In Depth Model Discussion and Summaries

[johnholmes]

dropping this in here as it may be easier to find it when I need to see how daft the comments in it are!

 

Changes in the 500mb noaa charts - looking 6 nov 13.pdf

[johnholmes]

been away fri am until 1 h ago but will try and assess how these models are coping and what looks like happening

[johnholmes]

The height The height predicted for English channel is about 6dm lower than 4 days ago and the flow on both 6-10 and 8-14 is n of west which is pretty different to 4 days ago when I last saw them and did my post in the technical thread.

The odds then I felt were against this pattern developing, see my last pdf, but with the probability that it could happen. It looks as if the lesser probability may well occur based on what I have seen so far. Not having seen any 500mb anomaly outputs since the 00z Friday ec-gfs I am not yet totally sure but it does look like yet again the anomaly charts gave a decent indication and a consistent one about a week ahead of a pattern change.

One thing which few will believe is that the small rise in the 30mb basic chart I use did give an indication that this might happen around the 3^rd week in November. That is in the very simplistic way I use it. It gave such indications in my view in 3 previous early winters but I suspect the more knowledgeable folk with Stratopshere work will not accept my suggestion for this. It went against every other teleconnection, MJO, AO, NAO, PNA and 500mb anomaly charts as far ahead as they went.

Never mind I am sure the cold lovers will be happy whatever anything shows if it snows in their back yard. Not time now but I will do an in depth look at the anomaly charts when I get time.

In the meantime below are the anomaly charts from today.

Ec-gfs

http://www.meteo.psu.edu/~fxg1/ECMWF_12z/hgtcomp.html

noaa

http://www.cpc.ncep.noaa.gov/products/predictions/610day/500mb.php

 

copied into the very busy model thread also

[joggs]

Thanks John,its good to get over here away from the hustle of the output thread and take time to view this,keep up the good work and yes this sort of output showing against most of the telecons!!!!!!.

Just shows that anything may/may not happen when things say otherwise=weather.

[johnholmes]

Apologies but its looking like I am not going to be able to do the detailed look at the lead up to the change on the anomaly charts that in turn predicted the colder spell of weather for about 5 daysfrom now.

 

In brief the anomaly charts, the 3 I use, plus NAEFS, all show the colder weather with an upper ridge west/NW of the UK and lower heights NE-SE of the UK taking hold from about 144h and keeping this out to 240h. Beyond that and the 'signal' for this pattern begins to look less reliable but 10 days is a long way off. What follows beyond that will be clearer in say 3-5 days time. It is possible that the colder type may last, once it starts, to about the end of November. beyond that is outside my ability to make sensible comment.

As always the detail on how cold, dry or not so dry and what falls from the sky is something the shorter term synoptic patterns are for-watch the Fax charts as a good guide to how these matters may develop and Exeter and ECMWF for T+72 down and GFS if it fits the other two models.

[johnholmes]

An in depth but easy to follow view of what the 500mb anomaly charts showed from last weekend and what they are suggesting for the next 14 days or so.

 

What can we deduce from the 500mb anomaly charts-14 nov 13.pdf

 

also copied into the model thread

[Gael_Force]

For those of you who like to examine the entrails of NWP verification statistics, there is a lot of ensemble performance analysis on this site. Some may be surprising - one thing that strikes me is the error increase during the period 2010 to 2012 across all models - what was causing this I wonder.?

 

http://tparc.mri-jma.go.jp/TIGGE/tigge_score_all.html

[Catacol]

For those of you who like to examine the entrails of NWP verification statistics, there is a lot of ensemble performance analysis on this site. Some may be surprising - one thing that strikes me is the error increase during the period 2010 to 2012 across all models - what was causing this I wonder.?

 

http://tparc.mri-jma.go.jp/TIGGE/tigge_score_all.html

 

The really long range stuff of 9 - 15 days has generally gone downhill in recent years looking at thos graphs and the running 91 day mean. Fascinating. Some new element is unsettling the models - the most likely it would seem to me would be solar - and if some are to be believed we are heading for a period of low solar activity. Bottom line is that something in our weather has changed... though observers know this very well without needing to look at statistics.

[forecaster]

Bottom line is that something in our weather has changed... though observers know this very well without needing to look at statistics.

 

I'm very sceptical of statements like these. Humans are fine tuned to find patterns. Sometimes this works well for us, sometimes our signals analysis is simply terrible and we spot a pattern which is not real. This is why we need things like statistics. We have internal biases and we need something to stop us getting carried away with them. See here (http://psiexp.ss.uci.edu/research/teaching/Tversky_Kahneman_1974.pdf) for an example study of all the sorts of illusions we internally (not necessarily on purpose) cook up. 

 

For me, the ultimate annoying sentence in this context is the one that starts with "scientists have found what __________ have known all along........". (Often, the ________ is "traditional people" or something like that).Well, there is plenty of stuff that we've "known all along" which has turned out to be a laughably inept representation of reality.  

