ATL: MARIA - Models

[jlauderdal]

No SE Florida hit this run, although it did shift a bit closer to the west.

the gfs historically underestimates the ridge, we saw it as recently as irma, the bottom line is euro and gfs are showing a potential major hurricane approaching the united states, plenty of time to see how the pattern develops but the gulf to maine to ots is in play..what seems clear is development of another solid ACE producer

[Nimbus]

Models showing a stronger system effecting the islands especially Puerto Rico, strong side of a 977 mb storm. Ridging further west is currently depicted fairly far north but with the forecast 8 or 9 days out all we can gather from that is a high probability of Maria missing the weakness left by Jose.

[Ken711]

Looks like the Outer Banks and mid-Atlantic will be the target for these future runs.

[lando]

At work so cant post pic but 06 gfs shifts w with sc strike now, really focusing on that area between n Florida and nc

[GCANE]

[TheStormExpert]

00z EPS Ensembles.

[Digital-TC-Chaser]

ATCF

[SuperMarioBros99thx]

00z GFS run has landfall in Hispaniola (10/11 past GFS runs have had landfall in Hispaniola):

So this is will be a more Hurricane David re-run rather than Ivan re-run.

[lando]

ATCF

What is atcf

[AdamFirst]

What is atcf

Automated Tropical Cyclone Forecast

I believe it uses data from the SHIPS intensity guidance but I could be wrong

[bob rulz]

At this point I am most concerned for the northern Leewards. Another hit on Barbuda or St. Martin or the Virgin Islands would be unbelievably devastating.

[Gustywind]

At this point I am most concerned for the northern Leewards. Another hit on Barbuda or St. Martin or the Virgin Islands would be unbelievably devastating.

Unrealistic, cannot imagine that!

[jksc]

Is there a site to be able to view the HCCA model (HFIP corrected consensus) that NHC frequently mentions and seems to be one of the better performers for track?

[RL3AO]

Is there a site to be able to view the HCCA model (HFIP corrected consensus) that NHC frequently mentions and seems to be one of the better performers for track?

Not that I've seen. Here's the most information I've found about it. It appears to be creating a bias-corrected forecast based off of past model runs and what type of errors they had. I'm this is probably a big reason why NHC went so heavily with the Euro during Irma. The bias correcting HCCA was weighing the Euro so heavily since it was so much more accurate.

http://www.hfip.org/events/annual_meeting_jan_2017/presentations/Day1/1430PM-Simon_Penny_01112017_Final.pdf

[ronjon]

Is there a site to be able to view the HCCA model (HFIP corrected consensus) that NHC frequently mentions and seems to be one of the better performers for track?

Not that I've seen. Here's the most information I've found about it. It appears to be creating a bias-corrected forecast based off of past model runs and what type of errors they had. I'm this is probably a big reason why NHC went so heavily with the Euro during Irma. The bias correcting HCCA was weighing the Euro so heavily since it was so much more accurate.

http://www.hfip.org/events/annual_meeting_jan_2017/presentations/Day1/1430PM-Simon_Penny_01112017_Final.pdf

I figured that was the case. I understand the reasoning. Go with the model thats performed the best. And I can see that for the days 3-5 forecast but for the 24 hour forecast is there really that much error between UKMET, GFS, and ECM? Wouldn't a consensus approach be better?

[kunosoura]

Is there a site to be able to view the HCCA model (HFIP corrected consensus) that NHC frequently mentions and seems to be one of the better performers for track?

Not that I've seen. Here's the most information I've found about it. It appears to be creating a bias-corrected forecast based off of past model runs and what type of errors they had. I'm this is probably a big reason why NHC went so heavily with the Euro during Irma. The bias correcting HCCA was weighing the Euro so heavily since it was so much more accurate.

http://www.hfip.org/events/annual_meeting_jan_2017/presentations/Day1/1430PM-Simon_Penny_01112017_Final.pdf

Really cool presentation, thank you for sharing that. Would have loved to have seen the talk that went with it. Your regular posting of model forecasting error vs. forecast time for recent systems made me wonder if weighting models differently due to their track records would lead to a more finely-tuned, continuously correcting TVCN. And it looks like that is the case!

I don't follow relative model performance across TC systems very much except to notice that the Euro is frequently at the top of the pile and that when the UKMET is right, it's really right.... Do you know if anyone is using this HCCA approach but restricting the training phase only to previous forecasts for the specific system being modeled? As in HCCA forecasts for 96L, for example, use training data sets that have only come from 96L forecasts? Obviously a TC would need to run for some days to get a useful training set, but for a long tracking CV system this could be cool. However I could possibly see having such a small training set leading to feedback issues and undamped oscillations in HCCA output, maybe. Just thinking out loud.

This sort of thing always interests me. Thanks for your contributions.

[toad strangler]

12z GFS coming in N of 6z early on

[gatorcane]

12z GFS coming in N of 6z early on

And stronger

[toad strangler]

Direct hit on PR this 12z GFS run

[Blown Away]

12z faster than previous runs...