ATL: MATTHEW - Models

[drezee]

There are major differences between the Op Euro and the ensembles. There is a very complex upstream interaction that would lead to that low near the Mid-Atlantic. Also, the upper trough depth is an outlier. I think the focus need to be on the immediate threat and not freak out about 10 day forecasts. We are pretty sure that a system will form with the potential to cause problems later. We will know more in three days....breathe...

[AtlanticWind]

To early to tell where this system will end up in 8 to 10 days, models are just not that reliable that far out.
5 days is the maximum I look for any kind of reliability.

[terstorm1012]

glad someone noticed the big cut off low forecasted to develop over the Ohio Valley. That will determine where 97L goes. Watch that.

[gatorcane]

00Z Guidance:

[Steve]

glad someone noticed the big cut off low forecasted to develop over the Ohio Valley. That will determine where 97L goes. Watch that.

It's been being discussed for several days. Initially it was a trough though. Something was going to be there in the 4-6 day period. GFS Operational had a small nose shortwave only as far south as Massachusetts on Thursday. Recently it and EC have shown a large cut off looping, progressing, retrograding, etc. the low itself was the system between the two highs. I think more important than what the low does is what the high following it does. Because it had been progged coming to the US East Coast and moving toward the SE. GFS this morning had it just sliding down to Texas as low pressure became more dominant in Eastern US. That's the big key weather feature in my opinion though the low is a wildcard.

[Hammy]

Model guidance grossly over-estimated Karl's intensity, but how do they do deeper in the tropics compared with mid-latitude?

[Alyono]

mega problem with the SHIPS output. It has massive shear over this. There is a less than zero percent chance this runs into 30 kts of shear

[tolakram]

It's a little premature to be calling an all clear for the USA when we do not even have a developed storm yet, the models have been awful this year..anyone taking these 200+ hour forecasts as written in stone need to go back and read through the model threads of the prior storms this year. Almost every storm this year went WEST of the model predictions prior to the Storm developing.

It's not an all clear but sometimes one has step back and take a look at the big picture.

A major hurricane has not hit the U.S. coastline for 11 years. Before Hermine, 11 years before the last hurricane of any intensity for Florida.

What does that suggest? In this current background state, it takes a damn near perfect setup to get a hurricane to strike the U.S. Squeaky clean.

A quick look at the 500mb map, even from yesterday, suggests that this is far from squeaky clean. Trof/closed low in the Ohio Valley. Narrow ridge that takes time to build back in (taking time is always a red flag in the model world).

Forecasts have to be made in context of climo and decadal trends. When a model latches onto a high-end signal that deviates significantly from climo and decadal trends, one must pause and think. If that signal subsequently weakens or reverses, that's usually right. It's just a probability thing: what is most likely going to be correct given what has been correct in the recent past?

I know people crave explanations for the string of good luck the US has had, but I don't think there is any scientifically valid reason. Sometimes you can flip a coin and get heads 11 times.

[wxmann_91]

It's a little premature to be calling an all clear for the USA when we do not even have a developed storm yet, the models have been awful this year..anyone taking these 200+ hour forecasts as written in stone need to go back and read through the model threads of the prior storms this year. Almost every storm this year went WEST of the model predictions prior to the Storm developing.

It's not an all clear but sometimes one has step back and take a look at the big picture.

A major hurricane has not hit the U.S. coastline for 11 years. Before Hermine, 11 years before the last hurricane of any intensity for Florida.

What does that suggest? In this current background state, it takes a damn near perfect setup to get a hurricane to strike the U.S. Squeaky clean.

A quick look at the 500mb map, even from yesterday, suggests that this is far from squeaky clean. Trof/closed low in the Ohio Valley. Narrow ridge that takes time to build back in (taking time is always a red flag in the model world).

Forecasts have to be made in context of climo and decadal trends. When a model latches onto a high-end signal that deviates significantly from climo and decadal trends, one must pause and think. If that signal subsequently weakens or reverses, that's usually right. It's just a probability thing: what is most likely going to be correct given what has been correct in the recent past?

I know people crave explanations for the string of good luck the US has had, but I don't think there is any scientifically valid reason. Sometimes you can flip a coin and get heads 11 times.

A bit OT and this is conjecture at this point, but it might have something to do with the fact it seems there's always been an ULL to either shear or steer away a storm threatening the U.S. Something related to the tendency for upper level troughs to cut off from the main jet stream and then migrate into the deep tropics as of late. Maybe related to Hadley cell changes, but I'm not an expert in that area.

Lo and behold what's the uncertainty related to 97L? An ULL cutting off from the jet stream. What steered Joaquin away? A piece of vorticity from the northeast that broke away from the main jet stream. What prevented Hermine from blowing up before FL? Shear from an ULL to the north.

11 years is about 1/5 of the satellite era so at some point you question if it's just luck. Maybe it is, I'm not sure. Perhaps it's an area of active research by scientists.

[Medtronic15]

GFS 18z with Cat 5 165mph (140 kt Haiti)

[wxmann_91]

It's a little premature to be calling an all clear for the USA when we do not even have a developed storm yet, the models have been awful this year..anyone taking these 200+ hour forecasts as written in stone need to go back and read through the model threads of the prior storms this year. Almost every storm this year went WEST of the model predictions prior to the Storm developing.

