December 2019 Weather Discussion for the PNW

[Frontal Snowsquall]

FWIW, today's 12z EURO Control run out to day 15.

 

[Deweydog]

FWIW, today's 12z EURO Control out to day 15.

 

Ugh.

 

That looks nothing like what I have on my easel.

[Front Ranger]

Nah... Just anyone who makes purely subjective, several-day extrapolations of 240 hour model solutions but refuses to look at 240+ hour model solutions because of they’re value as pure fantasy.

I don't think they are the same thing.

 

Pattern progression recognition is more akin to looking at the CPC ensemble analogs for clues to how the modeled pattern may develop.

 

And any type of speculation on here could be labeled "subjective", so that's a pretty empty statement.

[Frontal Snowsquall]

The thing I always have thought about the analogs is that, even if what is shows was EXACTLY the way it was in, say, December 1949, I doubt that the weather would play out exactly like it would through January 1950. So I really don't take much stock in them. Yes, things like ENSO and other indices can tell us the likelihood of how a winter will play out, but sometimes I see an analog on the CPC of something from 1949, and a few people will get excited because of January 1950. Or some other analog from a time that ended up being good several weeks later. I really doubt there has ever been two winters that played out pretty much the same from mid November through early March.

Yeah, there are way too many factors at play during the course of a season that we will never see a winter replay a previous one. I like the CPC analogs but from each run, the analogs it displays can change quite a bit showing that a slight shift in the pattern can lead to vastly different results. One of the reasons why I don't pay attention to them as much as I use to.

[SilverFallsAndrew]

Saw Fairbanks, AK ran a +10.1 departure in November. 

[Frontal Snowsquall]

[Jesse]

Looks like a 47/30 day at PDX. Putting up a -4F departure for the date.

[SilverFallsAndrew]

Looks like a 47/30 day at PDX. Putting up a -4F departure for the date.

 

Looks like -4 at SLE too, -5 departure at EUG. 

[SilverFallsAndrew]

My brother posted a 64/34 in Coquille today. Must have been lovely down in Coos County this afternoon.  

[Front Ranger]

Didn’t you just yesterday forecast a mid December pattern change. That’s outside of 72hrs.

It's possible to take yourself with a grain of salt.

[Guest CulverJosh]

Didn’t you just yesterday forecast a mid December pattern change. That’s outside of 72hrs.

No forecast. A wishcast. But seems like the patterns are every 3 weeks so far this fall. Not that it means anything.

[Jesse]

[Omegaraptor]

Looks like a 47/30 day at PDX. Putting up a -4F departure for the date.

 

HIO got to 50. No east wind there.

[Jesse]

[SilverFallsAndrew]

The huge positive departure in Fairbanks last month is probably why we had wall to wall arctic air and snow here for all of November. I just love when they torch.

 

OUCH

[MossMan]

[Front Ranger]

[Deweydog]

I don't think they are the same thing.

 

Pattern progression recognition is more akin to looking at the CPC ensemble analogs for clues to how the modeled pattern may develop.

 

And any type of speculation on here could be labeled "subjective", so that's a pretty empty accusation.

A 312 hour model solution isn’t subjective. It’s probably inaccurate, but it’s not subjective.

 

Anyhoo, imagining patterns moving forward from a ten day statistical model solution while immediately declaring any further statistical output on an operational level to be drivel just comes off as phunny. Especially when you do it with a solution which is most likely a snapshot of a progressive wave train.

[Omegaraptor]

Was pretty chilly for 2/3 of the lower 48.

 

Last1mTDeptUS.png

 

Washington County donut hole.

[Jesse]

Was pretty chilly for 2/3 of the lower 48.

Good point, I was totally wrong that November sucked for much of the west.

[Guest CulverJosh]

A 312 hour model solution isn’t subjective. It’s probably inaccurate, but it’s not subjective.

 

Anyhoo, imagining patterns moving forward from a ten day statistical model solution while immediately declaring any further statistical output on an operational level to be drivel just comes off as phunny. Especially when you do it with a solution which is most likely a snapshot of a progressive wave train.

Outside of 120 hrs is at best a trend. And that is pushing it.

