ATL: LEE - Models

[LarryWx]

0Z EPS through 240: not as many hits on US as 12Z run but there still were 7: 6 on ME and 1 on NY. There may not be any more US hits after 240. Nova Scotia gets clobbered even more than on the 12Z. Newfoundland also gets some direct hits in addition to leftovers from some of the Nova Scotia hits.

[RevanTheJedi96]

GFS shifted to the East again. With Lee weakening already, some great signs already in the works.

[PavelGaborik10]

GFS shifted to the East again. With Lee weakening already, some great signs already in the works.

The 06z literally landfalls slightly W of the 0z and is the polar opposite of the million mile shifting Euro.

[tomatkins]

GFS shifted to the East again. With Lee weakening already, some great signs already in the works.

The 06z literally landfalls slightly W of the 0z and is the polar opposite of the million mile shifting Euro.

It seems clear that the interaction with Margot is the key here. Euro has them close and that pulls Lee out to see - but also the models dont really do well with these interaction scenarios. GFS has them far enough part that they dont really interact, so it landfalls, and Margot continues on to pull Nigel up to the central Atlantic instead of allowing it to move into a similar position as Lee is now.

[crimi481]

SAL MAP
http://tropic.ssec.wisc.edu/real-time/s ... split.html

[rickcorvin]

The spread on the 00z GEFS has gotten considerably larger at hour 168. So much so that perhaps even South Carolina is in play. We’ve got a long way to go with this one.


Sent from my iPhone using Tapatalk

The general rule always used to be that wherever the landfall is 6-10+ days out, that’s usually not where it’s going. So those kind of shifts would not shock me.

Yeah, that was, and possibly still is the rule. The best place to be 6 to 10 days out is right in the bullseye. We still have much to learn about a science that is constantly in motion. Can anyone say chaos theory.


Sent from my iPhone using Tapatalk

I agree with the larger point about forecast uncertainty 6-10 days out, particularly with models. But we only need to look back a couple of weeks to see a storm (Idalia) that made landfall near the long range NHC track.

[Abdullah]

0z hurricane models. All of them show Lee achieving a first peak within 6 hours from now, but the two HAFS models show a secondary peak in 1.5-2 days (after an EWRC) with even lower pressure.

HAFS-A:

• Init: 941.0 mb, 148.9 kt
• 6z (now): 920.1 mb, 161.9 kt (First peak)
• Second peak: 18z 9/9 (42 hrs), 910.3 mb, 157.0 kt

HAFS-B:

• Init: 939.5 mb, 132.7 kt
• 6z (now): 929.5 mb, 144.7 kt
• First peak: 12z 9.8 (12 hrs), 919.5 mb, 160.6 kt
• Second peak: 6z 9/10 (54 hrs), 912.7 mb, 154.4 kt

HWRF:

• Init: 933.8 mb, 150.4 kt
• 6z (now): 926.7 mb, 137.0 kt
• First and only peak: 12z 9.8 (12 hrs), 922.4 mb, 140.2 kt

HMON:

• Init: 932.5 mb, 123.1 kt
• 6z (now): 916.3 mb, 147.5 kt
• First peak: 9z 9.8 (9 hrs), 916.2 mb, 147.9 kt
• Second peak: 21z 9/10 (69 hrs), 928.0 mb, 131.8 kt

Thanks man, this is a great overview.

Please consider doing it again at 12Z to see how it compares

[SouthFLTropics]

The general rule always used to be that wherever the landfall is 6-10+ days out, that’s usually not where it’s going. So those kind of shifts would not shock me.

Yeah, that was, and possibly still is the rule. The best place to be 6 to 10 days out is right in the bullseye. We still have much to learn about a science that is constantly in motion. Can anyone say chaos theory.


Sent from my iPhone using Tapatalk

I agree with the larger point about forecast uncertainty 6-10 days out, particularly with models. But we only need to look back a couple of weeks to see a storm (Idalia) that made landfall near the long range NHC track.

I can't argue with that point. The NHC has gotten damn good at doing their job over the years.

[tolakram]

These runs are getting really goofy at the end I cannot lie.

I like the end result of these runs if it keeps Lee off the East Coast!

Sure, but I csnt stand it from a forecasting standpoint.

This is downright embarrassing in 2023.

You're using the long term modelling wrong. Past 5 days accuracy falls off a cliff, the amount of raw data needed to predict that far out is bigger than huge. One small error, and there are many due to not sampling 100% of the atmosphere every second, just keeps compounding the further out it goes. Regardless of the quality of the model there is simply no way to be super accurate past 5 days. Then add to the fact that the further north you get the less data is available ... good luck!

[syfr]

But we only need to look back a couple of weeks to see a storm (Idalia) that made landfall near the long range NHC track.

That was a remarkably accurate prediction especially given a storm that was virtually stationary at the time.

[wxman57]

This Univ. of OK PhD student, Tomer Burg, has some excellent data on his site. http://arctic.som.ou.edu/tburg/products/realtime/tropical/?storm=AL132023

Here's a plot of all members of the GFS, EC, CMC, and UKMET ensembles. Plot identifies the chances of the center passing within 150km of a location:



Same plot, but with the operational and ensemble mean lines:



Here's a plot showing a trend of the ensembles for the past 4 runs. Clearly, the trend is farther east of the U.S.