 

So, when people say "something has changed in our recent winters", I want more proof than some vague observations without broader context. Now I don't have any strong personal opinions on this. Maybe something has changed (sea ice, snow cover, maybe, PDO...????), maybe it hasn't. I wait to be convinced. But if I am to be convinced, it will be with science and statistics, not solely by the claims of people. Observations have a major part to play (essentially, the major part!), but only in an objective and broad methodology. 

 

After all, even if something has changed in recent winters, we still experienced November 2011.

[Interitus]

For those of you who like to examine the entrails of NWP verification statistics, there is a lot of ensemble performance analysis on this site. Some may be surprising - one thing that strikes me is the error increase during the period 2010 to 2012 across all models - what was causing this I wonder.?

 

http://tparc.mri-jma.go.jp/TIGGE/tigge_score_all.html

 

Sorry, which graphs are we looking at?

From what I can see the anomaly correlations (AC) are going up, and the root mean square errors (RMSE) are going down which is good.

[Catacol]

I'm very sceptical of statements like these. Humans are fine tuned to find patterns. Sometimes this works well for us, sometimes our signals analysis is simply terrible and we spot a pattern which is not real. This is why we need things like statistics. We have internal biases and we need something to stop us getting carried away with them. See here (http://psiexp.ss.uci.edu/research/teaching/Tversky_Kahneman_1974.pdf) for an example study of all the sorts of illusions we internally (not necessarily on purpose) cook up. 

 

For me, the ultimate annoying sentence in this context is the one that starts with "scientists have found what __________ have known all along........". (Often, the ________ is "traditional people" or something like that).

Well, there is plenty of stuff that we've "known all along" which has turned out to be a laughably inept representation of reality.  

 

So, when people say "something has changed in our recent winters", I want more proof than some vague observations without broader context. Now I don't have any strong personal opinions on this. Maybe something has changed (sea ice, snow cover, maybe, PDO...????), maybe it hasn't. I wait to be convinced. But if I am to be convinced, it will be with science and statistics, not solely by the claims of people. Observations have a major part to play (essentially, the major part!), but only in an objective and broad methodology. 

 

After all, even if something has changed in recent winters, we still experienced November 2011.

 

OK - if you want data how about these 2 plots. The first is the anomaly dec-feb from the last 6 years when many observers believe that something changed:

 

 

 

and the second is the same seasonal anomaly for the 10 years prior to that:

 

 

 

There is a pretty obvious shift in pressure anomalies and therefore in the weather type we are having in the current sequence of years during winter. Most significant is all is the widespread positive anomalies that were globally present prior to 2007. Pressure has fallen at mid latitudes since then, and especially to our south over mainland Europe. Observers such as myself would bear this out - the element that has been absent in recent years is the Euro high, though last winter the azores high made a mess of many a potential easterly! Notice also the total flip in pressure patterns in the north pacific. Why have we gone from well below to well above average?? PDO phase only?

 

It is also evident from recent years that the wave pattern has been much more meridional - you can almost see the wave pattern from those anomalies. No such obvious meridionality from the period prior to 2007.

Edited November 21, 2013 by Catacol_Highlander

[Interitus]

 Most significant is all is the widespread positive anomalies that were globally present prior to 2007.

 

Caution with that particular conclusion from those plots - the bottom chart has three sub 10m bands which are all coloured white in the top chart so while it may be true they aren't a fair comparison in that respect.

[Catacol]

Caution with that particular conclusion from those plots - the bottom chart has three sub 10m bands which are all coloured white in the top chart so while it may be true they aren't a fair comparison in that respect.

 

Isnt that the point? The positive anomalies around much of the NH have declined, and in particular the big negative anomaly in the northern pacific has been transformed into a big positive anomaly. We are seeing lots of this again this year, with the aleutian high controlling the hemispheric pattern. 

 

So pressure patterns have, in that sense, flipped. I dont see how the comparison isnt fair. Can you be more specific?

[Interitus]

Isnt that the point? The positive anomalies around much of the NH have declined, and in particular the big negative anomaly in the northern pacific has been transformed into a big positive anomaly. We are seeing lots of this again this year, with the aleutian high controlling the hemispheric pattern. 

 

So pressure patterns have, in that sense, flipped. I dont see how the comparison isnt fair. Can you be more specific?

 

Sorry it's not really relevant to your argument, I was only referring to the lower latitudes where the pre-2007 chart seems to have almost complete +ve anomalies they are for anything >2.5m where as to be shaded on the other chart they have to be >10m.

So the +ve anomalies at lower latitudes could be pretty much as widespread more recently too.

[Catacol]

Sorry it's not really relevant to your argument, I was only referring to the lower latitudes where the pre-2007 chart seems to have almost complete +ve anomalies they are for anything >2.5m where as to be shaded on the other chart they have to be >10m.

So the +ve anomalies at lower latitudes could be pretty much as widespread more recently too.

 

Ah!!! Got it. Somehow the 2 charts I plotted have got a different scale. That's odd.... used the same parameters. I'll give it a go again.

 

 

EDIT - got the scale and parameters the same on the original post now.

