ATL: ISAAC - Models

[Portastorm]

Point of Clarification

I think it is important to be clear about something ... the image which shows a possible relocation of the center of Isaac is not any sort of official image or map or document from the NHC or a government agency. It was created by a professional meteorologist who is a member of another online weather forum.

[SunnyThoughts]

It's beginning to slow down and the center is starting to relocate to the north. Jmo though

Even if that is the case, the model doesn't know that, the 0z models are probably about completed..we just don't see them until later. Anything that happens now probably wouldn't show up in the models until tomorrow. If what others have said in here is factual.

[JTD]

Something people do need to remember is with landfalling systems, the models are typically too far left and a good amount of time the system will landfall say 50-60 miles east of where the models are expecting say 24hrs out.

Okay, so you just kind of confirmed what Senobia and I were thinking about the models.

Yes indeed, Stephanie. Just by way of historical footnote, Ivan, Charley, Andrew and Katrina to a small but significant extent, Opal, Rita and Ike are all examples of these eastward shifts near the end. Katrina could have went directly over NOLA and Andrew directly over downtown Miami but did not sparing some damage. Ivan missed Mobile Bay to the east. Charley was way east at Punta Gorda. Opal hit NW FL. And Rita hit extreme SW LA/SE TX after being progged intiially to possibly hit slightly south of Houston! Ike switched from southern TX initially I believe to the mid/upper TX coast Galveston landfall.

We may see something similar here with Isaac but I suspect nothing dramatic.

[Mandeville]

if the center relocates to the north, it means the official track will stay closer to the same as it is now since everything is trending west

[Hurrilurker]

A few questions about these model runs:

1. How long do they typically take to compute from start to finish once the data is fed in? Is processing power a major factor?
2. How long does it take to organize the data that gets fed into the model, including recon data from planes in the air? Does the model run have to wait for the plane to land and download the data? If so, could this data be transmitted back in real time to speed up the process?
3. Right now it seems there are 2 model runs per day, right (00Z and 12Z)? Any chance that in the future we'll see more than 2 model runs per day? Maybe even once per hour or continuously updated.

Seems everyone's always waiting on updated models and I assume faster updates could help with predictions and, if necessary, warning/evacuation decisions.

[WxGuy1]

It's important to remember that we're still looking at 60-84 hr forecasts from these models, and there are errors associated with those forecasts (obviously, since the forecasts have been varying and widespread). As it often the case, the future path of Isaac is going to be affected by its intensity, so how well a model handles the intensity of the storm will affect how well it handles the track/path. Of course, we also need to look at how well any particular model is handling the synoptic-scale environment -- errors in the forecast of 500 mb heights, for example, will affect the steering flow.

The fact that the models have shifted all over the place, and that one reliable model is so different from another, should tell us that this is a low-confidence forecast. As such, I don't think it's worth the time and energy to examine the exact model forecasts very closely. Just like it's often not worthwhile (dare say useless) to overanalyze each wiggle and wobble in the center of circulation, it's also probably not worth getting worked up about whether Model A shows landfall at Location A or 100 miles from Location A. We don't know which forecast will verify, so the best forecast method likely is to look at some sort of model consensus (where are the GFDL, HWRF, ECMWF, GFS, etc., from the past few model cycles showing landfall?). In addition, as we've seen with Isaac, model forecasts tend to "wiggle and wobble", and history shows that it tends to be better to adjust one's forecast only slowly and incrementally unless there is overwhelming evidence to suggest a radical change in forecast. A single model, or a single model run, or even 2-3 models and model runs, typically shouldn't be used to make a whole-sale change to your forecast. Imagine if the NHC flip-flopped their forecast as often as the models do (both run-to-run and model-to-model) -- eek!

There's a lot of money involved in the response of municipalities and states that will be affected by the NHC forecast and, ultimately, by Isaac. Ideally, it'd be great to evacuate a huge area of the Gulf coast right now, but that's not feasible. The more "busts" (i.e., false alarms), the more complacency can creep into people's decision making about future forecasts and events. Finding the balance between "crying wolf" (often resulting in fewer people leaving next time, since leaving town can be very costly for some people) and acting with sufficient time to enact one's emergency plans can be very difficult.

[Wx_Warrior]

Early 0z guidance



Uploaded with ImageShack.us

[PTPatrick]

Mos of that is still 18z data.

[ROCK]

yeah saw that earlier WX.....creeping west ever so slowly....is it done creeping west? IDK..need more model runs...

[Wx_Warrior]

Sorry didn't mean to re-post. Working at the desk tonight at the newspaper. Gonna be a long night or two, or three for me.

[N2FSU]

FWIW... 00Z NAM is running

[bonjourno]

Hello all. I'm a long-time lurker from around Gainesville, FL who just decided to register. Eagerly awaiting the latest on this perplexing storm.

[N2FSU]

00Z NAM: 24hr

[southerngale]

Timestamps of the maps are in the lower left corner

[N2FSU]

00Z NAM; 36hr

[N2FSU]

00Z NAM: 48hr

[smw1981]

It's beginning to slow down and the center is starting to relocate to the north. Jmo though

Even if that is the case, the model doesn't know that, the 0z models are probably about completed..we just don't see them until later. Anything that happens now probably wouldn't show up in the models until tomorrow. If what others have said in here is factual.

Hey Sunny, I thought the models started running at specific times automatically, so I don't think they have been completed yet. (I am not sure about this, so someone please correct me if I am wrong!) One of the METS explained the models last night, saying that the data goes into them automatically (no one actually sits down and enters it), and they initialize.. Sooo, the 00z models may have the current data in them..

[HurrMark]

Looks about the same as 12Z...maybe a tad NE.

[Senobia]

Timestamps of the maps are in the lower left corner

http://i.imgur.com/eyg7i.gif


http://i.imgur.com/qEzF8.gif

Okay, I see a split in the models in a 5hr span with the left ones being more clustered than the right ones. To my untrained eye, that does not bode well for SETX/SWLA - but what do you knowing types make of it? Anything?

[PTPatrick]

Yeah nams gonna be Mobile county...still showing a weakness over the Ohio valley at 48 hrs