Felly sgrifennodd John Hall :
Apparently the MO predict the coming winter's NAO based on May's
Atlantic SST anomalies. I suppose that those May anomalies are going to
be greatly influenced by what the previous winter's circulation pattern
was like.

I wonder why May is used. You would think that taking a later month
would give a better prediction. Surely it can't require several months
to process the data?

As I think I posted at the time this came up before, I don't understand
that either. You'd think that if May's was found the best, surely some
statistical input from other months would at least help the model somewhat.
So why not form a model on a number of months' data? A decent analysis
technique would chuck out data that wasn't important anyway, so there's
not a lot to lose (principle of parsimony excepted).

Adrian
--
Adrian Shaw ais@
Adran Cyfrifiadureg, Prifysgol Cymru, aber.
Aberystwyth, Ceredigion, Cymru ac.
http://users.aber.ac.uk/ais uk