[cycloneye]

 https://twitter.com/DerekOrtt/status/1700118073573794009

[invest man]

https://twitter.com/DerekOrtt/status/1700118073573794009

Question. So then can these models be considered useful for the track? Or should we consider anything passed the NHC forecast of no use for anything other than wishful thinking or entertainment purposes and say again the whole east coast is not safe until it is safe? Along with the slow down, is it possible for it to get trapped south of Hatteras- Bermuda to east of Bahamas way longer than the 6+ day tracks are showing?

[REDHurricane]

https://twitter.com/DerekOrtt/status/1700118073573794009

Question. So then can these models be considered useful for the track? Or should we consider anything passed the NHC forecast of no use for anything other than wishful thinking or entertainment purposes and say again the whole east coast is not safe until it is safe? Along with the slow down, is it possible for it to get trapped south of Hatteras- Bermuda to east of Bahamas way longer than the 6+ day tracks are showing?

At this point I wouldn't say that anything is impossible. Something about this setup really seems to be confusing all of the models, so no one actually knows where this storm will end up going 7+ days in the future with any certainty

[TomballEd]

The models seem to generally do well, sometimes 50 miles wrong, so that someone doesn't get their sensible weather and cries 'bust', within a few days of a NE USA snowstorm. I lived near NYC long enough my wife knew the KU monographs was a great birthday present. NE snowstorms can pump the downstream ridge and increase amplitude and decrease wavelength. Within a few days the 'bust' can be 50 miles, if it moves the R/S line to the 'wrong' side of a big city.

I understand understanding how strong a storm is essential to understanding how it might change the steering around it, but I'd be surprise if Lee hits CONUS anywhere except near Canada border Maine.

Tropical Tidbits recon map suggests weaker winds in a small moat-ish feature, or there is a big max eyewall, a small min, and another relative maximum. Or I think the higher pressure this morning may be an EWRC.

[rickcorvin]

https://twitter.com/DerekOrtt/status/1700118073573794009

Question. So then can these models be considered useful for the track? Or should we consider anything passed the NHC forecast of no use for anything other than wishful thinking or entertainment purposes and say again the whole east coast is not safe until it is safe? Along with the slow down, is it possible for it to get trapped south of Hatteras- Bermuda to east of Bahamas way longer than the 6+ day tracks are showing?

Keep in mind that Derek Ortt regularly trashes the model forecasts, particularly the GFS. He does it for every storm. Models are useful tools, but not perfect. Look to the NHC and ensembles 5+ days out.

[Hybridstorm_November2001]

Hmmm, curiouser and curiouser, overnight/early this morning the GFS runs had Lee arriving in Nova Scotia on Friday, September 15th while Euro runs had Lee arriving or going just south of Nova Scotia an entire day later Saturday, September 16th. Still a great deal of uncertainty here.

[Hybridstorm_November2001]

GFS shifted to the East again. With Lee weakening already, some great signs already in the works.

The 06z literally landfalls slightly W of the 0z and is the polar opposite of the million mile shifting Euro.

It seems clear that the interaction with Margot is the key here. Euro has them close and that pulls Lee out to see - but also the models dont really do well with these interaction scenarios. GFS has them far enough part that they dont really interact, so it landfalls, and Margot continues on to pull Nigel up to the central Atlantic instead of allowing it to move into a similar position as Lee is now.

Indeed, the Margot-Lee interaction may save any mainland area from a direct hit, here is hoping.

[PavelGaborik10]

I like the end result of these runs if it keeps Lee off the East Coast!

Sure, but I csnt stand it from a forecasting standpoint.

This is downright embarrassing in 2023.

You're using the long term modelling wrong. Past 5 days accuracy falls off a cliff, the amount of raw data needed to predict that far out is bigger than huge. One small error, and there are many due to not sampling 100% of the atmosphere every second, just keeps compounding the further out it goes. Regardless of the quality of the model there is simply no way to be super accurate past 5 days. Then add to the fact that the further north you get the less data is available ... good luck!

I'd concur IF this was a regular mishap, which we typically see in the long range

What we do not typically see is storms shift thousands upon thousands of miles in the span of a single run, be it 7 days or not. There are plenty of well respected mets on social media who concur with me as well, Derek in particular blames the current resolution of the Euro and deems the Euro leaping Margot across the entire Atlantic in the span of an entire run completely and totally unacceptable.

I'm not looking for anywhere near perfection but even a slight amount of consistency would be nice.

I'm in agreement with him, it's horrendous modeling :

 https://twitter.com/DerekOrtt/status/1700038744122819029




 https://twitter.com/DerekOrtt/status/1700040196165718309




 https://twitter.com/DerekOrtt/status/1700041220137885939




The only way to "properly" utilize long range modeling that's this poor is to throw it in the garbage :

[url]blob:https://www.tropicaltidbits.com/354a6a8f-fd17-464e-885c-4ed429b29455[/url]

And it starts with its complete and utter failure with Margot in the short to medium range, that's what leads to this mess.

[Spacecoast]

Perhaps Derek can plot a 10day forecast for Lee, and subject himself to actual verification.