Mid January/Mid February Medium/Long Range Discussion

[Maestrobjwa]

18 minutes ago, Always in Zugzwang said:

Well, if you're looking at the *total* QPF through that time (I'm looking at TT site), then yeah, it would appear like huge swaths of moisture because it's a total amount over all those hours.  Sort of like a Jackson Pollock painting (I use the term "painting" very loosely, hahaha!).  Much larger area then, of course, would be covered by something, especially when you get out in time.  But if you're looking at the 24-h amounts, you should see more "motion" (for lack of a better word) as a system moves through...it will increase, then decrease, over an area for various 24-h increments.

Now, I was actually talking about like the actual map that shows the rain/snow. Like when it looks like this:

[CAPE]

18z GEFS says fogedaboudit for the next 7 days(we have pretty much known this for a while now), and then this-

Way out there, but what else we got?

[psuhoffman]

4 minutes ago, C.A.P.E. said:

18z GEFS says fogedaboudit for the next 7 days(we have pretty much known this for a while now), and then this-

Way out there, but what else we got?

This is the one they want 

[mappy]

23 minutes ago, psuhoffman said:

This is the one they want 

I could be interpreting this map wrong, but I don’t think at 380hrs, having only a 50% chance of 3 or more is all that good. 

[psuhoffman]

1 minute ago, mappy said:

I could be interpreting this map wrong, but I don’t think at 380hrs, having only a 50% chance of 3 or more is all that good. 

Well there is absolutely nothing the first 7 days...so it’s really a 50% week 2. And that’s above climo. Avg chance of 3” of snow during that week is about 25%. Is it the best probability map I’ve ever seen no. But it’s better than those weeks of 10-20%

[jaydreb]

55 minutes ago, Maestrobjwa said:

Now, I was actually talking about like the actual map that shows the rain/snow. Like when it looks like this:

It should look the same no matter the hour. This storm looks like that because it is snowing over a large area in that 6 hr period.  Its a large swath of precip.  

[mappy]

Just now, psuhoffman said:

Well there is absolutely nothing the first 7 days...so it’s really a 50% week 2. And that’s above climo. Avg chance of 3” of snow during that week is about 25%. Is it the best probability map I’ve ever seen no. But it’s better than those weeks of 10-20%

Got it, I had a feeling I was interpreting it incorrectly. This is why I don’t analyze long range, and stats was never my strongest subject lol

thanks for the info!

[Bob Chill]

4 minutes ago, jaydreb said:

It should look the same no matter the hour. This storm looks like that because it is snowing over a large area in that 6 hr period.  Its a large swath of precip.  

Exactly. Those panels cover 6 hours of qpf. I always compare the 6hr output to the sim radar panel to get a better idea of the flavor of what type of event ops are predicting. In this case it looks exactly like the nicer waves in JFM of of 2014 & 15. 

[psuhoffman]

Just now, mappy said:

Got it, I had a feeling I was interpreting it incorrectly. This is why I don’t analyze long range, and stats was never my strongest subject lol

thanks for the info!

You’re welcome. You weren’t interpreting it wrong. It means exactly what it says. The DC area is between the 50-60% chance of 3”. Is what it is. I’m just adding historical context.  The fact that it’s all in a 5 day window though actually does make it more impressive.  It’s not from a shotgun effect of 5 or 6 low level threats over 15 days.  That’s typically a not so good thing.  Additionally, I know it being 10 days away sucks but seeing that high probability for events that are 9+ days out is actually impressive.  It’s hard to get high odds at that range, for obvious reasons.  

Keep in mind that map is very course and won’t pick up local meso scale features like elevation well. You are way closer to the 60 than the 50 so you are likely about 58 or 59% and maybe more since you have some oragraphic help. DC is probably about 52% guessing from that map.  

[psuhoffman]

[cbmclean]

Just now, psuhoffman said:

Would you consider that a (relative) fringe.

[jaydreb]

[WinterWxLuvr]

1 minute ago, psuhoffman said:

Start the thread

[aldie 22]

5 minutes ago, psuhoffman said:

I'd prefer to see the southern edge further south

[mappy]

6 minutes ago, psuhoffman said:

I would take that (figuratively speaking of course) 

[jaydreb]

15 minutes ago, Bob Chill said:

Exactly. Those panels cover 6 hours of qpf. I always compare the 6hr output to the sim radar panel to get a better idea of the flavor of what type of event ops are predicting. In this case it looks exactly like the nicer waves in JFM of of 2014 & 15. 

It’s the 6 hour period ending at the timestamp (i.e. 12z Feb 9), not starting, right?

[87storms]

19 minutes ago, psuhoffman said:

You’re welcome. You weren’t interpreting it wrong. It means exactly what it says. The DC area is between the 50-60% chance of 3”. Is what it is. I’m just adding historical context.  The fact that it’s all in a 5 day window though actually does make it more impressive.  It’s not from a shotgun effect of 5 or 6 low level threats over 15 days.  That’s typically a not so good thing.  Additionally, I know it being 10 days away sucks but seeing that high probability for events that are 9+ days out is actually impressive.  It’s hard to get high odds at that range, for obvious reasons.  

