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

[neige57]

Sounds as though the software is set up to reject GFS outliers.

 

I think not. There are not really any aberrant values ​​GFS last night

[chionomaniac]

I guess that bit doesn't matter for the forecast element - but I wouldn't want to see 'cherry picking' from the last 17 days actual recorded data.

[bluearmy]

I did wonder about this and was posting something in this direction yesterday evening before I got distracted and aborted.

It's good news, IMO, and gives the OPI numbers much more credence. I wonder if they are just looking at gfs op continuity or whether they have checks against the ens mean (or indeed other modelling such as ECM ens mean).

It does run the risk whereby a shift in modelling that the op has picked up ahead of the ens will not be included for a couple of runs. However, how often 500mb anomalys shift that markedly and prove to be correct must be open to question.

I am certain that there is no retrograde selection of OPI numbers - otherwise you would not see the wild swings from last year during the first half of the month. I have noticed that they select one particular run from the days output for the graph. One would hope that is selected purely for continuity purposes from the geopot anomalys rather than the number itself.

Edited October 18, 2014 by bluearmy

[Mucka]

I guess that bit doesn't matter for the forecast element - but I wouldn't want to see 'cherry picking' from the last 17 days actual recorded data.

 

We could do with knowing what the rules of the game are but the term "cherry picking" suggests selecting data to fit a preferred outcome and I see no reason they should want the OPI to be a particular figure since it is a forecasting tool not a hindsight measurement,

 

Ultimately the OPI is based on actual data anyway so it only affects the running total as far as I can see.

Edited October 18, 2014 by Mucka

[Formula_1_Fan]

Is it possible that it reaches this point in the month and they stop using forecast data and just input the actual data from each day? It might explain why the value stopped fluctuating as much last year.

I'll admit I know very little but with the pv typically getting stronger during the last part of the month do the models tend to struggle with its development if it's slightly abnormal or unusual?

[bluearmy]

Is it possible that it reaches this point in the month and they stop using forecast data and just input the actual data from each day? It might explain why the value stopped fluctuating as much last year.

I'll admit I know very little but with the pv typically getting stronger during the last part of the month do the models tend to struggle with its development if it's slightly abnormal or unusual?

They use the 10 day forecast data right out to the 31st. However, after the 21st, the forecast element drops by a day as each 24 hours passes. today, the OPI will have 17 days actual data and 10 days forecast data. There will be only three days of no data to extrapolate for those who can't wait!

[Mucka]

So I got an email response from Guido Guidi stating he will come to the thread and help answer questions if he could find the time.

[bluearmy]

So I got an email response from Guido Guidi stating he will come to the thread and help answer questions if he could find the time.

If so, we need to generate a list of relevant questions as he would not have time to trawl through the past 26 pages and dig out each question that needs answering.

Perhaps we could have an OPI question thread and the moderators could perhaps remove any non relevant ones. Guido can then answer queries more easily ?.

[Nouska]

I guess that bit doesn't matter for the forecast element - but I wouldn't want to see 'cherry picking' from the last 17 days actual recorded data.

 

They have certainly been omitting some of the less favourable runs in the past two weeks - it is a moot point as the final figure is based on the reanalysis of the whole month - at least, I hope it is.

[Interitus]

They have certainly been omitting some of the less favourable runs in the past two weeks - it is a moot point as the final figure is based on the reanalysis of the whole month - at least, I hope it is.

 

Yes that is the way it is described. As the formula requires a discrete or discernible vortex to be measured, at a guess a forecast one that is splintered could return some truly unrealistic OPI and hence AO values.

[lorenzo]

Couple of posts removed, we don't need any further sub-plots in this thread, nor for it to go off on one elsewhere. Thanks.

[bluearmy]

I didn't find your posts particularly aggressive but you made an assumption and statement that was incorrect. fwiw, relevant questioning of the theory is essential.

 

Edited October 18, 2014 by lorenzo

[karyo]

If so, we need to generate a list of relevant questions as he would not have time to trawl through the past 26 pages and dig out each question that needs answering.

Perhaps we could have an OPI question thread and the moderators could perhaps remove any non relevant ones. Guido can then answer queries more easily ?.

.

Excellent idea!

Edited October 18, 2014 by karyo

[bluearmy]

Looks like we now have three consec op gfs runs which are considered inconsistent enough to ignore. That should raise some obvious questions and the next number we see could well be quite different to the recent sub 2

[karyo]

Looks like we now have three consec op gfs runs which are considered inconsistent enough to ignore. That should raise some obvious questions and the next number we see could well be quite different to the recent sub 2

Are you sure the 6z won't update?

