OlaL

-1,5 or lower will do! I would certainly take one of these winters, which are the ones with OPI<-1,5 1976/77 1978/79 1984/85 1985/86 2009/10 2012/13 Out Of these 18 months, only 1 above normal for my area (Dec 1984). -1,5 seems to be a break point where you go from unclear signal(big spread in the plot), to clear signal (colder than normal).

Interitus Yes, you beat me with 5 mins there. Thanks, elipticity makes sense. Looking forward to reading the paper when it is released. In particular I want to see whether there is a reasonable link between the geophysics and the statistics. The correlation is obviously strong, but it should ideally be backed with a geophysical theory.

Dont know if you have noticed, but as far as I can see based on the site, http://app.til.it/opi/, the index is the sum of following two components: Elyptization[sic*]+axes parameters Now, I dont know what they mean by eliptization (elyptization?), but what I can see that this parameter is not changed much from the 06z run. The big change is in the axis parameters.

I did the same plot for the three winter months, individually, and interestingly, January got the best fit, with R^2 of .47, followed by February at .43 and December "only" at .27. Just speculation of course from my side, but is the reason to this "lag" that the pattern established in October is affecting the Stratosphere, which in turn is feeding back to the Troposphere and this takes some time? 1984/1985 is maybe a good example here, where we had a -1,8 OPI, followed by a very mild december, before there was a big freeze in January and February (as predicted by OPI)

Thank you for a very interesting topic and a lot of good posts. Have been following the forum for some years without posting. I see this index can show for amazingly high correlation with AO, although last year did not fit in this picture. However, the winter was extremely mild in my part of Europe(Southeast Norway). So I dediced to try to plot OPI index (thanks to people on this forum for posting the index numbers for each year) against winter mean temperatures. It actually shows a very good correlation. My interpretation of the data is that the index is predicting colder than average winters for Southeast Norway when you have an index value below -1 and milder than average winter with index values above 0. There is a somewhat grey area in between these figures where it probably has less of a predicting ability. Still I am amazed by the skill and predicting ability by this index and is looking forward to the coming winter with a lot of anticipation if we can keep the index value below -1 and preferrably lower until the end of october. Below is the plot of OPI(x-axis) against southeast norway DJF anomalies(y-axis):