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

[karyo]

of course - but then it could mean the opposite !

It is too late in the month to worry now.

[johnholmes]

 

judging from tonights NOAA cpc update, we may have to wait a few more days before we see a landing zone < 0.5 spread

'SIGNIFICANT UNCERTAINTY IS EVIDENT IN THE MODEL FORECASTS OVER ALASKA AND THE  NORTH PACIFIC EVEN AT RELATIVELY EARLY LEAD TIMES.' 

 

fairly consistently NOAA anomaly outputs have shown a +ve area of heights of SW Alaska, not coincident with any ridge. However this may be sufficient to help prolong the fairly mobile set up looking likely on these same anomaly charts for the Atlantic and into the UK. This does NOT of course mean the winter will be mild , wet and windy. But have a look back at October last year for the ridging and +ve heights around through much of October. It may well be simple coincidence.

[radiohead]

-2.37 this morning. Almost identical to yesterday's value following the 12Z.

Looks like the wild swings are over and we are closing in on a final figure.

Edited October 24, 2014 by radiohead

[Isolated Frost]

-2.37 this morning. Almost identical to yesterday's value following the 12Z.

Looks like the wild swings are over and we are closing in on a final figure.

The last 4 values before were

-1.89

-2.99

-2.75

-2.39

I'd be wary to say we're closing in, I think once we get to +96 (27th) we can be a lot more sure of a 0.2-0.4 landing figure; but the next couple days will change a bit I think.

[Radiating Dendrite]

I agree that there will still be variation but I think a sub -1.5 is a near cert now.

Let's be optimistic

[Penrith Snow]

-2.4 would still be very good and the OPI is already taking into account the less favourable synoptics over the next 7 days which ais hardly FI when looking at the broad scale pattern.

If it's below -1.5 then I think a cold winter is ahead and will put my reputation on the line by telling people so, it's the time of year everyone asks me what the winter will be like!

This year the answer to that might be easier than normal.

Andy

[Duane S.]

Sub -2 is looking likely but far from sure at this time. Sub -1.5 is almost in the bag though I would say and as the data data shows that is all we need for fun and games come wintertime!

One thing puzzles me with the OPI, the first calculation, so is it geopotential height anomalies? In which case, why is the heading 'elypticization' not a real word I don't think but implies shape (as in egg) more than a pressure anomaly to me

[lorenzo]

Don't worry about the twists and turns running out in this thread, once we have the final value the OPI team will also go on to release the predicted wave activity..

 

And we're off... again !

[Mapantz]

-2.4 would still be very good and the OPI is already taking into account the less favourable synoptics over the next 7 days which ais hardly FI when looking at the broad scale pattern.

If it's below -1.5 then I think a cold winter is ahead and will put my reputation on the line by telling people so, it's the time of year everyone asks me what the winter will be like!

This year the answer to that might be easier than normal.

Andy

 

You really think it's that easy? 

[Duane S.]

You really think it's that easy?

No but it would undeniably put us in a ten fold better position going into the winter than 13/14 ever did. There is no doubt that the early creation of the vortex is key to how the NH winter pans out, in general terms. If we have a vortex that is 'allowed' to grow as we head out of Autumn and into winter into its natural compact shape, history has shown how robust and difficult it is to shift and as a consequence HLB will simply not be able to form in any consequential format until the vortex is compromised and that can take until late winter in those cases. The QBO at least is easily forecastable and will be strong Easterly and likely, if not hinder, at least not aid the creation of a growing vortex as a W-QBO set up surely must do.

Same caveats as always though, all of the above does not = a cold winter for the UK as there are plenty of -AO set ups that won't bring cold to our shores but it does help load the dice that bit more. In real terms it would mean increased opportunities of cold spells throughout the winter and then we will need a bit of luck thrown in for good measure, as always.

Edited October 24, 2014 by s4lancia

[Yarmy]

You really think it's that easy? 

 

Indeed. A feature on the current charts and one that has persisted for most of the year (August aside) is for stubborn heights over Western Europe. The OPI site's map of Europe implies (if things stay as they are) that heights over Western Europe should be significantly below average for winter (assuming I'm reading it right and the unit in the legend is DAM). So let's see how that pans out.

