OCTOBER PATTERN INDEX (OPI) MONITORING WINTER SEASON 2014-2015

[SMU]

Steve - where are you getting those daily OPI figures from?

 

If it is from here:

 

Image-1.png

 

then I suspect that these are not the daily OPI figures for those particular days but rather the daily OPI forecasts given each day. So we don't actually know that rolling average! Frustrating!

 

 

It must be the rolling figure..... as again that's what the author said in the OP, also, a 12z forecast for the day is never going to be far wrong from what the actuals are....

 

S

[snowking]

This may have already been covered but my understanding of the tool mostly revolves around this quote:

 

 

For example on the 15th of October, the software calculates the OPI index based on the 15 

consolidated daily charts (1-15 October) adding 10 forecast charts as given by the GFS model (16-

25 October).

 

Therefore, each day the tool updates we should be getting a more and more realistic figure shown. The large fluctuations are still accounted for by the complicated nature of the index calculation itself, and how prone the index is to swings in NWP, particularly given some major hemispheric events currently, including large Typhoons and an impending MJO surge:

 

 

But my understanding is that the calculated figure for today of -3.15, is the most accurate assessment available to us. But as cautioned by previous posters it is probably wise to wait until the final third of the month before following every high and low, and given the current situation globally I would expect plenty more swings through the next 10 days owing to a likely increase in uncertainty within NWP as a whole.

 

SK

[chionomaniac]

It must be the rolling figure..... as again that's what the author said in the OP, also, a 12z forecast for the day is never going to be far wrong from what the actuals are....

 

S

If you look at last years' figures Steve, it becomes pretty clear that the daily figures are the forecast figures, not a figure for each individual day. That is why the fluctuations reduce towards the end of the month.

 

And last year the figure of 1.64 was pretty much figured by the 20th and didn't vary from then on in.

Edited October 12, 2014 by chionomaniac

[tcc]

Right then, see you all back here on the 20th! No point reading too much into the figures before then!

[Cloud 10]

Right then, see you all back here on the 20th! No point reading too much into the figures before then!

 

I think that ship has sailed.

[Yarmy]

The real problem is that not a single one of us knows how the figure is calculated. No paper has been published and the method described only vaguely. If there are no proprietary secrets or commercial interests here why not simply release the exact steps and let people verify independently. Until then, given the remarkable correlation, the suspicion is that we have 40 years of data curve-fitted and a 1 year AO forecast.

Edited October 12, 2014 by Yarmy

[bluearmy]

The real problem is that not a single one of us knows how the figure is calculated. No paper has been published and the method described only vaguely. If there are no proprietary secrets or commercial interests here why not simply release the exact steps and let people verify independently. Until then, given the remarkable correlation, the suspicion is that we have 40 years of data curve-fitted and a 1 year AO forecast.

Ooh you old cynic yarmy! This thought had flashed across my mind during the evening but I resisted the option to post it in the hope that Ricardo would appear at some point and explain the issues that have been raised.

Did I read that cohen's team had withdrawn from the joint work that was being undertaken ? Can anyone expand that or put me right?

[Blessed Weather]

[Penrith Snow]

Yes, why the secrecy and if no one knows how the index for any given day is compiled we have no way of double checking the data and indeed that very impressive 0.9 correlation graph.

We have no way of checking indeed whether the graph is correct!

I hope my sceptism is unfounded I really do, because this could be an excellent piece of work and I have always suspected that the October circulation pattern influences the following winter but come in Ricardo open up!

As I said a link between October circulation patterns and winter weather does probably exist and that's why we often see a warm, dry October prior to a cold winter but a claimed correlation of 90% in any long range weather forecast really would be remarkable.

I look forward to reading the peer reviewed paper in Weather Magazine.

Andy

[feb1991blizzard]

Totally lost now, I haven't a clue what that bar graph represents then!!!!!!!

[beng]

To be fair to Riccardo - he has said that they will be publishing a paper in due course - so in a way we're lucky to be able to see some of this in advance. I guess, it's no worse than in the middle of winter, following the ECM or GFS to see them predict blizzards at day 8, only for the following days runs to collapse the block and sweep in a warm South Westerly.  

 

The good news is that we can now start this emotional roller coaster a couple of months early 

Edited October 13, 2014 by beng

[Summer Sun]

Based on the 00z run the OPI is now at -1.54

[feb1991blizzard]

Yes, why the secrecy and if no one knows how the index for any given day is compiled we have no way of double checking the data and indeed that very impressive 0.9 correlation graph.

We have no way of checking indeed whether the graph is correct!

I hope my sceptism is unfounded I really do, because this could be an excellent piece of work and I have always suspected that the October circulation pattern influences the following winter but come in Ricardo open up!

As I said a link between October circulation patterns and winter weather does probably exist and that's why we often see a warm, dry October prior to a cold winter but a claimed correlation of 90% in any long range weather forecast really would be remarkable.

