GFDL has been relatively successful on two storms this season (there may be others but these two come to mind first) It was forecasting Edouard to back track SW, cross FL and then impact the northern GC several days before this actually came to fruition... it was WAY off on intensity, but it's forecast path was right on...

Regarding Hanna, GFDL was the one model that insisted on a more westernly component, taking the storm near the mouth of the MS river and east of NO... although not perfect, it's performance was better than most others 48 hours out..

I am not a believer in models per se, but when they start showing some accuracy, like the GFDL has recently, it certainly gives them a little more respectability than what I have given them in the past...

What would be really great if the models were rated on their performance... maybe someone is doing this, I really have not seen the data... however, it would be great to be able to review the data to see which models have performed the best over a given forcast range, and period of time... like how the stocks perform... of course this wouldn't tell you if the model would be accurate on its next storm, but at least you'd have a track record for it performance...

my opinion/guess on TD10 is that it will probably survive the brush with SA, the center will not relocate, and the system will continue to head for the central caribbean... whether this is a CA storm, a GOM storm or a FL storm remains to be seen.