This month's seasonal forecast update has been a fascinating process. Speculating on why the dynamical models insist on a +NAO winter, we noted a close resemblance to winters following very low Arctic sea ice in September.
Interestingly the past summer's North Atlantic MSLP pattern was also a very close match to previous low-ice years (perhaps inevitably, i.e. the circulation anomaly was a cause of increased melting).
http://twitter.com/WorldClimateSvc/status/1185233653267341312