ATL: FLORENCE - Models

[Ken711]

You should prefer the blend of models, EC, GFS, UKMET, with EC having a slightly heavier weight. To say the GFS is useless is extremely incorrect.

How are accurate is the CMC with regard to hurricane tracks, could the NHC use it in the blend?

It's not. I think it is in the blend, probably just not weighted very high. The main model used is the Florida super ensembles, which we don't have access too. My understanding is it processes all the ensembles of most of the models and has some kind of scoring system as to which ones do better, which changes day to day, year to year. It always scores near the top in track accuracy. TVCN utilizes it I think, not sure.

Thanks.

[chris_fit]

The NHC does not let Flo go below 24.9 - I think this will be a good benchmark to see if that verifies

So I said that about 20 hours ago...... at 5am Flo was @ 24.5N (and still appears to be going S of due W) - Which is pretty much where the UKMET had her, i don't think any other model other than a few ECMF Ensembles dropped her that low. Might be just a wobble, might be something to watch for, I don't know. Just making an observation.

[jlauderdal]

It could be right and they all start trending OTS. GFS has been right a few times recently to score a win.

Since last year the GFS has had a severe right bias to the track (similar to the NAM in fact), Irma last year was another good example where it kept trying to recurve it out to sea when it was moving SW and even the 48-72 hour forecast had it something like 150mi east of Florida. The FV3 and CMC have consistently had more realistic scenarios track-wise.

Why does it still maintain a severe right bias compared to the other models? What factors do you think contribute to that? I have read here many posts about the GFS's underestimated the strength of ridges, is that part of it.

the atmosphere is so complex that even the modeling cant resolve it all as we would like..its like googles algorithm, its good and getting better at delivering a great search result but its far from perfect and has certain bias because of the equations and how they weight things such as recency of updates to a site, bounce rate, time on site, and many other parameters, those are decisions a human is making as to what is more important than something else and then having the algorithm return results...in the gfs case it has proven to have some right bias and nhc knows that and every 6 hours they have to take this and everything else into account and make a track but this is why the forecast discussion is the first thing to look at it even before looking at their forecasts because you know what they are using to make these decisions and often times they will give you a heads up to where the bodies are buried in the track and intensity, you have to read the discussion to truly understand the forecast and that goes for local weather too. Also, error rate and this is extremely important, the cone isnt some random thing they draw up with a color marker...its all about statistics so to the newbies on the board you have to understand track error and why you will hear people like me and others say its better to have the nhc track on your doorstep at 5 days then on either side of the line, i have been in the cone, landfall whatever you want to call it many times on the the models and the nhc track and have had a total of three "major events" Wilma, Katrina and Irma in 20 years. If you are in the cone at 5 days you better be ready and if you are in the cone at 3 days you better have food and water and everything else because nhc has really nailed track inside of 72 hours the last few years, really nailed it!!

https://www.nhc.noaa.gov/verification/

[jlauderdal]

Seen this movie before, some models breaking down a ridge or lessening its SW extent too much and then the TC gets much further west. If there was some longwave trough swinging off the east coast you could trust it but in this scenario if I lived from Central Florida thru the Carolina's I'd spend the weekend getting some things squared away.

dean simplified everything in the post..good work

[invest man]

It just seems to me that the southerly route has storm slightly weaker than NHC projection of strength, whereas the stronger models shows a more northerly projection toward the coast. That being said if we took a blend of Euro and GFS outputs blend track to consider strengths and at this point landfall could be along NC Coast between Wilmington and Morehead City. . I understand the ridge but reading the NHC discussion does slightly imply that the western perimeter of the ridge by d3-5 might break down so slightly. Also read on many posts here back when Flo made her run toward cat 4 earlier this week how everyone thought the up and out had begun because of the NW movement at the time. Lots going on with this and a lot of things to consider. No pro here just an observer. So where is my thoughts wrong!

[p1nheadlarry]

Since last year the GFS has had a severe right bias to the track (similar to the NAM in fact), Irma last year was another good example where it kept trying to recurve it out to sea when it was moving SW and even the 48-72 hour forecast had it something like 150mi east of Florida. The FV3 and CMC have consistently had more realistic scenarios track-wise.

