OK, only one model has it going to New Orleans. Let's hope that the consensus of the other models is right, and it's heading for Mexico.

Or maybe Karl Rove is just revving up his black-killing weather machine one more time, for old time's sake.

Didn't most hurricanes last year on the same track - including those that passed over New Orleans and Florida - start with models looking like they do now, turning north as they entered the gulf?

Makes you wonder, doesn't it, about the accuracy of those weather models? Which means...hmm...when it comes to modeling the climate of the entire world for centuries into the future...

[WHOOP WHOOP WHOOP]

[sound of nightstick hitting thick skull]

Oh! Silly me, what was I thinking? Al Gore has patiently instructed us that it's totally unscientific -- practically Stone Age mysticism -- to doubt computer models, except when they're doing piddling things like predicting the track of hurricanes for the next two weeks.

Thank you, Mr. Policeman, I feel much better now. Thinking only correct thoughts. No need to stick around, really.

Carl, while it would completely delusionally to forecast the tracks of hurricanes more than a few days in advance, the climate models are on more solid ground because they are calculating large scale statistical quantities (eg, global temperature). I mean, once you have date, solar output, atmospheric composition, and maybe some info on ocean circulation and ground cover, that's really all you need to make calculations.

Karl,

Not to dispute your general point, but while it is true that the data you stipulate are "all you really need to make calculations," they don't seem to be all you need to make reasonably accurate long-term predictions. Else all the climatologists, who generally speaking have access to the same data, would be coming to the same conclusions.

Granted, the northeast quadrant of a hurricane is the strongest, but even that most easterly model track is a touch west of New Orleans. Not that it makes anyone in Morgan City (or the forgotten people who survived Rita) any happier.

Karl,

I WOULD like to dispute your general point. The dynamics of climate change are so poorly understood at this point that we cannot explain the weather we are currently experiencing, much less extrapolate temperature changes into the future.

It has been warmer than it is now in the past; what caused that? Whatever it was, it wasn't SUVs.

We have a lot of work to do before we understand how we got the weather we have now, much less how to control it into the future.

> I mean, once you have date, solar output, atmospheric composition, and maybe some info on ocean circulation and ground cover, that's really all you need to make calculations.

That may be all you need to "make calculations", but it isn't enough to make the calculations that would predict climate.

Moreover, even some of those things are not computable. For example, volcanos occasionally change the earth's albedo and atmospheric composition. There is considerable uncertainty as to when they do this, not to mention what effect it will have.

Hold your horses, Karl. Protein folding simulations take hours to get through microseconds, and we can predict into Futurama's time period? No way.

Big point:
We don't understand clouds. We can't predict their formation, color(albedo), location and certainly not their movements. Next week, I mean. We can't even measure the overall cloud cover of the world. Local factors might play significantly into all of this, as well. What a cloud covers is as important as whether it forms.

As an engineering graduate, I can't just factor in what happens to a beam if it gets loaded one way, I have to make sure I cycle test it for all stresses, even things like freezing and thawing over its life cycle. Modeling a beam is based on laws, and climate modeling is based on theories, which are hardly proven to be 100% true.

Small changes can make huge differences in models, especially due to the assumptions made which cut down the number of compensatory mechanisms, some of which we likely haven't even thought of.

Climate modelers are pulling the strings, and acting as though they have predicted the future. They've been wrong predicting up until now, and I have no doubt that my prediction of roughly similar temperatures for the next 100 years is going to have less error than all of theirs. If my job depended on a non-existant crisis, I would play it up too.

When a climate modeler can start with some defined set of conditions in the past, say 100 years ago, and have his model accurately show what actually happened, then perhaps I'll have some faith in the model's ability to predict what will happen 100 years from now. Of course, I'll want openness in his model to make sure he isn't hard-coding the results for the past 100 years. To the best of my knowledge, none of the existing climate models have passed this test. If they can't model what actually happened, how can we have any faith they can model what will happen? As it is, there are lies, damned lies, and computer models.

Karl H., it's not a bad argument, but far from certain.