I look forward to reading the peer reviewed paper in Weather Magazine.

Andy

 

I remember in the school week holidays in late October 1989 it was actually hot but we had a mild winter, a year earlier we had record cold in that said holiday period followed by another rubbish winter.

[Tom Quintavalle]

To be fair to Riccardo - he has said that they will be publishing a paper in due course - so in a way we're lucky to be able to see some of this in advance. I guess, it's no worse than in the middle of winter, following the ECM or GFS to see them predict blizzards at day 8, only for the following days runs to collapse the block and sweep in a warm South Westerly.  

 

The good news is that we can now start this emotional roller coaster a couple of months early 

 

If I can just add to Beng's comments above. I hope Riccardo won't mind me discussing a few of his thoughts from July, that he expressed on the Italian forum.

 

Riccardo is meeting Professor Judah Cohen at Christmas to discuss their joint research into this subject and they intend to publish their findings after this time, but until then, Riccardo will not be able to disclose anything further. There is no secrecy involved, simply a private agreement between both parties until such time as they are ready to divulge further information.

 

It's also obvious from reading Riccardo's posts that he is at pains to stress that the OPI should be treated with caution during the first half of the month, there are just too many variables and not until we hit the last third of the month, i.e when the consolidated data outweigh the predictive charts, can we then begin to have some confidence in that the OPI value reached by the end of the month will be a realistic tool, in terms of trying to predict the AO index, for the forthcoming Winter. 

 

Regards,

Tom.

Edited October 13, 2014 by Kentish Blues Man

[bluearmy]

That's all very well Tom but will it snow in my back garden !!!

More evidence that the thread should be quiet until the back end of this week - the polar profile looks to be well split for the next week to ten days but the last week of October could see that come under pressure. All conjecture until we get to nxt weekend and see where the OPI has begun to settle.

[BLAST FROM THE PAST]

I remember in the school week holidays in late October 1989 it was actually hot but we had a mild winter, a year earlier we had record cold in that said holiday period followed by another rubbish winter.

What it's like in our neck of the woods in October  is immaterial I would say.  Just look at the AO, even when negative it doesn't necessarily fall right with us.

 

BFTP

[radiohead]

Daily value is now just -0.19 following the 06z.

But as already been mentioned, it's probably best to ignore these figures until around the 20th.

[BLAST FROM THE PAST]

Looking at 00z of -1.64 and 06z of -0.19.....agreed let it be for a while

 

BFTP

[nick sussex]

Isn't the problem here the forecast value and too much emphasis on this.It's obvious that theres too much volatility when you add in 10 day projections and if we just concentrate on the daily figure then it might be better.

 

Lets reverse this if were sitting on a current positive OPI reading and the forecasts showed a big drop I'd be more worried. We've already banked negative upto today.The future forecast may well be wrong and so its probably not worth worrying about unless it actually comes to fruition.

[LomondSnowstorm]

Isn't the problem here the forecast value and too much emphasis on this.It's obvious that theres too much volatility when you add in 10 day projections and if we just concentrate on the daily figure then it might be better.

 

Lets reverse this if were sitting on a current positive OPI reading and the forecasts showed a big drop I'd be more worried. We've already banked negative upto today.The future forecast may well be wrong and so its probably not worth worrying about unless it actually comes to fruition.

 

 

 

Also worth considering what we know about the daily GFS runs - the 6Z has time and time again been the most progressive run of the day with the jet, with generally the 12Z being less progressive than the 00Z and the pub run being all over the place, so in that sense it probably shouldn't be surprising that the 6Z values have been the highest for the last few days.

[bluearmy]

one would suspect that the 12z would give a better OPI number than the 06z or the 00z. the height rise to our north is more in line with current ecm thinking.

[radiohead]

one would suspect that the 12z would give a better OPI number than the 06z or the 00z. the height rise to our north is more in line with current ecm thinking.

The OPI daily figure has been showing quite dramatic changes over the past few days regardless of which run though. But I suppose when you are dealing with 10 days of GFS output then you are well into FI territory and divergent output in the latter period is only natural anyway.

It makes me wonder what value the previous daily figures in the graph have until we get to within about about a week of the end of the month.

As to intra-run variance, I'm not sure if it makes that much of a difference really. It is afterall the same computer running the exact same program with the only difference being each run is updated with the latest initial conditions.

I think I'm going to stop looking at the OPI until we hear from Riccardo at the end of the month. It's interesting but each time I think about it it gets a little bit more confusing!

[KyleHenry]

Does last winters GFS upgrade have any effect on OPI, has all the bugs been ironed out?

[Allseasons-Si]

Just been racing through this thread and what a great thread it is,there is so much informative posts in here,thanks to all

 

i would just like to ask,is there a pattern index for every month or is it just October?

[bluearmy]

Does last winters GFS upgrade have any effect on OPI, has all the bugs been ironed out?

 

was there an upgrade ?  one due dec 9th. the last gfs op upgrade was a couple years back.