Why does it still maintain a severe right bias compared to the other models? What factors do you think contribute to that? I have read here many posts about the GFS's underestimated the strength of ridges, is that part of it.

the atmosphere is so complex that even the modeling cant resolve it all as we would like..its like googles algorithm, its good and getting better at delivering a great search result but its far from perfect and has certain bias because of the equations and how they weight things such as recency of updates to a site, bounce rate, time on site, and many other parameters, those are decisions a human is making as to what is more important than something else and then having the algorithm return results...in the gfs case it has proven to have some right bias and nhc knows that and every 6 hours they have to take this and everything else into account and make a track but this is why the forecast discussion is the first thing to look at it even before looking at their forecasts because you know what they are using to make these decisions and often times they will give you a heads up to where the bodies are buried in the track and intensity, you have to read the discussion to truly understand the forecast and that goes for local weather too. Also, error rate and this is extremely important, the cone isnt some random thing they draw up with a color marker...its all about statistics so to the newbies on the board you have to understand track error and why you will hear people like me and others say its better to have the nhc track on your doorstep at 5 days then on either side of the line, i have been in the cone, landfall whatever you want to call it many times on the the models and the nhc track and have had a total of three "major events" Wilma, Katrina and Irma in 20 years. If you are in the cone at 5 days you better be ready and if you are in the cone at 3 days you better have food and water and everything else because nhc has really nailed track inside of 72 hours the last few years, really nailed it!!

https://www.nhc.noaa.gov/verification/

Wonder if it has to do with how it handles the tropics since the Rossby number asymptotes at the equator

[Ken711]

It just seems to me that the southerly route has storm slightly weaker than NHC projection of strength, whereas the stronger models shows a more northerly projection toward the coast. That being said if we took a blend of Euro and GFS outputs blend track to consider strengths and at this point landfall could be along NC Coast between Wilmington and Morehead City. . I understand the ridge but reading the NHC discussion does slightly imply that the western perimeter of the ridge by d3-5 might break down so slightly. Also read on many posts here back when Flo made her run toward cat 4 earlier this week how everyone thought the up and out had begun because of the NW movement at the time. Lots going on with this and a lot of things to consider. No pro here just an observer. So where is my thoughts wrong!

Maybe the 06Z GFS is the start of a north trend.

[Hypercane_Kyle]

What a joke the gfs has become...I've been watching these models for 23 years and I think it is on par with what it could do 20 of those years ago.

What a disgrace. It once again is breaking the damn ridge down too fast.

Please pay attention to the the trend of the gfs, ecmwf, ukmet and cmc that all mostly discount this joke of a model run.

It could be right and they all start trending OTS. GFS has been right a few times recently to score a win.

Since last year the GFS has had a severe right bias to the track (similar to the NAM in fact), Irma last year was another good example where it kept trying to recurve it out to sea when it was moving SW and even the 48-72 hour forecast had it something like 150mi east of Florida. The FV3 and CMC have consistently had more realistic scenarios track-wise.

More than since last year, the GFS also did this with Matthew and Sandy, the latter being a massive failure as the ECMWF correctly showed Sandy bending westward into New England while the GFS showed it moving safely OTS until the last possible minute. Definitely tends to send systems more poleward.

[Blown Away]

Don't lose sight how amazing these models are, Flo is still >1500 miles from CONUS and they have narrowed to SC to OBX consistently. Still and likely subject to change, but the NHC 5 day track is usually pretty solid these days. Unfortunately they continue to report low confidence after day 5. I think the recon samples out in front of Flo this weekend will tighten things up.

[p1nheadlarry]

It could be right and they all start trending OTS. GFS has been right a few times recently to score a win.

Since last year the GFS has had a severe right bias to the track (similar to the NAM in fact), Irma last year was another good example where it kept trying to recurve it out to sea when it was moving SW and even the 48-72 hour forecast had it something like 150mi east of Florida. The FV3 and CMC have consistently had more realistic scenarios track-wise.