It's not an all clear but sometimes one has step back and take a look at the big picture.

A major hurricane has not hit the U.S. coastline for 11 years. Before Hermine, 11 years before the last hurricane of any intensity for Florida.

What does that suggest? In this current background state, it takes a damn near perfect setup to get a hurricane to strike the U.S. Squeaky clean.

A quick look at the 500mb map, even from yesterday, suggests that this is far from squeaky clean. Trof/closed low in the Ohio Valley. Narrow ridge that takes time to build back in (taking time is always a red flag in the model world).

Forecasts have to be made in context of climo and decadal trends. When a model latches onto a high-end signal that deviates significantly from climo and decadal trends, one must pause and think. If that signal subsequently weakens or reverses, that's usually right. It's just a probability thing: what is most likely going to be correct given what has been correct in the recent past?

I could not disagree with you more. Just because a major hurricane has not hit the USA in 11 years has absolutely nothing to do with the future of 97L. Climo can be used as a guideline and nothing more. And if you want to make a forecast based on climo please reference the storms which made abrupt north turns from the ABC islands in late September and later hit Hispanola or areas east of there. And as far as the steering pattern being depicted by the models goes it's about as clear as mud. Models have a notoriously diffult time forecasting cut off lows.

But that's what I'm doing... I'm using climo as a guideline. Forgetting models for a second, what has the past 10 years told us about the likelihood of any given disturbance making landfall in the U.S. as a hurricane? And from that base state, I can either say that (just as an example) a) yes, hurricanes do hit the U.S. often, so a major U.S. hurricane on that model is reasonable, or b) no, hurricanes almost never hit the U.S., so that seems like a suspicious outcome, and let's double check why it's showing that and how sensitive that solution is to perturbations upstream.

Hurricanes recurving from the Caribbean are not too abnormal. Flora 1963, Hazel 1954... and the GFS shows a much less sharp recurvature than does the Euro, with similar outcomes for the CONUS.

[northjaxpro]

It's a little premature to be calling an all clear for the USA when we do not even have a developed storm yet, the models have been awful this year..anyone taking these 200+ hour forecasts as written in stone need to go back and read through the model threads of the prior storms this year. Almost every storm this year went WEST of the models predictions prior to a LLC being established.

Excellent post!! I could not have said this any better!

[tolakram]

Ok,

my fault as well but let's take the discussion to the discussion thread and focus on model runs and direct discussions concerning those runs here. Thanks.

[TheStormExpert]

Question, why are the GFS Ensembles more clustered around Florida when the Operational GFS run is several hundred miles east?

[Hurricaneman]

Question, why are the GFS Ensembles more clustered around Florida when the Operational GFS run is several hundred miles east?

Probably less ridge erosion

[Hypercane_Kyle]

Impressive values out to 48 hours.

SHIPS Prob RI for 20kt/ 12hr RI threshold= 9% is 1.5 times sample mean ( 5.5%)
SHIPS Prob RI for 25kt/ 24hr RI threshold= 46% is 4.0 times sample mean (11.6%)
SHIPS Prob RI for 30kt/ 24hr RI threshold= 25% is 3.5 times sample mean ( 7.2%)
SHIPS Prob RI for 35kt/ 24hr RI threshold= 12% is 2.8 times sample mean ( 4.2%)
SHIPS Prob RI for 40kt/ 24hr RI threshold= 0% is 0.0 times sample mean ( 2.8%)
SHIPS Prob RI for 45kt/ 36hr RI threshold= 0% is 0.0 times sample mean ( 4.9%)
SHIPS Prob RI for 55kt/ 48hr RI threshold= 19% is 3.8 times sample mean ( 5.1%)

[CrazyC83]

GFS 18z with Cat 5 165mph (140 kt Haiti)

Those winds are at 850 but the resolution and real central pressure would support Cat 5.

That would make Katrina look like child's play. We'd likely be talking death tolls unheard of in the western hemisphere.

[Alyono]

Impressive values out to 48 hours.

SHIPS Prob RI for 20kt/ 12hr RI threshold= 9% is 1.5 times sample mean ( 5.5%)
SHIPS Prob RI for 25kt/ 24hr RI threshold= 46% is 4.0 times sample mean (11.6%)
SHIPS Prob RI for 30kt/ 24hr RI threshold= 25% is 3.5 times sample mean ( 7.2%)
SHIPS Prob RI for 35kt/ 24hr RI threshold= 12% is 2.8 times sample mean ( 4.2%)
SHIPS Prob RI for 40kt/ 24hr RI threshold= 0% is 0.0 times sample mean ( 2.8%)
SHIPS Prob RI for 45kt/ 36hr RI threshold= 0% is 0.0 times sample mean ( 4.9%)
SHIPS Prob RI for 55kt/ 48hr RI threshold= 19% is 3.8 times sample mean ( 5.1%)

and ignore those. The output is awful upon closer inspection. No GFS vortex, dead flat wrong shear values, etc

[floridasun78]

GFS 18z with Cat 5 165mph (140 kt Haiti)

could gfs over doing showing major hurr near Haiti that could send them stone age

[floridasun78]

i look latest gfs it look Haiti and Bahamas are doom if it right south fl and fl will get outter rain bands