Edited October 24, 2014 by Yarmy

[SomeLikeItHot]

One thing puzzles me with the OPI, the first calculation, so is it geopotential height anomalies? In which case, why is the heading 'elypticization' not a real word I don't think but implies shape (as in egg) more than a pressure anomaly to me

 

Could be a poor translation from italian. I presume they meant a quantity like eccentricity, but perhaps not defined in exactly the same way.

[Mapantz]

No but it would undeniably put us in a ten fold better position going into the winter than 13/14 ever did. There is no doubt that the early creation of the vortex is key to how the NH winter pans out, in general terms. If we have a vortex that is 'allowed' to grow as we head out of Autumn and into winter into its natural compact shape, history has shown how robust and difficult it is to shift and as a consequence HLB will simply not be able to form in any consequential format until the vortex is compromised and that can take until late winter in those cases. The QBO at least is easily forecastable and will be strong Easterly and likely, if not hinder, at least not aid the creation of a growing vortex as a W-QBO set up surely must do.

Same caveats as always though, all of the above does not = a cold winter for the UK as there are plenty of -AO set ups that won't bring cold to our shores but it does help load the dice that bit more. In real terms it would mean increased opportunities of cold spells throughout the winter and then we will need a bit of luck thrown in for good measure, as always.

 

As you well know, we've had all those factors many times before, how many times have you seen snow in the last twenty years? Would you be telling all your locals that we're in for a cold Winter because of a negative number?

[Penrith Snow]

Any winter forecast based on the OPI has got to be better than the formula the MetO use!

Remember their 2009/10 winter prediction lol

Had we known about the OPI in October 2009 none of us would have predicted a mild winter!

Andy

[mcweather]

As you well know, we've had all those factors many times before, how many times have you seen snow in the last twenty years? Would you be telling all your locals that we're in for a cold Winter because of a negative number?

 

To be fair Mapantz I cross referenced all the years with a with an OPI number greater ( more negative than minus one) and found a very good correlation

for worthwhile snowfall in Dorset, 

[karyo]

Excellent update again this morning based on the 0z gfs which is the least blocked output this morning. ECM looks much better with much higher pressure over the Arctic.

[Mapantz]

To be fair Mapantz I cross referenced all the years with a with an OPI number greater ( more negative than minus one) and found a very good correlation

for worthwhile snowfall in Dorset, 

 

How about from 1994 to 2009? I saw nowt until February 09.

 

Edit: Misread that. More negative than a minus one?

Edited October 24, 2014 by Mapantz

[phil nw.]

Let's not spoil this thread by debating past Winters peeps.There are other threads for that.

 

The negative readings look like they are steadying between -2 and -2.5 with only about 1 week to go. 

It 's looking more and more likely it will get a good test against the raised expectancy by many of a colder than average Winter.

[Interitus]

I don't understand why you are trying to correlate the AO mode in October, with the AO mode in the winter- the statistics - even the AO table on the website shows the data where the relationship isn't SIGNIFICANT.

Also the problem with doing 'all' months & segregating them into a mere pos or negative is you have no neutral point, with this what you actually doing is reducing your correlation.

 

 

 

The division of data into positive or negative has a very obvious neutral point of zero but is nothing to do with the correlation calculation, it is different statistical measure, binomial chance like tossing a coin as purely by guessing it should be possible to be in the right ball park for winter AO half the time - certainly as close as last year's OPI for example.

It is valuable to employ alternative statistical measures, consider this very basic example - a series of winters have AO -1, -3, -2, -0.5, -2 and -1 but our forecast gave 3, 2.5, 2.75, 3.125, 2.75, 3 and 1 - oh dear miles out - except they have a perfect positive correlation of 1. (Obviously this case would be trivial to correct).

 

Comparison of the October and winter AO represents a base state, the climatological persistence so this is a starting point to improve upon.

Thanks to BFTV converting the graph we had values to work with and see if there was anything to the index, but what are the variables measured?

An obvious starting point for the Arctic value is geopotential anomaly at the north pole, taking the height in metres and divide by -100 (or dam/-10) to give an approximate value for winter AO.