More than since last year, the GFS also did this with Matthew and Sandy, the latter being a massive failure as the ECMWF correctly showed Sandy bending westward into New England while the GFS showed it moving safely OTS until the last possible minute. Definitely tends to send systems more poleward.

Add Joaquin to the list

[Hypercane_Kyle]

Take it for what it's worth as the FV3 hasn't exactly done a skillful job of predicting Florence's track either, but the 06z FV3-GFS shows Florence on a path pretty much identical to the ECMWF.

[TheStormExpert]

I’m going to go with landfall somewhere near the NC/SC border next Thursday. Intensity is anyone’s guess but wouldn’t rule out a Cat.4.

[xtyphooncyclonex]

What a joke the gfs has become...I've been watching these models for 23 years and I think it is on par with what it could do 20 of those years ago.

What a disgrace. It once again is breaking the damn ridge down too fast.

Please pay attention to the the trend of the gfs, ecmwf, ukmet and cmc that all mostly discount this joke of a model run.

It could be right and they all start trending OTS. GFS has been right a few times recently to score a win.

The GFS literally plowed Flo into a ridge for several runs. It keeps on depicting a powerful ridge break apart; I believe even the southernmost run isn't the furthest west, and it still underestimates the ridge.

They can't score wins in storms like this. Seriously, plowing this through a ridge (and then eventually turning back)? I'm no met and I don't mean to be offensive but that sounds ridiculous. Pure entertainment---trash run from 06Z GFS.

[NDG]

Below is what last night's Euro Ensembles showed, so the majority of them are concentrating in a smaller area over South Carolina. The members that showed a weaker Florence in the short term are the ones that showed a track towards FL & GA, the ones that showed a much stronger Florence in the short term are the one tracking towards NC.

17 members into FL & GA
18 members into S.C.
8 members into N.C. & outer banks.
7 members OTS or meander offshore.

[BobHarlem]

FV3 GFS is an odd one again, Hits near Wilmington, then loops out over the outerbanks then back and hits Wilmington a second time and then rides the coast up and out.

[xtyphooncyclonex]

Btw, here's the steering assuming the storm strengthens

[p1nheadlarry]

FV3 GFS is an odd one again, Hits near Wilmington, then loops out over the outerbanks then back and hits Wilmington a second time and then rides the coast up and out.

Almost like Diana in 1984

[ronjon]

Don't lose sight how amazing these models are, Flo is still >1500 miles from CONUS and they have narrowed to SC to OBX consistently. Still and likely subject to change, but the NHC 5 day track is usually pretty solid these days. Unfortunately they continue to report low confidence after day 5. I think the recon samples out in front of Flo this weekend will tighten things up.

Sampling the ocean environment will help, but ultimately its the most accurate depiction of the Pacific coast trough and its evolution that will better predict the strength and orientation of the western atlantic ridge 5-6 days from now. Think sampling plays its most important role 2-3 days from landfall around the storm.

[ronjon]

Ukmet ensembles and CMC are showing strongest ridging so far. Interesting if Euro stays consistent with its SC landfall at 12z.

[tolakram]

What a joke the gfs has become...I've been watching these models for 23 years and I think it is on par with what it could do 20 of those years ago.

What a disgrace. It once again is breaking the damn ridge down too fast.

Please pay attention to the the trend of the gfs, ecmwf, ukmet and cmc that all mostly discount this joke of a model run.

It could be right and they all start trending OTS. GFS has been right a few times recently to score a win.

The GFS literally plowed Flo into a ridge for several runs. It keeps on depicting a powerful ridge break apart; I believe even the southernmost run isn't the furthest west, and it still underestimates the ridge.

They can't score wins in storms like this. Seriously, plowing this through a ridge (and then eventually turning back)? I'm no met and I don't mean to be offensive but that sounds ridiculous. Pure entertainment---trash run from 06Z GFS.

No it did not. Show me the map where it plowed a storm into a ridge. It's silly to think any model can't function as designed. It may break down a ridge too fast but it's not going to plow a storm into a ridge. Some people here are misreading the maps.