In essence, you are arguing the thermodynamic limit, that when you go to the limit of very large distance and time scales, all the medium-scale fluctuations just average out, and the only things that still matter are certain basic parameters, such as the solar energy budget, total average percentage cloud cover, et cetera.

Sometimes complex systems work out that way, yes. But sometimes they don't. For example, just consider an ordinary fluid at its freezing point. If you did not know ahead of time that it would exhibit a spontaneous breaking of (translational) symmetry and form a crystal, you would never predict it. It's quite an astonishing thing to have happen, if you just know that it's made of molecules with certain interactions. Why would order on an enormous scale break out, when the interactions are so short-range? This is something you'd never predict using your thermodynamic limit argument.

It's even hard to do a computer simulation of a liquid correctly so that it freezes into a solid, unless you cheat because you know the solid forms.

I'm not saying you're wrong. But lots of experience in the physics of complex systems says your argument is fairly speculative, a priori.

At 2pm EDT (8-17), the GFDL model has moved west. In fact, so west, that landfall is south of Galveston Island.

Not only are we going to the Antilles, we're going to Puerto Rico and the Dominican Republic and Haiti and Cuba and the Caymans, and we're going to Yucatan and Texas and Louisiana. And we're going to Arkansas and Tennessee and Kentucky and Virginia, and then we're going to Washington, D.C., to tear down the White House! YEEEARRGH!

Let's hope that the consensus of the other models is right, and it's heading for Mexico.

In other news, bloggers south of the border are wishing very much that that sucker makes landfall somewhere north of the Rio Grande.

Folks, Katrina id dnot hit or pass over New Orlens. The eye of the storm hit the Bay St. Louis area in Mississippi.

Few people in our nation nderstand that Mississippi was devesrated by Katrina-the entire coastline two miles in, was destroyed and people theree are still living in FEMA trailers.

New Orleans troubles were created by government corruption and neglect. If the levies had been maintained with the more than adequate funding that had been allocated for them over the years, nothing would have happened in New Orleans.

Here in Texas, we're recovering from the 7 inches of rain Erin dropped on us, so another visit isn't exactly wanted right now. I hope Jamaica is prepared, they've got a monster on their doorstep.

Great comment Alan K. Henderson ^_^

My brother is actually hoping that it hits Port A so it'll finally tear down the old building his restaurant's in so he can rebuild.

He rents so he doesn't have to worry about a house & he can go spend a coupla days in SA with Mom & Dad.

You better check your maps. The blue track does not go over New Orleans.

I lived on the gulf coast for the first 27 years of my life. Ran from at least two hurricanes; coincidentally the last one was named Allen. Missed Corpus Christi by 150 miles, but we got gale force winds and some flooding. Will never forget seeing video of people jogging on the seawall in the middle of that. Nobody was jogging on the shore at Port Isabel...

Larry J, your reasoned comments are sorely missed at the Habitable Zone SPACE board. I've long ago abandoned the other HZ pages.

I gave up on HZ when it started looking like any other leftwing kook site. For me, it became uninhabitable. I spend most of my time here and here.

http://tinyurl.com/yrz2dp

Shows the track predictions for Katrina in the days leading up to its landfall.

It was going to Atlanta and wound up in Louisiana.

What a bunch of marroons.

Karl Hallowell:

You need to look up Roger Pielke Sr. Recently on his site he discussed that very issue and came to the conclusion that it is much more difficult to predict climate decades ahead than to do weather forecasting a few days to a week ahead. The core of the argument is that longer term, lots more variables change while short term many can be assumed to be nearly constant.

I was watching a documentary that was discussing climate prediction using chaos theory. It seems that large scale atmospheric phenomena can be predicated by a series of very small, seemingly unrelated events. Literally, the flapping of a butterflies wings in the Yucatan could be unknowingly linked to a tornado in Texas.

If I remember correctly, Ivan was on a similar track as Dean. That storm turned NNE in a day and rode the Cuba/Yucatan gap and into Louisiana. I think the models on Dean are right though, into Mexico, but I keep remembring watching Ivan's track since it formed.

The latest GFDL suggests that Dean might bring some much needed rain to California.