My bad  I just discovered data set I looked at only went up to 2002.

Along with 2009 and 2012. Mean monthly AO values for October were also signicantly negative in 2002, 2006 & 2014, and again, were followed by mostly mild and wet winters overall (with the odd cold, snowy spell as would be expected)  2009/10 would seem to be the only exception in this group.

Presuming they're the years the October occurred in, I'd be happy to take a repeat of either of those following winters (especially the 2009/10 winter, which I think is the best of my adult life)

I know I'm in a minority here, but I really wish the creators of the OPI had continued with it last year, and this. Certainly an almost neutral AO index for Oct. 2015 would have precluded any thought of a cold winter in 2015-16, which was duly the case. The chaps behind it got a lot of stick on account of winter 2014-15 falling hilariously short of expectations the first time they went public with it, but it was still an interesting idea. There were at least enough previously fair matches with the following winter to suggest it had as much credibility as a forecasting tool as plenty of other factors, such as May SSTs, ENSO or sunspots, which are still treated with respect.

Some more recent research suggests that the OPI, in only calculating circ anomalies at the 500mb level, failed to take into consideration the pattern needed to be throughout the troposphere and into lower stratosphere.

http://onlinelibrary.wiley.com/doi/10.1002/joc.3968/abstract

If you look at October 2014 when the OPI was being measured - while 500mb looked ideal - the 100mb pattern was completely different. Included October 2009 100mb for comparison.

Just some food for thought while we are musing about October patterns.

More a case of what that was, Winter's Tale; October Pattern Index. Everyone got quite excited up to about 2 years ago about this relatively new indicator developed by Italian academics, relating the degree of snow cover over Russia in October to winter weather over our parts. There seemed for a moment to be some fair statistical evidence, mostly focused on the strong correlation in 2009 (I think), but it failed a couple of years back and that was on the back of indecisive correlations since, so it seems to have gone.

You might already know this..can't tell if there's irony in your comment!

Many thanks for that. Glad to see that the ideas have been the subject of serious research for a number of years. It's sobering though that your link (to a 2014 publication) describes a variation of the index which still delivers only a 0.58 correlation, while a further link there from a paper this year boosts that correlation to just 0.61. 

Would still be at least a more reliable seasonal forecasting tool than just about anything else though!

Agreed

December 81 was superb , I was stuck in Banbury for three days until my Dad could get out of the village of Hempton.

As you say only matched by 1987 and 2010

As far as I know they never published the details about how they got to an OPI index for a particular year but they did claim an amazing correlation between the OPI value and the average AO for the following DJF of over 0.9

Sadly on the years they used it for forecasting rather than hindcasting it was a total flop and has now faded into the mists of time

The "wayback machine" has their forecast for 2014-15 in Europe and USA and some clues as to how it worked. Sadly this was one of the flops

Basically they were looking at the degree of ellipticity of the polar vortex at 500mb every day in October (that is the theta and phi) and also the 500mb anomaly every day at various locations in the mod-latitides. In total this gives the "broadness" of the vortex and its displacement from a pure circular shape

Someone with sufficient skill and spare time could probably continue to calculate it

http://web.archive.org/web/20141219141504/http://app.til.it/opi/

Brian was mentioning a possible link between warm Septembers and mild winters on twitter the other day, but maybe it is more a sign of general warming rather than a definite connection having looked at the figures. Last month was the 6th plus 15C September since 1999 compared to none in the years from 1971 to 1998

Well I suggest running the archive charts, but with around two thirds of these winters in the mild or very mild category, the common factor is to see December zonality roll on thru January and into February.

ukmo has the high holding up but ecm & Gfs a classic sinker. 

But there were another 6 in that period of 14.5 or more. In the post 1998 era (18 years) there have been 7 below 14C. Depends how you pick.

Here are some more graphs, derived from data.

and

I don't know where he got that idea from because look at the evidence.

Just these Novembers alone

2009: exceptional mild and wet

1995: Mild and not that wet

1990: Just above Average and drier

1985: Cold but not wet

1984: mild and wet

1978: Very mild and drier than average

1962: cool but not wet

1946: very mild and wet

1939: very mild and wet

1928: milder and wetter than average.

Got some clarification from Judah Cohen about the OPI. He told me:

I met with the Italian scientists but I don't think that they have pursued it further.

Even the mild club would like that , I'm sure

Thanks for that. From now on I won't have to waste half an hour each autumn trying to discover whether it's still out there being calculated on some Italian classified advertising site.

Pity really. Fundamentally I agree with KevBrads1 that there's no real logical reason why average atmospheric conditions in one particular month, whether May, October or November, should determine those in Dec, Jan or Feb. But to be honest there's not much else out there with any more credibility for seasonal forecasting, and at least the OPI was trying something a little bit different, with emphasis on the shape as well as extent of pressure patterns.

I was joking, you plum.