What is the most reliable computer model

[Jim Cantore]

there was a poll on which was your fav but this is a question on which you find most reliable

[gkrangers]

None of those.

[yoda]

None of those.

[JTD]

From a NHC discussion about Frances by Avila last year:

NOW THAT THE RELIABLE
GFS AND GFDL ARE IN AGREEMENT WITH THE OTHER MODELS...THE
CONFIDENCE IN THE OFFICIAL FORECAST IS HIGHER.

So the GFS and GDL are best.

[DoctorHurricane2003]

really depends on what you are looking at

[deltadog03]

GFS=HAHAHA...its horrible

[yoda]

From a NHC discussion about Frances by Avila last year:

NOW THAT THE RELIABLE
GFS AND GFDL ARE IN AGREEMENT WITH THE OTHER MODELS...THE
CONFIDENCE IN THE OFFICIAL FORECAST IS HIGHER.

So the GFS and GDL are best.

Not always. The GFS is doing horrible with Irene. The GFDL always has trouble with the strength of a system.

[huricanwatcher]

From a NHC discussion about Frances by Avila last year:

NOW THAT THE RELIABLE
GFS AND GFDL ARE IN AGREEMENT WITH THE OTHER MODELS...THE
CONFIDENCE IN THE OFFICIAL FORECAST IS HIGHER.

So the GFS and GDL are best.

Not always. The GFS is doing horrible with Irene. The GFDL always has trouble with the strength of a system.

but its accuracy and point of landing is usually correct? ... just an opinion...

[x-y-no]

No one model is best in all situations.

If I were forced to pick one, though, I'd probably pick the European, which you did not include in your poll.

[gkrangers]

No one model is best in all situations.

If I were forced to pick one, though, I'd probably pick the European, which you did not include in your poll.

Agree.

[mahicks]

Can we get the mods to add Clipper and Xtrap??

These are by far the most accurate models we have available today.

GFS=Good For S..........

[BUD]

Can we get the mods to add Clipper and Xtrap??

These are by far the most accurate models we have available today.

GFS=Good For S..........

[clfenwi]

I interpret 'reliable model' to be one that isn't necessarily the most accurate, but the one least prone to offer cracked out forecasts, or forecasts that prove to be grossly wrong. By that measure, BAM, GFS, CAN, and UKMET are all tossed. I can't think of a terribly blown GFDL forecast (besides premature deaths) and it has offered some good forecasts over the life of a storm, but I'll throw it out thanks to guilt by association (with the GFS). That leaves NOGAPS as the only remaining choice in this poll. Maybe someone can point out a storm that NOGAPS did horrible with or remembers a specific situation in which NOGAPS offered an utterly ridiculous solution, but at the moment, I can't think of/am not aware of one.

[kmanWX]

there was a poll on which was your fav but this is a question on which you find most reliable

Within your poll Mr. Floyd I simply can not answer your question. Because there is no 'most' reliable computer model when it comes to inputting all sorts of variables. However there are somewhat pretty reliable models and that is the GFS and GFDL models. Even those two models in some cases turn out to be unreliable.

Some computer models could draw up the right situation however it's hypothesis is utterly nonsense.

[WindRunner]

We all know the MM5 is the best, admit it

[caneflyer]

Well, NHC posts verification reports. Last year's is here:

http://www.nhc.noaa.gov/verification/pdfs/2004.pdf

The FSU superensemble had the lowest average errors last year. Here is something else from their web site - a plot of model performance year by year:

http://www.nhc.noaa.gov/verification/fi ... th_TDs.gif

[mahicks]

We all know the MM5 is the best, admit it

LOL...I'll add that one to my "most accurate" section with clipper and xtrap...LOL

MM5 sure is great for anyone that wants a storm to develope... The only thing that bothers me about that statement is currentyly MM5 isn't developing anything right now!

I guess it's a slight lull.

[mahicks]

Well, NHC posts verification reports. Last year's is here:

http://www.nhc.noaa.gov/verification/pdfs/2004.pdf

The FSU superensemble had the lowest average errors last year. Here is something else from their web site - a plot of model performance year by year:

http://www.nhc.noaa.gov/verification/fi ... th_TDs.gif

The FSU Super IS the best....It's a weighted "average" of the other models. It is NOT a model itself.. Therefore if you accurate weigh and average the other models, you can pretty much be the "best"

I guess a good analogy of the FSU super would be the NHC's usual track? An "average" of the other models + good common sense + stick your head out the window weather prediction=NHC[/u][/i]

[curtinnc]

Exactly right, the FSU product is that, an ENSEMBLE of the driving factors computed by the many models, and it's tailorable in a way. Has no particular single weakness...

[Derecho]

The FSU Superensemble incorporates the previous NHC forecast track, so it's not really a pure model.

Also, NHC notes that they don't get it in time to actually use it for forecasts most of the time, which I thought was interesting.

The accuracy of GUNA (which is a simple average of the GFDL, UKMET, NOGAPS, and GFS) is so close to the FSU Supensemble that the difference doesn't matter much, and the public can get a look at it (with the appropriate software to average the model tracks) unlike the FSU.

People need to stop thinking about having a favorite single "model" and recognize that consensus models are the promised land. Most of the time the NHC forecast tracks are close to or even identical to GUNA.