• Hello, please take a minute to check out our awesome content, contributed by the wonderful members of our community. We hope you'll add your own thoughts and opinions by making a free account!

Pattern Failboat February

KyloG said:
Past dozen runs…🤷‍♂️

View attachment 167713
Click to expand...
Man that's the first run that's gotten Northern MS/AL in the game. Feels great to see the bounce back to the South. I was honestly pretty worried she was gonna drift North a little more and take us out of the game.

And we still might whiff... I'd be feeling really good right now if I lived around the North Central NC/VA border. Like, really good.
 
rburrel2 said:
Man that's the first run that's gotten Northern MS/AL in the game. Feels great to see the bounce back to the South. I was honestly pretty worried she was gonna drift North a little more and take us out of the game.

And we still might whiff... I'd be feeling really good right now if I lived around the North Central NC/VA border. Like, really good.
Click to expand...

Yeah I was about to make mention that, the snowfall footprint especially on the last two runs stretched further to the SW over the Deep South


Sent from my iPhone using Tapatalk
 
Myfrotho704_ said:
AI gonna pull a CMC in the next couple of days or take a W here ? I’d take the middle ground with initial slop south of I-40 and I-40 north doing better. But it’s far more encouraging to see the AI with that vs say the GFS/CMC
Click to expand...
I mean why can’t it just work out for once. It can’t always go the wrong way. Right?
 
I do believe the article was posted on here that the Euro AI has now claimed the top spot in accuracy among all models at this range ... with the GFS coming in way less accurate. Did I read that correctly?
 
Landof2rivers said:
Why not? And I have no evidence either way bc I lurk and try to learn. There are posters here I pay attention to and you are one of them so if you don’t mind could you please elaborate?
Click to expand...
I appreciate that. For me, it's a couple of things. One, fundamentally, you should never ever ever ever give much credence to any deterministic model showing a snowstorm 7+ days out. Not ever, particularly when there's not broad support from other models/ensembles.

Two, just personally, I haven't observed it outside of this season, so as far as I'm concerned, it's the new kid on the block and needs to prove itself. Holding onto this storm and being generally accurate with it from this far out, would help. I'm not saying it's wrong, but I just haven't seen enough wins to trust it yet.

It's pretty much all the eye candy we have to look at for now, so we'll take it.
 
SnowmanVol99 said:
But the GFS seems to be on a island with the other models leaning more toward Al right now. Hard to go with GFS all by itself. Not saying it can’t be right but it doesn’t have much support and some of its ensembles don’t agree.
Click to expand...
Agreed!
 
Rain Cold said:
I appreciate that. For me, it's a couple of things. One, fundamentally, you should never ever ever ever give much credence to any deterministic model showing a snowstorm 7+ days out. Not ever, particularly when there's not broad support from other models/ensembles.

Two, just personally, I haven't observed it outside of this season, so as far as I'm concerned, it's the new kid on the block and needs to prove itself. Holding onto this storm and being generally accurate with it from this far out, would help. I'm not saying it's wrong, but I just haven't seen enough wins to trust it yet.

It's pretty much all the eye candy we have to look at for now, so we'll take it.
Click to expand...
I'm still digesting AI tech but aren't AI forecast models still "learning". They're not infallible (prob never will be) but should get better with time. With that said I believe they're verification scores are decent and let's hope it's spot on this time
 
RBR71 said:
I'm still digesting AI tech but aren't AI forecast models still "learning". They're not infallible (prob never will be) but should get better with time. With that said I believe they're verification scores are decent and let's hope it's spot on this time
Click to expand...
Yeah that's supposedly true. And that reminds me of the other thing I wanted to say. The verification scores seem respectable. But I'm not taking that to mean a particular solution, say a future snowstorm, has a higher chance of coming true because the model is showing it. I know you're not saying that.

I do like that its scores seem to be good overall. Hopefully, it will continue learning and just get better at everything. No idea how that process actually works or how long it takes, though.
 
Rain Cold said:
I know the GFS sucks, but I still don't trust the AI yet.
Click to expand...

Not at 8 days out. But 5 days in I’ll admit it’s been pretty accurate iirc. Last storm we were begging for a nw trend and it wouldn’t give it to us. When it caved to suppressed we were done. It didn’t have the warm nose on the one before that, but it had a lot of others specifics pretty well.

Long range though yeah, ensembles is what I trust the most, relatively speaking. If we can get the EPS to match this, game on.
 
EURO AI’s consistency is remarkable so you gotta give it that much. I doubt it’s completely right. I still think south of 40 is much more of an Ice issue but that AI is giving me pause that we could be getting more of a snow look with better confluence and suppressed height field. I could care less what the GFS shows. It’s a garbage model and I’ll die on that hill
 
SnowNiner said:
Not at 8 days out. But 5 days in I’ll admit it’s been pretty accurate iirc. Last storm we were begging for a nw trend and it wouldn’t give it to us. When it caved to suppressed we were done. It didn’t have the warm nose on the one before that, but it had a lot of others specifics pretty well.

