Google DeepMind has really, really nailed Melissa. It's one of the most difficult forecasts of recent times (if not ever), and it was pretty much the only one that had been consistently right on track and intensity.
NHC's discussions had repeatedly highlighted the GDM ensembles, at times even following their ensemble intensity to the point of issuing forecasts higher than all physics-based models (and yes, that included HAFS-A/B and HWRF at some point). That ended up verifying.
















