ATL: IRMA - Models

[DESTRUCTION5]

New member here, my 1st post. I've been a guest and frequent visitor since 2004. I live in Martin County, FL. I just wanted to thank everyone that post the model runs, images, and explanations of the complex science I try to understand. I really can't explain how much I appreciate this website and it's members. So thank you for your efforts.

Welcome fellow County resident. Stuart here.

[tolakram]

Here's the Matthew plots from last year, about the same number of days out. This might be instructive as for the amount of error to be expected. Check out the UKM track. Nailed it.

[emeraldislenc]

If the models shift east would that mean more of a threat to NC?

[Hurricane Andrew]

So did a little analysis on model pressures; no matter what the initialization pressure is, pretty much all of the models show an increase over the D1-3 period, followed be fairly significant deepening over D4-D5 up to landfall. Quite concerning...

Yep I've noticed this as well. Which has me confused, because the NHC predicts fairly significant weakening before landfall (30-40mph less than now). So why would the models show a deepening storm but NHC expect it to weaken?

It's possible that the weakening happens earlier, but when Irma passes over the astronomical OHC/SST values off of SFL, it restrengthens.

[JPmia]

Hey all, been busy getting ready down here so I haven't had the time to review every model like I usually do.. what's the summary this morning on the models?

[DESTRUCTION5]

Here's the Matthew plots from last year, about the same number of days out. This might be instructive as for the amount of error to be expected. Check out the UKM track. Nailed it.

UKie absolutely nailed Matt to the wall here.

[tolakram]

0Z UKmet

[tgenius]

Hey all, been busy getting ready down here so I haven't had the time to review every model like I usually do.. what's the summary this morning on the models?

Appears alot of the 00Z guidance shifted east (but Euro Ensembles still are concentrated over SFL) and I believe the 06Z HWRF shifted a degree west, and the 06z GFS also moved a little bit west. UKMET also appears to parallel the coast.

[Blown Away]

06z GFS was slightly W near FL and big jump W up near GA/SC border... Many of those guidance lines based on GFS, so 12z may shift slightly back W...

[WilmingtonSandbar]

So did a little analysis on model pressures; no matter what the initialization pressure is, pretty much all of the models show an increase over the D1-3 period, followed be fairly significant deepening over D4-D5 up to landfall. Quite concerning...

Yep I've noticed this as well. Which has me confused, because the NHC predicts fairly significant weakening before landfall (30-40mph less than now). So why would the models show a deepening storm but NHC expect it to weaken?

Might have something to do with interaction with the Gulf Stream

[Vdogg]

I'm not buying the east shifts to much and sending this out to sea. If that happens no one will ever believe the NHC again and the panic they created for no reason would be just awful. Huge inconvenience right now for people who rely on getting gas for a living. Just crazy why people are panicking without even a hurricane watch or warning. Just an opinion not a forecast.

The NHC didn't create any panic, none. They simply stated the facts as they presented themselves, and always cautioned that the track could change. People not understanding the track and models and the media themselves are responsible for the panic.

[chris_fit]

From the 5AM NHC Update on 9/5 Irma had a the following forecast.
INIT 05/0900Z 16.6N 57.0W
24H 06/0600Z 17.7N 61.5W


Here are the actual results:

06/0600Z 17.7N 61.8W

So just in 24 hours, Irma's Lat was correct, but Log was off by about 20 miles. Irma is moving faster than forecast. 20 miles a day for 5 days is 100 miles. Accuracy goes down the more days you go out... You can connect the dots together.... we have a long ways to go.

For now, I would venture out to say the models that have this moving faster might have a better handle on the storm.

[Vdogg]

06Z GFS Ensembles.

Big shift east on the ensembles with many making a classic Northeast turn.

[Mouton]

From the 5AM NHC Update on 9/5 Irma had a the following forecast.
INIT 05/0900Z 16.6N 57.0W
24H 06/0600Z 17.7N 61.5W


Here are the actual results:

06/0600Z 17.7N 61.8W

So just in 24 hours, Irma's Lat was correct, but Log was off by about 20 miles. Irma is moving faster than forecast. 20 miles a day for 5 days is 100 miles. Accuracy goes down the more days you go out... You can connect the dots together.... we have a long ways to go.

For now, I would venture out to say the models that have this moving faster might have a better handle on the storm.

I noticed the difference in speed today also. I think you might be onto something here with your supposition.

[xironman]

06Z GFS Ensembles.

Big shift east on the ensembles with many making a classic Northeast turn.

By day 6-7 the low off of Nova Scotia starts to move east allowing the ridge over the east coast to go away, at that point a NE track can happen. The 4 runs that are truly OTS are way north right away, which is hard to imagine.

[chris_fit]

From the 5AM NHC Update on 9/5 Irma had a the following forecast.
INIT 05/0900Z 16.6N 57.0W
24H 06/0600Z 17.7N 61.5W


Here are the actual results:

06/0600Z 17.7N 61.8W

So just in 24 hours, Irma's Lat was correct, but Log was off by about 20 miles. Irma is moving faster than forecast. 20 miles a day for 5 days is 100 miles. Accuracy goes down the more days you go out... You can connect the dots together.... we have a long ways to go.

For now, I would venture out to say the models that have this moving faster might have a better handle on the storm.

Just to addon -

If we go back to the 48hr forcast point from 9/4 (N 17.8 W 61.2) the error is 40 miles.

If we go back to the 72hr forecast point from 9/7 (N 17.8 W 60.6) the error is 80 miles.

.... And this is the 'easier' part of the forecast with much more model agreement.

[lovingseason2013]

From the 5AM NHC Update on 9/5 Irma had a the following forecast.
INIT 05/0900Z 16.6N 57.0W
24H 06/0600Z 17.7N 61.5W


Here are the actual results:

06/0600Z 17.7N 61.8W

So just in 24 hours, Irma's Lat was correct, but Log was off by about 20 miles. Irma is moving faster than forecast. 20 miles a day for 5 days is 100 miles. Accuracy goes down the more days you go out... You can connect the dots together.... we have a long ways to go.

For now, I would venture out to say the models that have this moving faster might have a better handle on the storm.

I noticed the difference in speed today also. I think you might be onto something here with your supposition.

So help us out a bit, which models do you see having it going faster? The more west leaning ones?

[SouthFLTropics]

Well after looking at the model runs this morning I don't know if I should feel better or worse up here in Port St. Lucie. The runs moving East are somewhat encouraging because maybe it is a trend to a further East shift. However, the runs further NE also bring into play the possibility of getting a landfall right along the coast. Not sure what to make of all of it this morning.

Especially concerning is the spread on the GFS ensembles. So much for a consensus. I feel like we're back to square one.

[tarheelprogrammer]

Any chance this goes OTS? Seems models are trending east in the ensemble members. What did the ECMWF ensembles look like?

[chris_fit]

So help us out a bit, which models do you see having it going faster? The more west leaning ones?

That, I am no where nearly qualified to answer - I'm more of a 'analyze after the fact' and see how we did, haha

I have not checked which ones are faster/slower. Not really sure how to do that based on historical runs without going through one manually. Maybe someone on this board has a better way?