 

Inside 72 hrs=50% at least, 72-120 hrs=20% at best, more than 120 hrs= 5-15-%; closer to 5%

[Frontal Snowsquall]

HIO got to 50. No east wind there.

Aurora and Kelso also got to 50.

[Jesse]

Aurora and Kelso also got to 50.

Phoenix, AZ was probably even warmer!

[Front Ranger]

A 312 hour model solution isn’t subjective. It’s probably inaccurate, but it’s not subjective.

 

Anyhoo, imagining patterns moving forward from a ten day statistical model solution while immediately declaring any further statistical output on an operational level to be drivel just comes off as phunny. Especially when you do it with a solution which is most likely a snapshot of a progressive wave train.

Of course it's speculative. Like many other things on here. But there is a difference between statistical model output and the human ability to recognize pattern progressions.

 

I'm not saying my speculation is more valuable than LR model output. But there are classic precursors to many patterns...and there's nothing inherently more subjective about that than anything else posted on here.

[Deweydog]

Outside of 120 hrs is at best a trend. And that is pushing it.

 

Inside 72 hrs=50% at least, 72-120 hrs=20% at best, more than 120 hrs= 5-15-%; closer to 5%

Huh?

[Omegaraptor]

[Guest CulverJosh]

Huh?

Accuracy.

[Deweydog]

Of course it's speculative. Like many other things on here. But there is a difference between statistical model output and the human ability to recognize pattern progressions.

 

I'm not saying my speculation is more valuable than LR model output. But there are classic precursors to many patterns...and there's nothing inherently more subjective about that than anything else posted on here.

All true. But I’m sure on some level you can probably see why it might come across as delightfully hypocritical.

[Guest CulverJosh]

[Deweydog]

Accuracy.

20% accuracy 3-5 days out?

[TigerWoodsLibido]

40F and clear. Should be a nice night.

[Front Ranger]

Good point, I was totally wrong that November sucked for much of the west.

Did I say you were wrong?

 

Just pointing out that the warmth in AK did correlate to cold in the lower 48. Sometimes that includes the PNW, sometimes it does not, but regardless - when it's warm in AK during the cold season, it usually gets cold downstream to the south.

 

No exception last month.

[SilverFallsAndrew]

Looks like November was the inverse of October. Let's keep that up through December and then flip it for January. Ok guyz?!

[Front Ranger]

All true. But I’m sure on some level you can probably see why it might come across as delightfully hypocritical.

Subjectively, yes.

[Brian_in_Leavenworth]

Yeah, there are way too many factors at play during the course of a season that we will never see a winter replay a previous one. I like the CPC analogs but from each run, the analogs it displays can change quite a bit showing that a slight shift in the pattern can lead to vastly different results. One of the reasons why I don't pay attention to them as much as I use to.

I get using the analogs for a predicted upcoming pattern, like the 6-10 or 8-14 day range, but to use it to say "that analog from 19xx was 3 weeks before a great arctic outbreak, so that means we may get a repeat" is totally meaningless IMO.

[Front Ranger]

I get using the analogs for a predicted upcoming pattern, like the 6-10 or 8-14 day range, but to use it to say "that analog from 19xx was 3 weeks before a great arctic outbreak, so that means we may get a repeat" is totally meaningless IMO.

Totally agree. Whether you're extrapolating a 10 day out pattern or looking at 348 hour maps, if the 10 day modeled pattern isn't close, anything past that point is totally useless.

 

Regardless, anything past 10 days will be low probability...but if you have a high degree of model agreement in the believable range, it helps a bit.

[Dave]

It did not take long for the fog to reform. Blech.

[Timmy Supercell]

Saw Fairbanks, AK ran a +10.1 departure in November. 

 

Their MEAN? Or High? 

[Rubus Leucodermis]

Inside 72 hrs=50% at least, 72-120 hrs=20% at best, more than 120 hrs= 5-15-%; closer to 5%

 

Consistency matters a lot as well. Consider two scenarios.

 

1. Models show an arctic blast five days out, but they've been flip-flopping about it.

2. Models show an arctic blast five days out, and they've all been the same thing for the past five days, too.

 

(1) wouldn't get me excited about an arctic blast in five days, but (2) would.

[Timmy Supercell]

Balmy 47 for a high and 34 for a low, also balmy.