Keep in mind that map is very course and won’t pick up local meso scale features like elevation well. You are way closer to the 60 than the 50 so you are likely about 58 or 59% and maybe more since you have some oragraphic help. DC is probably about 52% guessing from that map.  

it's an interesting topic, actually.  relying on probabilities that far out is also assuming models can accurately portray probabilities that far out.  i'm not sure what feeds those stats.  if indices play a role in the formulas, then i think that's more reliable than ensembles.   otherwise, it's going to change a lot from run to run.

[psuhoffman]

15 minutes ago, 87storms said:

it's an interesting topic, actually.  relying on probabilities that far out is also assuming models can accurately portray probabilities that far out.  i'm not sure what feeds those stats.  if indices play a role in the formulas, then i think that's more reliable than ensembles.   otherwise, it's going to change a lot from run to run.

They simply use the output of each member to calculate odds. Eps is less jumpy because it has more members and more spread typically. Gefs is underdispursed. NCEP is aware of it. The whole system is being changed to operate off the new FV3 core soon. 

I’m not sure an index probability would be better because the index is based off the individual members output.   And their results are impacted by that too. We only know what the indexes are now. (Bad just in case you weren’t sure) but we don’t know in 5/10/15 days...those estimates are based off the models so can be just as wrong. 

There are some pattern analog progression models but they don’t seem to be any more reliable because patterns often evolve different based on different factors. 

There are SST long range models but they mostly busted BAD last year and this year. I don’t blame them...I busted both years because I tend to lean heavily on SST analogs for long range and lately the correlations are failing. I need to look more into some of the methods Anthony and Tom use wrt angular momentum, walker cell strength. Those correlations seem to be more important that canonical sst anomalies in a changing climate where waters are in a warmer state. I know HM busted this year but he has stated that was partly because he broke some of his own rules this year. Everyone makes mistakes!  

[Ralph Wiggum]

47 minutes ago, jaydreb said:

It’s the 6 hour period ending at the timestamp (i.e. 12z Feb 9), not starting, right?

i though the comp reflectivity map was what the radar is forecast to look like at a specific time and not accums? Forecast of dbz's on a radar essentially.

[Bob Chill]

13 minutes ago, Ralph Wiggum said:

i though the comp reflectivity map was what the radar is forecast to look like at a specific time and not accums? Forecast of dbz's on a radar essentially.

Correct. It's an instantaneous snapshot of sim radar at the time stamp

[AlexD1990]

We abscond.

Sent from my SM-G975U using Tapatalk

[jaydreb]

24 minutes ago, Bob Chill said:

Correct. It's an instantaneous snapshot of sim radar at the time stamp

Right.  My question was about the 6 hour qpf map that we were discussing earlier.  I copied the wrong map in my post.  Here is the map I’m asking about.  6 hrs before after this time?  Sorry.  

[Maestrobjwa]

1 hour ago, psuhoffman said:

Yep--that would be almost right for our every 3-4 year footer!  But seriously, I do hope we still see this potential by the end of the week...if we do, it might turn into the best chance we've had in four years, lol

[87storms]

45 minutes ago, psuhoffman said:

They simply use the output of each member to calculate odds. Eps is less jumpy because it has more members and more spread typically. Gefs is underdispursed. NCEP is aware of it. The whole system is being changed to operate off the new FV3 core soon. 

I’m not sure an index probability would be better because the index is based off the individual members output.   And their results are impacted by that too. We only know what the indexes are now. (Bad just in case you weren’t sure) but we don’t know in 5/10/15 days...those estimates are based off the models so can be just as wrong. 

There are some pattern analog progression models but they don’t seem to be any more reliable because patterns often evolve different based on different factors. 

There are SST long range models but they mostly busted BAD last year and this year. I don’t blame them...I busted both years because I tend to lean heavily on SST analogs for long range and lately the correlations are failing. I need to look more into some of the methods Anthony and Tom use wrt angular momentum, walker cell strength. Those correlations seem to be more important that canonical sst anomalies in a changing climate where waters are in a warmer state. I know HM busted this year but he has stated that was partly because he broke some of his own rules this year. Everyone makes mistakes!  

Yea that’s a good point re indices. It would end up being the same result since indices are just another way of explaining what the current conditions are. With precip being essentially a weekly event here, I still think tracking cold/overall patterns is a better vibe. Those are the topics I’ve been gravitating towards (pna, nao, etc).  With the Pacific being our enemy this year, it sounds like we may need an atmospheric traffic jam to shunt some of that milder air south.

[clskinsfan]

0Z NAM has some snow out this way on Saturday morning. But overall it is pretty lame.

Haha....Ninja'd

[psuhoffman]

29 minutes ago, jaydreb said:

Right.  My question was about the 6 hour qpf map that we were discussing earlier.  I copied the wrong map in my post.  Here is the map I’m asking about.  6 hrs before after this time?  Sorry.  

It’s the previous 6 hours precip rate. 

[Ji]

Too bad the gfs is 1004 mb and not 988

[Scraff]

7 minutes ago, Ji said:

Too bad the gfs is 1004 mb and not 988

It’s pretty disheartening to see a storm like that not deliver the goods. So close but...

[peribonca]

Anyone feeling lucky and want to start a new thread for February only?

[Ralph Wiggum]

Not a cozy feeling seeing guidance almost completely lose the strong  -NAO being forecast for a while in the LR and replace it with a +NAO and SE ridge almost across the board.