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

[mulzy]

as I thought mate!!!12z looked pretty damn good for a negative opi!!!

 

Good call shaky!  I got it wrong...

[MrsDee]

While I appreciate that the NH October circulation might indeed give an indication of what the dominant winter circulation might be, I find the statistics of the OPI 'averaging' really odd.  

 

E.g.  the 31st October OPI (which is seemingly the overall October OPI) depends on the 22nd October OPI... which in turn depends on a simulated 0-9 day forecast. So my (surely flawed?) interpretation gives a final 2014 OPI that depends on forecast approximations.

 

It's possible that a forecast contains hidden info that's lost with the actual evolution, but it seems odd.

 

Also, surely the 1980s data points weren't calculated using 10 day forecasts... modelling at 10+ days was really poor in those days, so that graph with nice correlations must use different calculations in different decades.

 

But then that does away with any reason for using 10 day forecasts now.

 

If last century's data points used reanalysis data, then strictly speaking to be consistent we need to redo the calculations after the 31st October using actual weather data, not forecast data.  Or perhaps they do this anyway, I don't know.

 

And why not use 3rd October - 2nd November:  does the theory just collapse then or what?

 

(A confused Mrs Dee)

 

 

[lorenzo]

Here are a few composites from the figures on the original OPI original chart for comparison with October to date and projected NH runs.

 

Negative (All)  / Negative (range -0.5 > -2.0)

 

Neutral

 

Positive

 

I think the axis angle is still the thing that is the great unknown in this index, not withstanding the original query around the bust years.

 

Read today that the bust years correlate somewhat with Solar Max, although we have last year with a Solar Max that looks like previous Solar neutral phases for flux values, so not entirely convinced this explains the random predictions in the time series.

 

With some other very solid known analogs on the cards, then it will be an interesting test for this index to see if it can get back on track with correlation over this winter, following which a paper would be good to read !

[Mucka]

 

With regards to the geopotential anomaly part of the equation (in the Elypticization box) the formula is (Asia + America - Arctic)/10 so a high north pole anomaly can be cancelled out by smaller positive anomalies in the other two regions.

The axis angle has generally been the larger factor but is very sensitive to the locations of low pressures skirting round the edge of the vortex, both from timeframe to timeframe and run to run.

 

 

Good information, thank you.

So as I understand it that translates as a perfect set up for high negative OPI  being low pressure over Asia and America (North or South?) and high pressure over the Arctic?

[shaky]

Good call shaky!  I got it wrong...

its pretty tricky cos every run seems to be different!!but with plenty of blocking over the Arctic for the next 10 days this month should be a negative one!!!dont forget the models have all of a sudden picked up on some pretty high heights across Scandinavia in the next week to 10 days!!!only god knows whats around the corner!!lol

[easton888]

Anyone know how the Siberian snow cover is developing? And how does it compare to last year?

[SE Blizzards]

Anyone know how the Siberian snow cover is developing? And how does it compare to last year?

 

Click on link below - Northern Hemisphere snow and ice cover. There you can compare years etc.

 

 

http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=10&fd=16&fy=2013&sm=10&sd=14&sy=2014

 

Siberian snow cover developing very nicely by the look of things.

 

Great to see the OPI still in negative territory. Some great posts in here guys. Keep up the good work 

Edited October 16, 2014 by SE Blizzards

[Duane S.]

While I appreciate that the NH October circulation might indeed give an indication of what the dominant winter circulation might be, I find the statistics of the OPI 'averaging' really odd.

E.g. the 31st October OPI (which is seemingly the overall October OPI) depends on the 22nd October OPI... which in turn depends on a simulated 0-9 day forecast. So my (surely flawed?) interpretation gives a final 2014 OPI that depends on forecast approximations.

It's possible that a forecast contains hidden info that's lost with the actual evolution, but it seems odd.

Also, surely the 1980s data points weren't calculated using 10 day forecasts... modelling at 10+ days was really poor in those days, so that graph with nice correlations must use different calculations in different decades.

But then that does away with any reason for using 10 day forecasts now.

If last century's data points used reanalysis data, then strictly speaking to be consistent we need to redo the calculations after the 31st October using actual weather data, not forecast data. Or perhaps they do this anyway, I don't know.

And why not use 3rd October - 2nd November: does the theory just collapse then or what?

(A confused Mrs Dee)

Some good points there, to be honest though I don't think anyone here is any wiser now than at the start of the month regards the exact workings of this formula. Possibly even less so!