Long range though yeah, ensembles is what I trust the most, relatively speaking. If we can get the EPS to match this, game on.
Click to expand...
Well the EPS is the best looking ens for us as of now so there’s that
 
SnowNiner said:
Not at 8 days out. But 5 days in I’ll admit it’s been pretty accurate iirc. Last storm we were begging for a nw trend and it wouldn’t give it to us. When it caved to suppressed we were done. It didn’t have the warm nose on the one before that, but it had a lot of others specifics pretty well.

Long range though yeah, ensembles is what I trust the most, relatively speaking. If we can get the EPS to match this, game on.
Click to expand...
The EPS is much more closer to the AI vs the GFS. The CAD regions have been consistently getting hit on the EPS during this timeframe
 
The AI model was trained on historic data already so yeah it's still learning but it's got a good base already. I think the issue with it may be 2MT more than 500mb or at least that's how i understood some of the results on the EC site. To me it's like any other model we roll out here, it's not infallible but it's probably not wildly incorrect either.
 
SD said:
The AI model was trained on historic data already so yeah it's still learning but it's got a good base already. I think the issue with it may be 2MT more than 500mb or at least that's how i understood some of the results on the EC site. To me it's like any other model we roll out here, it's not infallible but it's probably not wildly incorrect either.
Click to expand...
Yeah that's how I understood it as well when i read it. It's respectable verification scores are at 500mb, seems all modeling struggles to nail down 2mt, especially in CAD/micro-climate regions.
 
RBR71 said:
I'm still digesting AI tech but aren't AI forecast models still "learning". They're not infallible (prob never will be) but should get better with time. With that said I believe they're verification scores are decent and let's hope it's spot on this time
Click to expand...
Yall gone make me have to post these charts again it looks like :)

Over the previous 3 winters (21-22. 22-23, 23-24), the Euro AI (AIFS) and the AI GraphCast (I'm guessing this is the Euro AI Graphcast and not the GFS AI Graphcast) were the best performing models here on the chart for the Day 5 forecast at 500mb in the Northern Hemisphere Extratropics.

AI Model Perf.jpg


Here are the charts for the Euro AI (AIFS) outperforming the Regular Euro (IFS) last winter (23-24). These charts are for days 1 thru 10. Note that the chart on the left is anomaly correlation, so the top line on the chart is the better performer (Euro AI). Upper right chart is the 850mb temperature error, so the bottom line on the chart is the better performer (Euro AI). Bottom right chart is in the Tropics.

AI Model Perf vs Euro.jpg
 
griteater said:
Yall gone make me have to post these charts again it looks like :)

Over the previous 3 winters (21-22. 22-23, 23-24), the Euro AI (AIFS) and the AI GraphCast (I'm guessing this is the Euro AI Graphcast and not the GFS AI Graphcast) were the best performing models here on the chart for the Day 5 forecast at 500mb in the Northern Hemisphere Extratropics.

View attachment 167727


Here are the charts for the Euro AI (AIFS) outperforming the Regular Euro (IFS) last winter (23-24). These charts are for days 1 thru 10. Note that the chart on the left is anomaly correlation, so the top line on the chart is the better performer (Euro AI). Upper right chart is the 850mb temperature error, so the bottom line on the chart is the better performer (Euro AI). Bottom right chart is in the Tropics.

View attachment 167728
Click to expand...
I'll be sure to save them this time 😂
 
Rain Cold said:
Yeah that's supposedly true. And that reminds me of the other thing I wanted to say. The verification scores seem respectable. But I'm not taking that to mean a particular solution, say a future snowstorm, has a higher chance of coming true because the model is showing it. I know you're not saying that.

I do like that its scores seem to be good overall. Hopefully, it will continue learning and just get better at everything. No idea how that process actually works or how long it takes, though.
Click to expand...
I know we all desperately want these maps to come to fruition but we have been fooled to many times in the past to fall for this.
The AI is the new shiny toy in the tool box but this just looks to good to be true!
I don't believe that our climate has changed so much that this can't happen but,
I'd be more out to believe this in January than mid Feb.
If I'm not mistaken based upon some of the posts from the Jan storms that Burrel was posting it was extremely to high on totals.
Would love to see half of this happen in that foot print,
It would be like times of old...
 
Gonna be a better run from the GFS here but still generally pretty ass. I hate that much ridge over head. When I get this kind of MSLP look and a ridge spoils the party I get triggered IMG_1389.png
 
Billypg70 said:
I know we all desperately want these maps to come to fruition but we have been fooled to many times in the past to fall for this.
The AI is the new shiny toy in the tool box but this just looks to good to be true!
I don't believe that our climate has changed so much that this can't happen but,
I'd be more out to believe this in January than mid Feb.
If I'm not mistaken based upon some of the posts from the Jan storms that Burrel was posting it was extremely to high on totals.
Would love to see half of this happen in that foot print,
It would be like times of old...
Click to expand...
This is pretty much where I am too. Its pretty to look at, but I mean, there is just no way the whole upstate is getting a foot plus. It hasn't happened in nearly 40 years. I know isolated totals like that have happened quite a bit but not like that is showing. My guess is, even if the upstate is cold enough for frozen at go time, it will be a change over or mainly ice event. The AI spit a lot of these maps out for the upstate for the first Jan system too but it was showing all frozen as snow and that didn't come close to verifying.
 
You must log in or register to reply here.
Back
Top