Major Hurricane Irma

[Griizzcoat]

Good Lord.

[WishingForWarmWeather]

Goodnight everyone. I am assuming Cat 5 upon waking up. Here's to another day of tracking and obsessing. 

[Morris]

150, landfall in SWFL

[seether]

First time poster, long time lurker but does anyone else believe the models are having a hard time digesting the new data introduced and may take another run to even out?

[hawkeye_wx]

A short-term benchmark with some importance is the northern islands.  Models suggest that if Irma is going to remain over water north of Cuba, the eye will have to track north of the northernmost islands, while any track over the islands will send Irma into Cuba.  So far, there is no sign of any wnw bend.  If the bend doesn't start by morning, the islands are in trouble.

[Hoosier]

Euro is basically a farther east version of GEM.  Overall upper air features more similar to that than the GFS.

[bch2014]

I just have a hard time believing that a storm is going to grind away in Central Cuba yet still make landfall at 940mb in SW FL.

[seether]

I believe these models are just as confused as most of us are haha.

[Griizzcoat]

[phimuskapsi]

5 minutes ago, bch2014 said:

I just have a hard time believing that a storm is going to grind away in Central Cuba yet still make landfall at 940mb in SW FL.

Central is the flattest part. 

[Griizzcoat]

[Mountain_Patch]

[seether]

This run.....Anyone have any insight?

[Mountain_Patch]

1 minute ago, seether said:

This run.....Anyone have any insight?

Wait until the 12z. Might be a further westward trend, or it could have overdone the impacts of the data from earlier. Regardless, it has designs on Florida.

[seether]

Alright agreed, this model is just screaming confusion, im assuming due to the new data introduced.  Either way I think FL is taking a major hit one way or another.

[seether]

Anyone look at the CMC?  Completely west of florida.

[ScottB]

Would wait for the 12z suites to look at trends. Still have quite a ways to go with Irma. 

[csnavywx]

The sat presentation and T numbers suggest Irma is nearing cat 5 status.

[whisp]

ive been lurking since like page 20 and made an account to ask a question...

has anybody noticed that the Japanese model seems to be showing the same trends as GFS & Euro but much earlier.

to me it looks like whatever way the JMA goes the other 2 go the same way the next day 

 

[Mountain_Patch]

1 minute ago, csnavywx said:

The sat presentation and T numbers suggest Irma is nearing cat 5 status.

next recon takes off in 40 minutes

[Griizzcoat]

IRMA looking pretty. About ready to dance with the islands. I hope the islands can dos-à-dos.

 

http://col.st/0b9aP

[LibertyBell]

1 hour ago, seether said:

I believe these models are just as confused as most of us are haha.

I saw someone saying earlier that models are better than human beings (or vice versa) and I don't think either is better than the other.  Models don't have some magical knowledge from some advanced alien species that we don't know about.  All they are is a combo of the data we feed into them plus our knowledge of atmospheric physics- both of which are far from perfect.  So if the models are wrong, that is just a reflection of our lack of complete knowledge.  They can by definition neither be better than us or worse than us, since they are an artificial construct.

 

[Griizzcoat]

6 hours ago, Wow said:

Trend loop of the hurricane models 

06Z--- Wagons West.

[TPAwx]

Please stop with the models are confused or wrong nonsense.

EPS mean stays north or Cuba, kicks out, and eventually landfalls east of Panama City.

 

[LibertyBell]

11 minutes ago, TPAwx said:

Please stop with the models are confused or wrong nonsense.

EPS mean stays north or Cuba, kicks out, and eventually landfalls east of Panama City.

 

Interesting trends we are seeing tonight for this to go into the Eastern Gulf/west side of Fla.

[skierinvermont]

33 minutes ago, LibertyBell said:

I saw someone saying earlier that models are better than human beings (or vice versa) and I don't think either is better than the other.  Models don't have some magical knowledge from some advanced alien species that we don't know about.  All they are is a combo of the data we feed into them plus our knowledge of atmospheric physics- both of which are far from perfect.  So if the models are wrong, that is just a reflection of our lack of complete knowledge.  They can by definition neither be better than us or worse than us, since they are an artificial construct.

