Which model(s) you prefer the most?

[cycloneye]

I posted this thread for all of the members to say what model or models are the best around to see a good guidance when threats arise in the tropics.

Me personnally dont have one model as a favorite however UKMET and NOGAPS I like the most but I know that others will say distint models.

[OtherHD]

GFDL is by far one of the best regarding cyclone tracks. Intensity?...Well...uh...it sucks.

[wxman57]

The GFDL was the only model that correctly predicted (from 24 hrs out) the demise of Lili about 12hrs before landfall. But I certainly wouldn't say that the GFDL did the best job with hurricane tracks for at least the past few years. The GFS (formerly MRF) has done far better than the GFDL (for the most part) since 1999. The GFDL is prone to wild swings in direction, like with Michelle a few years back, forecasting Florida panhandle one run, Miami the next, then Tampa. Hopefully, the 2003 modifications will make it more reliable from farther out.

[Stormsfury]

In regards to tropical forecasting, I do not actually know the best track record for the models, but I do know the significat biases ... like the GFDL does have a tendency to bomb out storms, etc., etc.,

Anyway, in regards to a model I prefer ... there are several in different timeframes ... and depends on the weather pattern as well ... different models have better handlings on certain longwave patterns ...

In the medium range, the EURO by far is the one I consult first when the new runs come out, I do not use the EURO within 48 hours most of the time (though I still consult it)...

[Scott_inVA]

For entertainment value only, here are report cards on Izzy and Lili.

Criteria:
I have a minimum of 20 runs per model in database,
Margin of error ranking is at 72 hours (3 days prior to landfall). #1 has the lowest margin of error at 72 hrs.

IZZY

1. NOGAPS
2. UKMET
3. AVN
4. Official TPC
5. EC
6. GFDL


LILI

1. UKMET
2. Official TPC
3. LBAR
4. AVN
5. EC
6. NOGAPS

Obviously, this is inconclusive. 24 hour margins are different (GFDL was smokin' hot at 6 hours, puked at 48 and 72). I use 72 hours b/c 3 days out is critical for evacs and emerg planning.

What is critical to me is how TCP ranks; they consistently verify in the top 3 inside of 72 hours.

FWIW, I don't have favorites. I like trends. AVN was excellent last year; I like seeing the EC and the UKie as close as possible to the AVN (boosts confidence ). When the GFDL is "on" a storm, it is pretty good and hopefully will be better this year. The MM5 knocked the cover off the ball a couple times last year (notably, Izzy) and is worth a look especially near land, IMO.

Scott



Scott

[Stormsfury]

Scott, thanks for posting those stats and model verification scores anyway ... that helps a lot ...

[JetMaxx]

I really like the EURO (ECMWF)....but try to always utilize a range of model data; along with climatology in creating an forecast track analysis.

[wxman57]

I guess I really don't use the ECMWF much because it's outdated by the time it arrives (12Z run is released around 02Z the next day) and only a low-resolution version is released to the public. The Europeans can keep their model. We should make them pay for GFS and satellite data.

[Colin]

Heh... with Izzy the NOGAPS was first, and with Lili, NOGAPS was last... just goes to show you can't only rely on one model.