OceanStWx

I’m pretty sure the 5SD is verifying, what’s not is location and thus getting it to the ground/forcing the QPF.

I’ll give you that if you showed people unlabeled 500 mb charts it might be tough to really tell the flow pattern apart, but that’s a bullet I’m saving for the real deal.

Verbatim maybe, but it’s also a failure of model interpretation. We as an enterprise need to do better to interrogate QPF fields. Scoots mentioned some red flags with dry air in the lift generating layers. But it’s easier to rip and read a QPF map than make your own.

I agree it's not a great look, but sometimes in the short term you just have to go with what you think is going to happen, it's in the best interest of your users. I stand by my 0.25" since it hasn't rained yet at ASH.

It better pound between now and 18z if the hi-res stuff is going to be right. Inch an hour for the next 4 hours?

I think one of the hardest things to do is pull the plug on your own forecast. And I know people I work with struggle to know when to take the forecast in a new direction (positive or negative busts) in the near term. Like it's fine to say gusts to 70 mph are possible on the Cape, but if that's the upper bound what's the lower bound? 48 hours out 30 mph might be just as likely as 70 mph, but we never hear the 30. That I'll totally agree with. The NWS is struggling with that mentality given all the social media-rologists out there. My opinion is I would rather the NWS be the right answer than the first answer. If nothing else maybe the HazSimp process will move us towards a more traffic light mentality. Have a yellow light/caution product that "gets the word out" without details, and when it's time to honk you have the red light/warning. As things currently stand we don't have a great product set for events we know will be below warning criteria but will still have impacts.

I think there is a tendency to get too specific too early. It's one thing when we talk about impacts here on the board, but when actual forecasts are going out that describe in detail how the storm will evolve 72+ hours in advance stretches beyond our capabilities. QPF is a great example, it's just a poorly forecast variable by the models and even over 48 hours it can change quite a bit. I know with my colleagues I see way too much trust in models, and especially the latest model run. The hill I'm prepared to die on this winter is holding onto watches longer. We shouldn't be crying wolf on 40% of them.

I gotta say, it feels good to be using ensemble sensitivity again.

There's nothing like a good Plains blizzard. We get the Atlantic moisture, but true blizzards are so hard to come by around here.

While the pattern may ultimately be a colder one, that FV3 cold bias is going to be a tough one to shake this winter. Because of the seasonal progression cold is easier to believe, but given the old GFS propensity to rush cold and the current version biased at long ranges it's probably better that cold shot expectations are tempered.

Barrington may as well have been EYW during most nor'easters.

I don't live in the tropics.

Classic radiating night up here. A little WAA does wonders to decouple quickly after dark. BML 52 > 42 > 36 in 2 hours. Even SFM did 12 degrees in an hour after sunset.

Yeah, overseeded during the driest Sept in 70 years.

GEFS doesn't move to FV3 core until late 2020 I believe, so for this whole winter the GFS and GEFS won't be running the same physics. Making them even less useful.

Well a cold bias would tend to do well with a snowstorm. Feel free to read up on the latest (https://www.emc.ncep.noaa.gov/users/meg/fv3gfs/) But long story short: FV3 gets progressively colder with longer lead times, and the bias is worse near the surface up through about 700 mb, then worse again around 200 mb.

It's definitely going to be a problem for 240 hour snow map rip and readers.

Good time for a reminder that the FV3 continues to have a cold bias, especially with increasing lead time. In other words, the cold bias gets worse as you move towards the end of the run. Unfortunately something we're going to have to mentally QC out for the time being until they can upgrade/fix it (but GEFS moving to FV3 is a priority first).

It was wild to be sweating it out in near 90 degree heat at a wedding with brown foliage halfway up the hills around Beacon, NY.

4 weeks ago some of these same places were experiencing record heat.

We actually have a lot of trees locally dumping their leaves already without much of a splash of color. My streets trees have given up most of their leaves now, but they did have a good initial burst of red before fading quickly.

Drove through your 'hood this weekend on the way to the Hudson Valley and it's all very orange/brown.

GTFO with Kuchera Basically at all NWS offices the forecasters will create grids for temps, PoP, QPF, snow ratio. From that you get Wx grids, and from that you can determine where snow you be output in a snow grid. You can also choose how you want to display it temporally. A one hour grid will produce a more realistic snow fall than a 3 or 6 hour grid that has parameters averaged over that time. It is up to the forecaster whether they want to hand draw grids, pick a model of the day, or use some form of a blend. Personally I use a blended approach, and then tweak for things like local effects or forecast position of banding. We have grids every hour with thousands of grid points, so we rarely are editing just one gridpoint. Rather we try and lock down the base grids and everything sort of fall out from there (i.e. if your temp, dewpoint, PoP, QPF, snow ratio are good, then your Wx and snowfall should be good too). We could automate a lot of that, but you would lose local knowledge and QC. Like we don't want to blend GFS 2 m temps in a damming scenario, but the computer doesn't know that.

Only one day has been warmer than yesterday at PWM this late in the season. 90 on 9/26/07 And we tied a record yesterday (89) set in 1895 when the thermometer was on the top of some tar roof in Old Port.

My hydrangeas are still flowering!