This showed a marked improvement over the October AO so could be an indication that the OPI was on to something, but just as interesting were the values from 1980, 1993 and 1994: -0.05, -0.15 and 0.7....remarkably these were exactly identical to the values from BFTV.

Now remember the OPI formula that I worked out -

(Asia + America - Arctic)/10 + F(34.37 - angle) ....where F is some exponential function.

Observation showed that the result of right hand side of the equation was restricted to negative values and with large (>34.37) angles = 0

The left hand side is simple, the division by 10 suggests the variables are anomaly heights in dam, and the whole OPI calculation can be reduced to one variable if the other two cancel out - so it is possible that those years did boil down to just the north pole anomaly. All three predictions were close with an average error of 0.14 - from just one sampling point.

Whilst toying around with other values to plug into the formula, the link to the Taymyr paper appeared and as it isn't freely available one cannot say how closely it resembles the Polar/Eurasia pattern - however their winter AO prediction figures are much better.

The value I calculated correlates quite highly to, but definitely isn't the Polar/Eurasia pattern - it is measured at a single arbitrary point at the tip of the Taymyr peninsula irrespective of anomalies elsewhere and it too outperformed that pattern whether measured by chance, correlation, RMSE or whatever, and much better than the north pole anom.

Once again the point is that far from debasing the OPI, I am finding evidence supportive of it in that relatively simple measurements derived in October can be indicators for the winter.

With all respect to the creators, anybody with a credible theory can just post a list of values and claim a breakthrough.

There is 4th way of dealing with the lack of recorded visible data

* investigate and examine the calculation to try to understand the results - or you can ramp and rant.

[karyo]

I am keen to see the index update based on the 6z as it has more blocking in the Arctic (similar to ECM in that respect). 

[karyo]

-2.28 based on the 6z so very stable now 

[Crocodile23]

I have made a simple calculation about the correlation between October's OPI and the average temperature of following winter in the 850 hPa level for a big portion of UK. (I provide the values i used in the end of this post)
 
So i took the October's OPI value and the following winter's average temperatures in degrees Celcius in the 850 hPa level for december+january+february, for the area that is defined by 5° W to 0° E and 50° N to 55° N (that contains most of England and Wales) and have calculated Pearson's r correlation coefficient and Kendall's tau.
The values of temperatures have been acquired from NOAA and the values for OPI from a post here.
 
So the calculated values between October's OPI compared to following December+January+February average temperature in 850 hPa were:

 

Pearson's r = 0.76

 

Kendall's tau = 0.55

2-sided p-value = 0.0000014
 
Pearson's r value is relatively high and shows that something might going on with this OPI thing.

Even more, in the Kendall's calculation, we see a smaller correlation but a very interesting p-value, since the almost zero 2-sided p-value shows that assuming(the null hypothesis) October OPI and average temperature of DJF are independent, we have an extremely small chance of obtaining this 0.55 correlation if October OPI and average temperature of DJF are independent.

So we must assume they are dependent with a +0.55 (positive) correlation.

Positiv emeans that when the value of October OPI is large(get's larger) the average temperature of DJF is large(get's larger) also and when the values of October OPI decrease, the value of average temperature of DJF decrease also.

 

So all in all there is something interesting with this OPI.

And it seems to suggest that a correlation with your winter's temperature lies between them. A positive OPI brings warmer winter and a negative colder. With the aforementioned correlations of course. This is NOT a strict 1 to 1 rule of course.

 

 

The data:

(For example for year 1980:

OPI value is for October 1980

Average temperature(in 850 hPa) is for December 1980 + January 1981 + February 1981)

YEAR	OPI	AVG1976	-1,75	-3,311977	-0,95	-1,901978	-1,80	-3,081979	-0,50	-1,711980	-0,05	-2,051981	-0,40	-1,771982	-1,10	-1,841983	-0,30	-1,511984	-1,80	-2,191985	-1,90	-3,311986	-1,30	-2,381987	-0,45	-0,981988	+1,45	+0,491989	+0,25	+0,061990	+0,55	-2,611991	+1,10	+1,651992	+1,75	-0,011993	-0,15	-2,141994	+0,70	-0,841995	-0,65	-2,051996	-0,75	-1,021997	-0,75	+0,311998	+0,10	-1,061999	-0,20	-1,462000	-0,75	-1,262001	+0,45	-0,482002	-0,90	-1,232003	-0,70	-0,622004	+0,30	-1,572005	-0,70	-1,102006	+0,85	+0,472007	+0,75	+0,902008	+0,25	-1,032009	-3,15	-3,882010	-0,85	-1,902011	+0,65	-0,852012	-1,65	-1,902013	+1,60	-0,53

