have to think this accidentally slipped out to the winter page. We aren't even required to have QPF this far out, and WPC doesn't even provide the probability information to produce the rest of the graphics on the winter page.

It's happening more often, but I don't know for sure where to find reliable information on plans.

The sensitivity is highest tonight around CA and NV, so those raobs will be important.

Toaster bath for the chickens?

I don't think so. The NBM is pretty sophisticated, bias correcting a lot of variables at each individual grid point. But it can have limitations when ensemble make big swings or you have dramatic pattern changes. It also struggles with mesoscale features at longer ranges (obviously), but that is what the forecaster is for. Make the right adjustments vs messing around with day 7 sky cover and dewpoints. The problem before the NBM was that each office was starting with whatever they felt like, and for people who live close to the border of two offices you could have wildly different forecasts. It was also pretty poor practice to try and cherry pick the model of the day, nobody is actually any good at that.

It will also be interesting to see how the NWS forecasts go today. Primarily starting with the NBM is producing better and more consistent forecasts, but the data is a cycle old. So the afternoon forecasts today will be using mostly 00z data fed into the NBM. That means chances of snow will be lower with this NBM run because of the inclusion of the paltry GEFS/GFS (and at this time range it is almost entirely GFS/Euro based). Blindly populating with NBM may lead to artificially lower snow forecast if the 00z guidance was indeed a burp.

I meant what I said. When you have your mean sitting higher than the 50th percentile, you're dragging the mean up. We've actually seen in the regional verification that quite often our (the NWS') forecast is closer to the 90th percentile than the 10th of the forecast spread. We almost always forecast too much snow, except for the picnic tables and First Connecticut Lake.

Those are pretty tight ranges for 25/75 too. That would suggest high confidence from the Euro. However, we've seen in the past few years how the ensembles can be underdispersive (i.e. not capturing the true spread of potential outcomes). It's why we tend to see the ensemble mean shift in tandem with the op so often.

It's important to look at these panels like Ginxy has here. The ensemble means may look nice, but don't tell the whole story. Like for Mount Tolland for instance, the EPS mean is about 5.5 inches. But the 25th/75th range is 1.5 to 8.5 inches. So some bigger members are dragging the mean up, while the floor still remains quite low. When you can start setting the floor higher, that's when you can start honking for a storm.

Was that why we stopped having GTGs?

Two maybe three chances for at least 3-6" for many by the middle of the month? That's pretty much average for an entire month right there.

I think they have some select locations available for free from the EC site. May be garbage graphics though.

He could learn a thing or two from Kev. Just double down on your next call.

Yeah, there are plenty of ways to salvage 1/10 up here. If we have snow on the ground after 1/7, I wouldn't be calling for a total pack reset on 1/10.

Don't really have the players on the field for another 36-48 hours. But I would start to feel more confident around that time.

The clusters with the sharpest shortwaves are definitely the whitest ones for the region. Driest cluster is more of a Southeast ridge looks with trof in Canada, so it just goes through the meat grinder.

You can already see the CAD on that weenie Euro gust forecast. Instead of 60+ mph through interior Maine it's already forecasting 40s in a CAD shape. Not hard to sell that right now.

The EFI looks at the 5 week window centered on the forecast period and compares model climate to ensemble distribution. The more ensembles higher or lower than the m-climate the higher or lower EFI will be. Irrespective of sign, EFI closer to 1 is more unusual. Shift of tails looks at the top 10% of ensemble members. The more extreme those members, the higher the shift of tails will be. That can clue you in to just how big the potential is with certain events.

The Euro EFI stuff can help with that too. The shift of tails shows you what the outliers at the top end are saying, but the shading tells you where the meat of the ensemble is.

Right on time to set everyone's expectations unreasonably high.

But how much snow?

DESI tossing out 4 clusters, but there are two dominant ones accounting for 64% of ensemble members. More or less varying degrees of amplification. But the snowier solutions are more amplified trof. The more you lower heights locally, he snowier the outcomes. Which is a pretty wild difference when you think about it. Some clusters feature a ridge overhead 08.00z, vs other clusters featuring a trof. So nothing is locked.

Just some ?