Im Impressed with the GFDL Model this Year

[gatorcane]

Just remember the GFDL (and AVNS) models 2-3 days ago called for this NW movement of Ernesto up into Cuba and then a turn WNW through Cuba or just north (exactly what it is doing now) and then a gradual turn to the NW and then N over the eastern GOM with a hit near or just north of Tampa.

The GFDL also nailed Chris and predicted it to just go poof (lost it). I saw some other think the GFDL was also doing well this year.

So far the GFDL has been an excellent model this year. Thoughts?

[StormWarning1]

Your picking just one run of the GFDL. It has been all over the place with Ernesto,
Texas, New orleans, Mobile, Fla. West coast.

[Cyclenall]

I have to agree, the GFDL has been excellent. Ioke was amazing since the first 9 days of Hurricane/Typhoon Ioke's life, the GFDL just NAILED it to the point. I'm not sure about right now, but the GFDL really surprised me at how well it did.

[Downdraft]

Can't blame the GFDL for being all over the place with this storm everyone else has too. A whole bunch of people are going to be really embarassed this year publishing their verifications for this system. Can't say I blame them though the weaker the system the harder it seems to forecast. Sometimes a little humble pie is just the ticket to improve next time. So it's not all bad news now is it?

[Patrick99]

It seems to me that the GFDL frequently gets the "right idea," even if the actual landfall/path is a bit off. Similar to last year with Katrina - Katrina didn't go as far south as the GFDL predicted, but the model did sniff out the SW turn. Same here - the model did sniff out this general motion with a much more northerly track initially than the other models.

[dougjp]

If you mean in comparative terms, then perhaps yes.

Although for this Hurricane going near Jamaica and then touching the extreme west of Cuba on its way into the Gulf as a Cat 2/3, this Tropical Storm about to reach south Florida is not an example of where I would dredge up the term "impressed" to be used in any context. Embarrasment or ineffectual seem more appropriate terms to me.

[Jim Cantore]

GFDL is one of the better models with the tropics I've found, the biggest botch I've seen was with Wilma last year.

[clfenwi]

It's the GFS, not the GFDL that rocked the house as far as the landfall forecast goeswith this storm. Remember that 12Z run on Saturday?

A refresher

IT SHOULD BE NOTED THAT THE GFS SOLUTION OF
BRINGING THE LOW-LEVEL CENTER OF ERNESTO TO SOUTH FLORIDA LOOKS
DUBIOUS

Turned out to be not so dubious. As one can see in the graphic, the GFDL was forecasting a panhandle hit.

Flipping back a day earlier, the GFDL did have the better short term forecast, but lost out after that. From that point on this was the GFS' storm.

[THead]

It's the GFS, not the GFDL that rocked the house as far as the landfall forecast goeswith this storm. Remember that 12Z run on Saturday?

A refresher

IT SHOULD BE NOTED THAT THE GFS SOLUTION OF
BRINGING THE LOW-LEVEL CENTER OF ERNESTO TO SOUTH FLORIDA LOOKS
DUBIOUS

Turned out to be not so dubious. As one can see in the graphic, the GFDL was forecasting a panhandle hit.

Flipping back a day earlier, the GFDL did have the better short term forecast, but lost out after that. From that point on this was the GFS' storm.

Its a tough business forecasting these things. I wonder if in the weatherman locker room, some co-worker makes 100 copies of that quote and wallpaper the guy's locker with it.

[Thunder44]

Just remember the GFDL (and AVNS) models 2-3 days ago called for this NW movement of Ernesto up into Cuba and then a turn WNW through Cuba or just north (exactly what it is doing now) and then a gradual turn to the NW and then N over the eastern GOM with a hit near or just north of Tampa.

The GFDL also nailed Chris and predicted it to just go poof (lost it). I saw some other think the GFDL was also doing well this year.

So far the GFDL has been an excellent model this year. Thoughts?

I disagree. The GFDL did not nail Chris. It was awful in track and intensity. And you can't use just a couple model runs to judge the performance of a model.

[clfenwi]

Grr... with the days running together, I got the days mixed up...
this is the 12 Z Saturday run, that the "dubious solution" quote was referring to:

http://euler.atmos.colostate.edu/~vigh/ ... 082612.png

and here's a day later with the GFS about dead on, with the others out to the west

http://euler.atmos.colostate.edu/~vigh/ ... 082712.png

And in fairness to Beven, there wasn't exactly a flood of people on this board hailing the GFS solution; it was met with much derision.