[bluearmy]

Are you sure the 6z won't update?

You're right karyo - I'm an hour early with my observation.

[Cloud 10]

When is this chart valid for?

 

 

It implies well below normal heights for some of Europe,but when for?

 

With my snow goggles on i see it as an anomaly for winter.

[Summer Sun]

-2.96 on today's 06z

[karyo]

You're right karyo - I'm an hour early with my observation.

Ok, we have an update: -2.96 so not bad at all! 

[bluearmy]

the 06z lost the Asian anomoly anomoly! The 21st not too far away now.

Would like to see the next few runs generate an acceptable number rather than further inconsistency.

I guess we now realise why the 18z doesn't generate an OPI.

[KyleHenry]

Current OPI is -2.96 from 6z.

So we learn some more of the technique used. If certain Parameters are present the at the 500 level then the run will be discarded until they are verified. In this case the Asian anomaly .

Edited October 18, 2014 by KyleHenry

[Timmytour]

Fascinating thread and a very entertaining read. It was like watching a TV series for me....i had to read it all the way through and not skip to the end to see what the current position was!   Glad I did as I've learned a thing or two but that's still leaves lots i don't know.

Good to see Matt Hugo on here as well.   I used to enjoy following him on Twitter and thought it strange I hadn't seen him for a while. Turns out he's blocked me!!....for what reason I can't begin to imagine! 

The strange thing for me is that I have this big gut feeling this winter won't be anything special.  But I did a little thing on monthly mean CETs stats a few weeks back..just for fun, that indicated a cold February or March could be on the way. So it's fascinating to see something with a hell of a lot more science behind it that could possibly be indicating the same. 

Some of the criticism on here was reminding me of the Great North Sea Storm prediction    that proved a damp squib but was enjoyable to follow...and then I saw Lochlomond's mention of it. I heartily agree that this is something completely different.....and hats off to those giving us the low down on it.

The main  thing I suppose is that this provides no guarantees only indicators....and I guess there's always cars that have indicated and not turned or turned without indicating.
 

Edited October 18, 2014 by Timmytour

[Mucka]

THe GFS has been the most progressive model re the attempted breakdown of the Scandi ridge with the GEM and UKMO building a cut off Scandi/Arctic high at times. This obviously has big knock on consequences for the amount of Northern blocking and the battle of the models continues today with GFS continuing to be very (even more) progressive yet UKMO if anything go the opposite way. Here are the 12z 5 day comparisons.

 

 

If the UKMO is closer to the reality then would expect the actual OPI value to be lower (more negative) than the GFS projected value (barring any big swings in last week of October that have the PV forming strongly)

Edited October 18, 2014 by Mucka

[Mucka]

I had a reply from Guido stating he has joined the forum and is awaiting posting permissions where upon he will join the thread and try to answer questions and clarify one or two things.

 

From Guido,

Ok, I created an account in order to post in the forum. Waiting for validation and permissions.

As soon as I get them I will pop up in the forum to join the discussion.

 

We or the mods can decide whether this requires a dedicated thread. I hope some of our more technical and knowledgeable members are around to join in.

 

I will post a couple of less technical questions below which I don't think have been awnswered.

 

1. How is daily index figure arrived at and what is the number posted to your Website? For example the 00z must surely be the 00z index but is the 06z figure purely based on the 06z or is it a mean value that takes account of the 00z as well and likewise is the 12z figure specific tot he run only or does it take the other two runs of the day into account? In the graph you post below is the final figure a mean average of the 3 runs?

 

2. Could you explain more about why some runs are discarded and how that affects the daily data>?

 

3. In the graph are they simply the daily totals of each day or is it a mean average of all days calculated plus any forecast days thus far?

 

I expect some would want to know more about the calculations used to arrive at the index figure and hopefully the more technically minded will have specific questions but can you explain any more about that? 

Edited October 18, 2014 by Mucka

[karyo]

 

I had a reply from Guido stating he has joined the forum and is awaiting posting permissions where upon he will join the thread and try to answer questions and clarify one or two things.

 

From Guido,

Ok, I created an account in order to post in the forum. Waiting for validation and permissions.

As soon as I get them I will pop up in the forum to join the discussion.

 

We or the mods can decide whether this requires a dedicated thread. I hope some of our more technical and knowledgeable members are around to join in.

 

That's great news!