ATL IKE: Models Discussion

[JenBayles]

AFM, I need to thank you for your very thoughtful model analysis here. When the official track went south of Corpus, I knew it was bogus based on your knowledge and explanation of what models handle what atmospheric features the best or worst. At least I got my immediate family and friends to keep an eye on it instead of writing off a HOU/GAL threat. Thanks for hanging around and putting up with our constant questions!

[curtadams]

I'm going to bat for Avila and the Baffin Bay call. The NHC has a policy of making each track a blend of the prior track and the current models. Because of the noise factor in model runs, this is theoretically the way to maximize accuracy. It also obviously damps down variability. The NHC also gains credibility with the public by damping down swings in forecast paths, even to the point of reducing accuracy by overdamping, since swings tend to make the public throw up their hands and say "nobody knows what's going on". The NHC *overdamps* in the accuracy sense, as shown by the fact that CONU (model consensus) is more accurate than NHC up to 48 hours (if the NHC had accuracy-maximizing damping, it would be *more* accurate than CONU due to noise suppression.)

Avila's call was in line with NHC practice and as such, likely moved too *little* far south in the strict sense of making the most accurate prediction with the then-available data. Houston/Galveston remained in the cone and if EOM was distracted by the line moving away from them I agree they need (and are likely about to get) an attitude adjustment. But if the officials had done their job and called for mandatory evacs of places like Galveston in time for complete evac, - when it WAS in the cone - we would be fine. Some of the public would have been casual because of the line being near CRP but enough would been serious that a complete evac would be possible now. It wasn't Avila's call that put us in the situation of an incomplete evac of Galveston with a significant possibility of a hurricane strike that could overwhelm even that city's formidable hurricane defenses (as well as similar situations throughout NE Texas.)

In short, the NHC has the right policy on calls. The fact that an even more extreme damping policy would have produced a better call *in this particular case* doesn't change the fact that in general the NHC policy is better. There was no way for Avila to know that this time a generally inferior strategy of extreme overdamping would have produced a better result than the generally superior strategy of mild overdamping.

[haml8]

Any news on the models now... GFDL, GFS runs?? I have not seen much.. Maybe the Florida guys can help and pitch in to help us Texans out that can do research cause we are trying to get ready Any help you guys provide is appreciated!

[dwg71]

Any news on the models now... GFDL, GFS runs?? I have not seen much.. Maybe the Florida guys can help and pitch in to help us Texans out that can do research cause we are trying to get ready Any help you guys provide is appreciated!

GFDL shifted south the last run to around just north of palacious.

GFS has budged from middle texas coast.

HWRF is just about Port Lavaca south of Matagorda Bay.

[haml8]

Any news on the models now... GFDL, GFS runs?? I have not seen much.. Maybe the Florida guys can help and pitch in to help us Texans out that can do research cause we are trying to get ready Any help you guys provide is appreciated!

GFDL shifted south the last run to around just north of palacious.

GFS has budged from middle texas coast.

HWRF is just about Port Lavaca south of Matagorda Bay.

Thanks for the update! Waiting for the updates on weatherunderground.... they are taking forever to update..

[Sanibel]

Well the benefit of being an internet hurricane enthusiast (yes, there are other words used to describe that) is I can clear all that NHC track business out of my mind and concentrate only on what I see as the best detectors of track. Which is why I was staying to a simple observation of GFDL and its performance. I watch GFDL because if you learn its tendencies it very often comes close to registering the track in advance. Pro mets and NHC have more responsibility I admit. There's no real argument here, NHC's models tend to read synoptics in real time and the near future and apply longer term tracks to the present synoptic. As Ike showed us it almost always changes in the three day plus range, which is why you don't see me making any landfall calls far out. The plunging ridge was undeterminable in advance, as well as the depth of the weakness in between the Highs or the advance of the Pacific trough. By the way the GFS observation I made last night and was hammered for appears to be coming true. But for my limited technique I'll stick to seeing how GFDL performs on its landfall prediction and how that lines up with its behavior over the last 60 hours. Gus was easier to predict because it was under a firm High. Ike is in a little bit more of a tricky synoptic - but coming into NHC's accuracy range now. The wild card is always poleward with an approaching front.

[Bolebuns]

Somebody help me out here...

The NHC track just shifted east once inland, yet all the models seem to be west of the NHC track. What is causing that? Are models delayed to us?

[Sanibel]

Don't slam me for this, but I saw that too and thought NHC is 'leading' what they expect the storm to do against the models on final approach. It would make sense since the NHC track is rightward in front of the trough. The models have already pulled NHC too far south before. NHC said it was a blend of HWRF and GFDL I believe.

[rtd2]

Somebody help me out here...

The NHC track just shifted east once inland, yet all the models seem to be west of the NHC track. What is causing that? Are models delayed to us?

Could be picking up on any slowing of the systems fwd speed

[x-y-no]

Somebody help me out here...

The NHC track just shifted east once inland, yet all the models seem to be west of the NHC track. What is causing that? Are models delayed to us?

I don't think it's accurate to say all the models, the ECMWF was right of NHC last I checked.

At any rate, they're applying their meteorological understanding to the situation. And note that they're close enough to the model consensus as to make little difference given the large eind-field of this storm.

[Bolebuns]

Don't slam me for this, but I saw that too and thought NHC is 'leading' what they expect the storm to do against the models on final approach. It would make sense since the NHC track is rightward in front of the trough. The models have already pulled NHC too far south before. NHC said it was a blend of HWRF and GFDL I believe.

But both HWRF and GFDL are west of NHC (If I have the latest models). And if it is because they are gauging the speed of the system, wouldn't the models pick up on that?

[dwg71]

HWRF and GFDL blended would split matagorda bay.

Ike has picked up forward speed and 3 hour heading based on VDMs is about 275.

If it can keep heading for a few more hours, HWRF and GFDL would be more accurate around matagorda bay.

[A1A]

The other models you are looking at may not be updated. i.e. I look at the swf ensemble and the NHC is 9 hours later than the more reliable others right now.

[DESTRUCTION5]

GFS Rolling...Looking much further North so far?

http://www.nco.ncep.noaa.gov/pmb/nwprod ... loop.shtml

[DESTRUCTION5]

GFS Rolling...Looking much further North so far?

http://www.nco.ncep.noaa.gov/pmb/nwprod ... loop.shtml

Looks like Ike nails Freeport in this run? Help me TX people..

[Wx_Warrior]

GFS

[Portastorm]

Yeah, looks pretty close to Freeport and what the 0z Euro showed.

Oh well, so much for my thinking we could see a shift leftward again due to the cutoff low in the Four Corners area!

[PTrackerLA]

That's certainly further north than the 06z, I'd say maybe 50-75 miles? Unfortunately it looks to absolutely nail Galveston.

[pablolopez26]

GFS shows direct Galveston hit now.

[pablolopez26]