We have to assume that the forecast elements are being overridden constantly by actual data. If there is forecast data in any element of the final OPI then it will be flawed indeed. Reanalysis is fine as it will only be subject to actual data, albeit possibly compromised slightly in terms of absolute accuracy.

Edited October 16, 2014 by s4lancia

[terrier]

Just hope this opi is onto something because all long range models I'm seeing are going for another above average winter with the jma updating today also looks westerly based for uk.

[Interitus]

While I appreciate that the NH October circulation might indeed give an indication of what the dominant winter circulation might be, I find the statistics of the OPI 'averaging' really odd.  

 

E.g.  the 31st October OPI (which is seemingly the overall October OPI) depends on the 22nd October OPI... which in turn depends on a simulated 0-9 day forecast. So my (surely flawed?) interpretation gives a final 2014 OPI that depends on forecast approximations.

 

It's possible that a forecast contains hidden info that's lost with the actual evolution, but it seems odd.

 

Also, surely the 1980s data points weren't calculated using 10 day forecasts... modelling at 10+ days was really poor in those days, so that graph with nice correlations must use different calculations in different decades.

 

But then that does away with any reason for using 10 day forecasts now.

 

If last century's data points used reanalysis data, then strictly speaking to be consistent we need to redo the calculations after the 31st October using actual weather data, not forecast data.  Or perhaps they do this anyway, I don't know.

 

And why not use 3rd October - 2nd November:  does the theory just collapse then or what?

 

(A confused Mrs Dee)

 

No by the end of the month there will be no forecast data, just reanalysis data of what actually happened, and this is how the previous years were calculated. The forecast data just adds a bit of interest!

 

Further to the image from Nouska, here is one which they posted last year in an analysis of October 1984 which gives a good idea what they do - sorry it's very small because the original has been removed from the host site, this was taken from Google cache -

 

 

The axis is clear to see from the mean geopotential height for the month on the left, compared to looking at the anomaly chart on the right. 

 

Good information, thank you.

So as I understand it that translates as a perfect set up for high negative OPI  being low pressure over Asia and America (North or South?) and high pressure over the Arctic?

 

Yes that would give a lower OPI.

Edited October 16, 2014 by Interitus

[LomondSnowstorm]

While I appreciate that the NH October circulation might indeed give an indication of what the dominant winter circulation might be, I find the statistics of the OPI 'averaging' really odd.  

 

E.g.  the 31st October OPI (which is seemingly the overall October OPI) depends on the 22nd October OPI... which in turn depends on a simulated 0-9 day forecast. So my (surely flawed?) interpretation gives a final 2014 OPI that depends on forecast approximations.

 

It's possible that a forecast contains hidden info that's lost with the actual evolution, but it seems odd.

 

Also, surely the 1980s data points weren't calculated using 10 day forecasts... modelling at 10+ days was really poor in those days, so that graph with nice correlations must use different calculations in different decades.

 

But then that does away with any reason for using 10 day forecasts now.

 

If last century's data points used reanalysis data, then strictly speaking to be consistent we need to redo the calculations after the 31st October using actual weather data, not forecast data.  Or perhaps they do this anyway, I don't know.

 

And why not use 3rd October - 2nd November:  does the theory just collapse then or what?

 

(A confused Mrs Dee)

As far as I'm aware, Riccardo can correct me if I'm wrong, the final OPI is based entirely on the actual axis and geopotential values rather than the predicted ones, which are simply to give us their up to date best estimate of what the OPI will look like. It would make more sense, to me at least, if the website graph showed the actual values rather than the predicted ones from each day, but that's how they've set it up. The choice of October I think is because it's the month where the vortex begins to form properly, although it would be interesting to see if moving the dates a few days either way changed the indices, and in particular the correlation with the AO.

Cheers for posting those Tony, the main indicator seems to be the height anomaly over the pole, while the rest isn't as immediately obvious.

[Duane S.]

Further to Lomond's excellent post above. Maybe it would be better to have one running actual figure and one with forecast data.

It would indeed be interesting to see if the already impressive 0.9 correlation to the winter AO could be bettered by altering the duration and/or start finish dates to find the optimum period to record this. Logic says there must be scope for further improvement as it is currently based only on data from an exact calendar month. But then again I guess it serves its purpose.

[KyleHenry]

I wanted to make a point and a hat tip to Riccardo's instinct to open the OPI thread here on Netweather.