 

Like the other poster said, this models are confused or wrong or humans are better nonsense needs to stop. 

Of course models are better than humans. No human can look at a real-time 500mb map or any combination of atmospheric parameters and make a 5+ day prediction about the weather any where close to as accurately as a computer model. They had accurate data in the 1970s and 1980s but they didn't even issue 5 day forecasts for hurricanes until 2003. And the track error today is less half what it was in the 1990s. It's not because humans were stupid in the 1990s or a lack of information. It's because no human can come even come close to a computer model for predicting the weather. Humans weren't even as good as computers in the 1990s. And models today are far better than the 1990s. Any human prediction today is almost entirely based upon model output.

Which is why the NHC basically copies the model consensus for hurricane track. And insofar as the NHC track slightly differs from the model consensus track, the model consensus track usually wins.

[MillvilleWx]

Doesn't get much better than this folks. This storm will be one to remember for many in the future. 

[LibertyBell]

3 minutes ago, skierinvermont said:

Of course models are better than humans. No human can look at a real-time 500mb map or any combination of atmospheric parameters and make a 5+ day prediction about the weather any where close to as accurately as a computer model. They had accurate data in the 1970s and 1980s but they didn't even issue 5 day forecasts for hurricanes until 2003. And the track error today is less half what it was in the 1990s. It's not because humans were stupid in the 1990s or a lack of information. It's because of no human can come even come close to a computer model for predicting the weather. Humans weren't even as good as computers in the 1990s. Any human prediction today is almost entirely based upon model output.

Which is why the NHC basically copies the model consensus for hurricane track. And insofar as the NHC track slightly differs from the model consensus track, the model consensus track usually wins.

I mean machine logic is faster than human beings sure, but the knowledge that is fed into the machine comes directly from our knowledge of atmospheric physics and our data.  What I was trying to say is you can't blame the model for it being "wrong"- what it outputs is directly related to what we know of the atmosphere and the data we feed into it.

Completely agree about models being much better than they were in the 80s and 90s, and that's a direct result of us improving them through gaining better knowledge of atmospheric physics.

[skierinvermont]

7 minutes ago, LibertyBell said:

I mean machine logic is faster than human beings sure, but the knowledge that is fed into the machine comes directly from our knowledge of atmospheric physics and our data.  What I was trying to say is you can't blame the model for it being "wrong"- what it outputs is directly related to what we know of the atmosphere and the data we feed into it.

OK I do get your point, the issue was more with a few of the other comments others have been making about models.  Models aren't some basic formula that can be reduced down to some basic human logic. They're millions of variables all interacting simultaneously according to complex mathematics. It would take 1,000 mathematicians doing computations 24 hours a day for a century to produce a 24hr forecast as accurately as a modern computer. Then it would take another 9 centuries to produce the 10 day forecast. And that's probably a massive understatement of their complexity.

Also from what I understand the bigger limiting factor for models in this day and age is the quality of the initialization data. 

[LibertyBell]

8 minutes ago, skierinvermont said:

Models aren't some basic formula that can be reduced down to some basic human logic. They're millions of variables all interacting simultaneously according to complex mathematics. It would take 1,000 mathematicians doing computations 24 hours a day for a century to produce a 24hr forecast as accurately as a modern computer.

Sure, that's why they must be run on supercomputers, what you described is actually atmospheric physics.  As we gain more knowledge of it, we improve the models further (and also being able to ingest more data through satellites, dropsondes, etc, than we could do 20-30 years ago because we're further into the space age and have better technology than we had back then.)  You're absolutely right that people weren't "dumber" in the 80s and 90s, but we did have less knowledge and not the level of technology we have today that makes the models so much better than they were back then.

But (as I'm sure you know) perfect forecasting will never be achieved (until the day that we are controlling the weather ourselves, that is) because what can be done is limited by chaos theory and random interactions (which, in addition to the possibility that our data might be slightly off, and both have a stronger influence at longer lead times,) is why we run ensembles and why they are more accurate at those longer lead times than operational runs.