No one here today - so rather than lurk as I have done for the past 5 years I will post.
Given up with GFS this winter. I ignore all its output post 96hrs.
I used to think it a good model for trend spotting. However it only spots trends as it is run so often it shows every variable possible. Which isn't really trend spotting more a model akin to monkeys, typewritters and the complete work of shakespeare.
Thus I think its output not just tonight but for ever should be ignored - I recall Ian F has commented that it's what the professionals do at the meto with regard the gfs.