The knowledge on this forum is in itself, really a peer review.

Every foreseeable question is being asked and to a high level of where the doubts and true understandings of the Index model are.

Criticism is lurking for such a ground breaking method of long range forecasting. I wish him well.

[Duane S.]

Let's hope 2013 is typical in terms of when we should expect the OPI daily value to start showing more consistency.

The 18th of the month was nearly smack on the final OPI value (as was the 16th, with the final spike of the month on the 17th) and then the day to day variance was no more than +-0.25 than the OPI value of the 18th thereafter.

[nick sussex]

At the end of the month will there be a re-analysis of the whole month to get to the actual final figure?

 

In terms of last year is the lack of vatriation more to do with the fact that the pattern remained stable and so I'm thinking we shouldn't use that as a guide because surely if the last week of this October showed a marked change this could have a more detrimental effect to the final figure.

 

Aswell as this is it possible to do a correlation between first half/second half of month and following AO, do we have a connection between negative OPI that was caused by a more negative second half than first half and the following winter.

 

Logic dictates IMO that the more important part of the month is the second half as this is where the "average" year is more likely to see the PV winding up so set against that a negative second half would be more symbolic of the actual state of the NH.

 

What I'm asking is do we have a list of all data for OPI that gives us a full breakdown of daily figures going back 30 years?

Edited October 16, 2014 by nick sussex

[Mark Bayley]

A very interesting topic and great to see the level of debate that this new index has created. As its not peer reviewed, nor are there any available published studies (im aware there an article which was removed?), i will treat the subject with caution (and urge others too). However, the correlations and analysis do point to this OPI index being significant in forecasting the winter AO. Will be interesting to see how well this predicts the coming winters AO.    

Edited October 16, 2014 by Mark Bayley

[Nouska]

Here are a few composites from the figures on the original OPI original chart for comparison with October to date and projected NH runs.

 

Negative (All)  / Negative (range -0.5 > -2.0)

OPI below -0.5.pngOPI below -0.5 to -2.0.png

 

Neutral

OPI Neutral -0.5 to 0.5.png

 

Positive

OPI Positive +0.5.png

 

I think the axis angle is still the thing that is the great unknown in this index, not withstanding the original query around the bust years.

 

Read today that the bust years correlate somewhat with Solar Max, although we have last year with a Solar Max that looks like previous Solar neutral phases for flux values, so not entirely convinced this explains the random predictions in the time series.

 

With some other very solid known analogs on the cards, then it will be an interesting test for this index to see if it can get back on track with correlation over this winter, following which a paper would be good to read !

 

When I was looking at the graph in the EPF research paper I posted in the Stratosphere thread, it immediately caught my eye that there was a correlation with the misfit parts of the OPI graph in the early decades. As solar max progresses, the fit gets better - pattern established by then whereas transition is more difficult to call?

 

No match in the super Nino years of 97/98 though. There was the double whammy of solar max and high energetic particle count on two out of three occasions - with what we now know about solar effects I wouldn't rule it out as a contributor to the discrepancy.

 

 

[Interitus]

At the end of the month will there be a re-analysis of the whole month to get to the actual final figure?

 

In terms of last year is the lack of vatriation more to do with the fact that the pattern remained stable and so I'm thinking we shouldn't use that as a guide because surely if the last week of this October showed a marked change this could have a more detrimental effect to the final figure.

 

Aswell as this is it possible to do a correlation between first half/second half of month and following AO, do we have a connection between negative OPI that was caused by a more negative second half than first half and the following winter.

 

Logic dictates IMO that the more important part of the month is the second half as this is where the "average" year is more likely to see the PV winding up so set against that a negative second half would be more symbolic of the actual state of the NH.

 

What I'm asking is do we have a list of all data for OPI that gives us a full breakdown of daily figures going back 30 years?

 

I suspect that they just use the overall monthly mean and anomaly charts, doubt that there are extensive daily figures.

One point about the pattern remaining stable, possibly but there is also something else about the way the axis angle value is calculated in that it appears to only give a negative contribution to OPI.

The lower the angle, the exponentially lower the OPI, angles nearer to 34.37° give increasingly insignificant negative values and above 34.37° all OPI contribution is zero. Hence an axis near this cut-off angle will have OPI largely determined only by the geopotential anomalies,

[nick sussex]

I suspect that they just use the overall monthly mean and anomaly charts, doubt that there are extensive daily figures.

