GFS test

[hurricanetrack]

Ok- once again the GFS shows a Cape Verde type system forming and heading across the Atlantic- this time affecting the Greater Antilles and then hitting the road- out to sea. Putting aside the usual jokes about GFS bull-hockey, let's see what happens this time. There is energy over Africa and water temps will support development NOW. So, I will not post a new GFS update every six hours until either a landfall occurs or nothing happens at all- but I will be very interested in seeing if this idea of development out of the deep tropics this early has any merit. If so, we have plenty of time to watch and worry- if not, we know that the model is seeing something that is simply not possible yet. I'll post an objective critique of what happens next in one week- by then we'll know.

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

[Derek Ortt]

GFS seems to be having problems spinning up phantom TC's due to its increase to 27km resolution and the convective feedback issues. Some changes need to be made with the model regarding how latent heat release is handled or els this problem will continue

[MWatkins]

I am not sure that the GFS isn't just spinning up a phantom out of an ITCZ thunderstorm complex...it just kind of pops up a tropical low between the v+72 and v+78 hour timeframes then holds it for several days thereafter. Seems a bit too fast to spin up...

That would be a late June cape verde storm that would make it mostly across the Atlantic...which is of course unheard of.

However...this does...perhaps signal that the Cape Verde season...given all of the warm water out east...could get a quicker than normal start once the thermodynamic environment improves (in terms of tropical development) as we get into July.

My guess, just a guess, is that things get started late July early Aug out there.

And...of course...the desire for a quiet couple of weeks may be clouding my thinking as well...

MW

[dhweather]

In all, the GFS has been doing some odd stuff so far this season.

[Steve]

The European does the same thing but doesn't try to recurve the system the way the GFS is doing down the line.

Steve

[MWatkins]

The European does the same thing but doesn't try to recurve the system the way the GFS is doing down the line.

Steve

Just looked at the 0Z Euro...not much of a reflection if any up at 850MB. The GFS recurves the system in the model...probably...because it overdevelops it.

Also looks like the new 12Z GFS is backing down with no real development through 7 days.

M

[Anonymous]

Yea Mike, I don't know how the system will gets it's act together before 45 W.

[cycloneye]

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

12z GFS loop in 384 hours.

It shows weak low pressures moving from east to west in the Atlantic.But no development in this run only that the low when it gets north of the Greater Antilles deepens a little.

[cycloneye]

18z GFS.

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

18z GFS loop.It is more agressive as time goes by and by day 11 it has a strong low near Puerto Rico and near hurricane strengh storm between Puerto Rico and Bermuda.But what I would like to see is a consensus of the models.

[Derek Ortt]

a note about the 6 and 18Z forecasts

the initial conditions are merely extrapolations of the previous 6 hour forecast. Basically, the analysis is the 6 hour forecast from before. They really shouldn't be used as sole forecast indicators, only the 0 and 12Z when the observations are included in the analysis

[hurricanetrack]

Thanks Derek- I did not know that. So the 0z and 12z models are the really thorough ones, right? And the 6z and 18z are more intermediate in nature?

[cycloneye]

a note about the 6 and 18Z forecasts

the initial conditions are merely extrapolations of the previous 6 hour forecast. Basically, the analysis is the 6 hour forecast from before. They really shouldn't be used as sole forecast indicators, only the 0 and 12Z when the observations are included in the analysis

Good to know that thanks Derek.

[MGC]

Good info Dereck, all the less reasons to trust models.....MGC

[Scorpion]

Looks like a threat to Florida in the near future. Hopefully this doesnt pan out, im not quite ready for a hurricane.

[SouthernWx]

I never put too much stock in the 06 and 18z GFS runs (or NAM), as compared to the 00 and 12z runs, regardless of the season and type of storm progged (hurricane/ snow-ice storm/ tornado outbreak).

PW

[MWatkins]

To expand just a bit...

Basically, the 0z and 12z models get all the new observations, and the 6z and 18z models are updates with the features moved along 6 hours then initialized. Basically the 6/18z runs are reruns with interoplated data.

Sometimes, but not often, some new obvs will get into the models at 18z. Every now and then the SPC will request special soundings for anticipated severe events, or the NOAA dropwindsondes will work their way in, etc. But by and large, there are really only 2 runs per day with a couple of updates at 2am and 2pm respectivelly.

This is why, generally, any big changes in forecast thinking occur in the 5pm and 5am advisory packages and not many (usually anyway) big model-dependant changes occur in the 11am/pm discussions.

MW

[cycloneye]

To expand just a bit...

Basically, the 0z and 12z models get all the new observations, and the 6z and 18z models are updates with the features moved along 6 hours then initialized. Basically the 6/18z runs are reruns with interoplated data.

Sometimes, but not often, some new obvs will get into the models at 18z. Every now and then the SPC will request special soundings for anticipated severe events, or the NOAA dropwindsondes will work their way in, etc. But by and large, there are really only 2 runs per day with a couple of updates at 2am and 2pm respectivelly.

This is why, generally, any big changes in forecast thinking occur in the 5pm and 5am advisory packages and not many (usually anyway) big model-dependant changes occur in the 11am/pm discussions.

MW

Thanks Mike for posting that and I am sure that many(Including me) didn't know all about why the 00z and 12z runs are the most important.This is like a teaching lesson that you gave us Mike and the members will know now how the models operate.

[Swimdude]

Hate to agree with this notion... But the GFS has been *predicting* all sorts of systems since the beginning of June, 90% of which never occured anyway. Of course it's good to be on the lookout, but the GFS seems to be taking this a bit far this season.

[Derek Ortt]

Thats why I only put any stock in the 0 and 12Z runs, not the 6 and 18 as they have about as much value as the German currency did in the late 1920's and early 1930's... none

[jimvb]

The GFS created phantom storms last year. I go to the Southeast Unitarian Universalist Summer Institute (SUUSI - http://www.suusi.org) which occurs in late July every year. Last year's SUUSI (July 18-24) was threatened by a phantom GFS hurricane that struck South Carolina on July 18, causing attendees from that region to rush home instead, then hit Virginia Tech (location of SUUSI) on July 20 with Isabel/Richmond or Fran/Raleigh intensity, devastating the Institute. A few runs later, the "hurricane" missed the coast, and on the next run, it completely disappeared, never to reappear, and the weather at SUUSI was better than any other weather that summer.

I see that GFS is doing it again. SUUSI (July 17-23) is being threatened by a hurricane between Puerto Rico and Bermuda on the 13th. Further, the 0 and 12Z runs show the hurricane threatening the coast (at least 12Z does) and the nonsensical 6Z and 18Z runs show it turning away. I will believe this when I see it, 12Z or no. It could actually strike, or it could, like last year's SUUSI hurricane, simply disappear in model dust.