I remain of the belief that it is not possible to reliably forecast winter weather.However I do think that there is chance of of forecasting broad weather type as indicated by the NAO type.The issue is therefore the advance forecasting of the winter NAO
The met office do there own NAO forecast and sometimes the values are mentioned in their seasonal contingency forecasts.I would be very interested to know how these validate.It should be possible to get a time series showing the forecast NAO value and the actual.At the minimum figures showing how many NAO + forecasts were right and how many NAO- were right could be produced.The actual CET for each winter could also be added as he NAO is not a perfect correlation with + or - CET anomalies.(correlation is around 0.66)
Any suggestions about the best way of getting the winter NAO forecasts from METO?
Exactly a year ago the Met Office revealed that they could skilfully predict the winter NAO up to 2 seasons ahead, This got faithfully reported as such in the press. The paper in question was paywalled in Nature Geoscience, but it did appear for a time on a Russian server, though the link is no longer working.
Basically there was an r-value of 0.62 for the coming season, which fell to just 0.42 for the following one. This does NOT mean 62% accuracy, as the press reports claimed (and which the Met Office didn't attempt to correct, as far as I know). It's just an expression of the scatter of actual past values from the ones the model would have thrown up.
Thus an r-value of 0.62 means that their model can explain 38% of past variations in the NAO, which is fairly impressive as these things go (although the other way of looking at it - that 62% of what goes on with the NAO is still a complete mystery 3 months out - sounds a bit less impressive).
An r-value of 0.42 for the winter after is a lot less "skilful", as they say in these circles. That figure implies that just 18% of past variation is explained by their model, which is pretty useless on the face of it.
As to your question of how to get hold of MetO winter NAO forecasts, I have no idea, but the following figure, courtesy of the abstract to the above paper, shows how the model they use fared in terms of hindcasting, for what that's worth:
(The figure in the top left is for the year ahead, the figure in the top right is for 2 years ahead. The red line is the ensemble mean, the black line actual values.)
http://www.nature.com/ngeo/journal/v9/n11/abs/ngeo2824.html
2 miles west of Taunton, 32 m asl, where "milder air moving in from the west" becomes SNOWMAGEDDON.
Well, two or three times a decade it does, anyway.