One point about the pattern remaining stable, possibly but there is also something else about the way the axis angle value is calculated in that it appears to only give a negative contribution to OPI.

The lower the angle, the exponentially lower the OPI, angles nearer to 34.37° give increasingly insignificant negative values and above 34.37° all OPI contribution is zero. Hence an axis near this cut-off angle will have OPI largely determined only by the geopotential anomalies,

Thanks for that. The problem is though that without daily data when we look at those previous years we have no way of knowing where the main negativity was , perhaps those negative AO months were connected with a certain sub section of the month rather than the month itself.

 

IMO and this is just using logic is that the beneficial angle is more likely to come under pressure towards the second half. Another factor which is difficult to determine, what is our normal baseline, without many more years of data we don't know what to regard as normal. It could be that October when viewed over a 100 years will normally have a tendency to produce a  negative value.

 

Equally how do we know whether another factor or factors is working in tandem with the OPI to produce the negative AO. Is there another variable that we haven't accounted for?

 

I certainly do hope that the OPI is indeed the key factor as it certainly would be a big help to forecasting and of course without people doing this research we wouldn't increase our understanding.

Edited October 16, 2014 by nick sussex

[KyleHenry]

As Riccardo pointed out in first post, from today the 16th October to months end.

The daily OPI figures will use more solid consolidation data from GPS, less forecast.

I see the final OPI more directly based on winter in the Central European Continent.

As we move further out towards the UK there would need to be an adjusted OPI for us based on geographical location.

SE, East and NE coasts with less deductions and SW and Ireland with the most.(Height above sea level also a feature)

Today's -3.26 is a very impressive starting point.

I'm starting to see how maybe the equation for OPI was discovered by reversing through all analogs, all data from Solar minimum years(snow cover etc) a pattern was found. I'm guessing through many mathematical corrections to the OPI the equation was discovered.

Very clever and observant. How many years of work for Riccardo and Mr Cohen to achieve this , I wonder.

Edited October 16, 2014 by KyleHenry

[bluearmy]

another sub 3 number from the 00z.

Looking through the gefs and gem ens , those members keen to build a compact vortex with a lack of wave breaking into the polar field are very much in the minority. yesterday's musings about a second half October with a much higher OPI could well be awry.

[Blessed Weather]

Yes bluearmy, a very encouraging Index of -3.36 from the GFS 0z run.

 

Yesterday I posted the ECM/GFS 8-10 day Nth Hemisphere mean height anomaly from Weds 12z runs and the heights had disappeared from the Pole. I'm glad to say that last night's 12z output have the heights back again, as well as (to my eye) better positioning of anomalies in relation to the top ten -ve OPI composite chart.

 

I have no idea what signal the GFS had picked up to drop the heights on Weds, then have them back on Thurs (bearing in mind this is a 3 day mean chart out at days 8 to 10). But a wise post from John Holmes yesterday saying not to worry, you shouldn't take just one run!

 

Height Anomaly for 24th - 26th Oct:

 

Edit: Meant to say GFS has heights back, ECM remains reluctant. 

Edited October 17, 2014 by Blessed Weather

[MPG]

So the OPI follows the GFS output? So we are happy the heights are back therefore the OPI stays negative........I thought the GFS wasn't rated compared to the ECM?

 

So if the ECM was being used the OPI could be in the positive?

Edited October 17, 2014 by MPG

[KyleHenry]

another sub 3 number from the 00z.

Looking through the gefs and gem ens , those members keen to build a compact vortex with a lack of wave breaking into the polar field are very much in the minority. yesterday's musings about a second half October with a much higher OPI could well be awry.

I myself was one of those who felt that way. The second half of October must carry a higher percentage of the final OPI than the first half, maybe 65% to 35% .

I'm the first to say I was wrong and it does now look like it could be an impressively negative OPI .

Forgot to mention Lt Col Guido Guidi in my last post.

[LomondSnowstorm]

So the OPI follows the GFS output? So we are happy the heights are back therefore the OPI stays negative........I thought the GFS wasn't rated compared to the ECM?

 

So if the ECM was being used the OPI could be in the positive?

It still wouldn't be positive since only around 1/5 of the value comes from the 'unreliable' timeframe as opposed to the actual or reliable, but it would've likely been higher on yesterday's 12Z.

The 00Z ECM looks like a significant improvement on yesterday's run, which was a particularly poor run as far as the Arctic profile goes compared to the rest of the model output, so I reckon there isn't that much between them this morning.