Edited October 24, 2014 by Crocodile23

[SomeLikeItHot]

 

I have made a simple calculation about the correlation between October's OPI and the average temperature of following winter in the 850 hPa level for a big portion of UK. (I provide the values i used in the end of this post)

 

So i took the October's OPI value and the following winter's average temperatures in degrees Celcius in the 850 hPa level for december+january+february, for the area that is defined by 5° W to 0° E and 50° N to 55° N (that contains most of England and Wales) and have calculated Pearson's r correlation coefficient and Kendall's tau.

The values of temperatures have been acquired from NOAA and the values for OPI from a post here.

 

So the calculated values between October's OPI compared to following December+January+February average temperature in 850 hPa were:

 

Pearson's r = 0.76

 

Kendall's tau = 0.55

2-sided p-value = 0.0000014

 

Pearson's r value is relatively high and shows that something might going on with this OPI thing.

Even more, in the Kendall's calculation, we see a smaller correlation but a very interesting p-value, since the almost zero 2-sided p-value shows that assuming(the null hypothesis) October OPI and average temperature of DJF are independent, we have an extremely small chance of obtaining this 0.55 correlation if October OPI and average temperature of DJF are independent.

So we must assume they are dependent with a +0.55 (positive) correlation.

Positiv emeans that when the value of October OPI is large(get's larger) the average temperature of DJF is large(get's larger) also and when the values of October OPI decrease, the value of average temperature of DJF decrease also.

 

So all in all there is something interesting with this OPI.

And it seems to suggest that a correlation with your winter's temperature lies between them. A positive OPI brings warmer winter and a negative colder. With the aforementioned correlations of course. This is NOT a strict 1 to 1 rule of course.

 

 

Interesting post however the independence tests don't really work when you have selected specific data and combined them in a particular way to match the historical series which is what has been done here.  Its not like the OPI is say a relatively simple temperature or pressure measurement at one point.

[radiohead]

I'm not sure if this has been highlighted already. This was part of the OPI team's forecast for North America last year :

 

 

 

 

And this is how last DJF actually turned out.... :

 

[SMU]

The division of data into positive or negative has a very obvious neutral point of zero but is nothing to do with the correlation calculation, it is different statistical measure, binomial chance like tossing a coin as purely by guessing it should be possible to be in the right ball park for winter AO half the time - certainly as close as last year's OPI for example.

It is valuable to employ alternative statistical measures, consider this very basic example - a series of winters have AO -1, -3, -2, -0.5, -2 and -1 but our forecast gave 3, 2.5, 2.75, 3.125, 2.75, 3 and 1 - oh dear miles out - except they have a perfect positive correlation of 1. (Obviously this case would be trivial to correct).

 

Comparison of the October and winter AO represents a base state, the climatological persistence so this is a starting point to improve upon.

Thanks to BFTV converting the graph we had values to work with and see if there was anything to the index, but what are the variables measured?

An obvious starting point for the Arctic value is geopotential anomaly at the north pole, taking the height in metres and divide by -100 (or dam/-10) to give an approximate value for winter AO.

This showed a marked improvement over the October AO so could be an indication that the OPI was on to something, but just as interesting were the values from 1980, 1993 and 1994: -0.05, -0.15 and 0.7....remarkably these were exactly identical to the values from BFTV.

Now remember the OPI formula that I worked out -

(Asia + America - Arctic)/10 + F(34.37 - angle) ....where F is some exponential function.

Observation showed that the result of right hand side of the equation was restricted to negative values and with large (>34.37) angles = 0

The left hand side is simple, the division by 10 suggests the variables are anomaly heights in dam, and the whole OPI calculation can be reduced to one variable if the other two cancel out - so it is possible that those years did boil down to just the north pole anomaly. All three predictions were close with an average error of 0.14 - from just one sampling point.

Whilst toying around with other values to plug into the formula, the link to the Taymyr paper appeared and as it isn't freely available one cannot say how closely it resembles the Polar/Eurasia pattern - however their winter AO prediction figures are much better.

The value I calculated correlates quite highly to, but definitely isn't the Polar/Eurasia pattern - it is measured at a single arbitrary point at the tip of the Taymyr peninsula irrespective of anomalies elsewhere and it too outperformed that pattern whether measured by chance, correlation, RMSE or whatever, and much better than the north pole anom.

Once again the point is that far from debasing the OPI, I am finding evidence supportive of it in that relatively simple measurements derived in October can be indicators for the winter.

With all respect to the creators, anybody with a credible theory can just post a list of values and claim a breakthrough.

There is 4th way of dealing with the lack of recorded visible data

* investigate and examine the calculation to try to understand the results - or you can ramp and rant.

 

Afternoon- Thanks for the reply, & perhaps shed some light on your thoughts & with this in mind looking at the information of course a fair way of challenging the process, IE whilst we don't have the specifics in terms of numbers if you breakdown the equation we can more or less get the answers anyway without the actuals.

 

Im perhaps still perhaps surprised you chase the AO though as my personal time & research lead me to believe that 'same' correlation with October that you have identified, which is that positive Octobers ( all be it slightly) end up with a higher correlation to negative Winters than a negative link-

However the caveat that I have noticed is more pertinent for recent years is the correlation of winter AO with significantly negative October months- We are talking in the order of indexing around 1.5.

So whilst you go along a reasonable poor correlation for low index negative Octobers all of a sudden you reach a point ( in recent years) where theres a sharp correlation.

 

I do believe this is in relation to the ever changing arctic & how the Taymyr region in particular produces a feedback that has an inverse impact on the height anomaly which is-  the more negative the sea anomaly to the Norm the more positive the height anomaly is. This manifests across late September & early October where this region at sea level is warmer than the average so the ice growth at the edges recedes further to the pole & takes longer to grow back.

Because the overall mechanism that is creating this feedback is becoming self perpetuating & the signal for melt becomes larger each year ( over the period of time ) my thoughts on this feedback is that we are yet to see the full extent how 'deep' this anomaly will get. 

 

Heres an October anomaly graph at 500MB

 

1948-2006

 

2007 -2013  - post lowest ice extent.

 

Surface temp anomaly.

NOTE the direct feedback related to the Taymyr region.

 

Relating that back to the equation...

 

(Asia + America - Arctic)/10 + F(34.37 - angle)

 

You have correlated the October AO correlation based on the fact the A+A is cancelled out.

 

Well

 

I think your going in the wrong direction - Why? Because the Artic signal on the loading diagrams isn't the one that has the biggest variable & its slowly being muted out by the other anomalies, moreover this is why I believe the OPI index has good legs.

 

The Calculation towards negative is driven by the 2 subsequent anomalies that are created when you get a huge Pos height anomaly in the Taymyr region, we are of course referring to Russia west & the western Atlantic.

 

So instead on focusing on the AO, we should perhaps focus on the TCI ( or whatever it was in that link)-

I don't personally label this index, Perhaps TCI, perhaps POL -

Either way its the Sea ice feedback anomaly in that region.

 

is it of no surprise that the record year for the OPI is the year is closely correlated to the record Negative POL for October 2009 @ -1.98.

 

Something else to consider is the effect of this Taymyr anomaly on the UK temps for October.

Look at the small euro height anomaly - it will in the long run make UK Octobers warmer !!!

 

Further to this the October Taymyr Anomaly drives the SAI index more positive towards Quadrent 4 - NW Europe so that this feedback Also impacting on the AO suggests that there are 2 clear mechanisms working to make NW europes & central Europes Winters Colder.

 

In Summary,

 

I would say using the Oct AO is a little pointless as its not the variable on the move , the OPI clearly earmarks the 2 locations that are correlated to nearly r=1 in terms of feedbacks to the TAYMYR anomaly area. & as a result because the feedback is stacked towards negative expect the OPI to sway more negative in the years to come as that Anomaly area becomes more Stark.

 

Best regards

Steve

Edited October 24, 2014 by Steve Murr