Edited November 21, 2013 by Catacol_Highlander

[johnholmes]

pdf also in model thread

 

What might the upper air pattern be as we move into December-22 nov 13.pdf

[lorenzo]

Been a bit of a week in the model thread ! Good this lazy one is here, helps to keep track of things.. 

 

Looking at December and with such variety of divergence as the models get to grips with a pattern change it is tricky to see a way ahead.

 

So much going on to get through - The wave activity in the strat. showing on the models has been inconsistent, the Pacific SST profile is cooler, but with a warmer profile beneath the surface. The AO threatens after some massive model spread earlier in the week to set it's positive stall out, and we are seemingly at the mercy of the subtle orientation if the ridge to our west.

 

This not withstanding the ECM upgraded this week so who knows if that requires fine tuning.. GEFS and ECM Ens quite different on the AO outlook, although the means are similar, the control runs are in complete contrast.

 

Made some composites based on perhaps more predictable factors for the 1st - 10th December.

 

These are based on the 50mb QBO, ENSO MEI index and AAM. QBO figure is 7 to 10 ms, ENSO is a wider filter -1 to 1. All dates from 1974 to 2009 using the daily dataset

 

First is for QBO and ENSO Neutral, some similarities to current model output, namely the troughing into the heart of Europe, this has to be a signature pattern now surely, also the mid-Atlantic ridging.

 

Next is QBO without the ENSO Filter

 

Next is QBO for dates with a lower AAM

 

Recent long range mean plots from GEFS and ECM Ensemble.

 

GEFS days 10 and 15

 

ECM day 15

 

Variations on a theme, and perhaps a backdrop for the start of December.

 

 

 

 

 

[Catacol]

Anyone know why some copy and pasted charts disappear a few days later?

[lorenzo]

MJO starting to feature a little more as it has left the Circle of Doom ( origin).

 

GEFS into phase 2 and ECM into Phase 3

 

UKMO nearer the ECM side of things.

 

Couple of composites similar to previous, taking account of Westerly QBO and Neutral MEI. Also filtered for the GWO moving through lower AAM phases 1-4.

 

 

Low AAM

 

Matt just tweeted that EC32 goes for the anti-cyclonic outlook and these 'background' states would support this. Now to figure out the rest of the month !!

[lorenzo]

Very interesting update, some further information on the mysterious October Pattern Index (OPI) from a co-author Alessandro, this explains a little more around the science and method they are using to calculate the OPI.

 

 

Hello guys. First of all, forgive me the language .. I do not speak well English. My name is Alessandro and I am the co-author of the research OPI together with Riccardo.  want to compliment you for the wonderful topic and specify only a concept about the difference between OPI and SAI .

 

OPI essentially measure the position and degree of intrusiveness of the planetary waves in the month of October by two parameters:

1) degree of ellitticizzazione / elongation

2) axis of polar vortex

 

When we have a very tilted axis ( iceland -east Siberia / Aleutine ) and a pv stretched in October, like 2005/2006.

a part of the rate of snow cover can not be measured as the snowfall occurring in the sea under the 60th parallel. 

In these cases the SAI approximates very well unless the AO . 

 

For this reason it is important to measure the position and the intrusiveness of the planetary waves rather than the snow which is nothing other than a consequence of the pattern of October.

Soon we'll end the English translation and will submit to the competent authorities in the USA .

 

Best regards ;

Alessandro

[Cloud 10]

Just been having a browse through the AO index for November's and it seems as though we are on for a new record +ve for November 2013,with the final figure for the month set to be around +3

 

 

AO monthly records (back to 1950)

 

http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/monthly.ao.index.b50.current.ascii.table

 

 

The only November that is anywhere near is 1978 which came in at nearly +2.5,composite below.

 

 

 

When looking at the composite for December 1978,which sees a complete pattern reversal to a negative AO,it does make me wonder what winter 2013/14 really has in store.

 

 

 

[Gael_Force]

I see there was a Canadian warming in December 1978 - that may have had a bearing on the turnabout in the heights. From what I can see that was also a westerly QBO and must have been close to solar maximum as the ssn was high in January.

[lorenzo]

GLAAM tendency still falling into the low momentum phases.

 

GEFS forecast sees the orbit through phases 2 and 3

 

Composites below

 

Then onwards to

 

MJO plots for the models are still quite scattered, GFS bringing in phase 2, however ECM and UKMO not in support.

 

From Adrian Matthews page, only the briefest of forcing which possibly will not qualify as an active MJO.

 

Saw a good post where it was discussed that some anomalies creating transport on the wheeler diagram for MJO forecasts are a result of CCKW, Convectively Coupled Kelvin Waves, they create a noise signature in the models which mimics OLR and hence we get an MJO forecast. Might need to start looking more closely at OLR - Outgoing Longwave Radiation plots in conjunction with whatever is forecast for verification.

 

With the above in mind, would be following what the vortex is up to and comparing what is the lead on the 100mb strat charts.

 

 

 

 

[forecaster]

Saw a good post where it was discussed that some anomalies creating transport on the wheeler diagram for MJO forecasts are a result of CCKW, Convectively Coupled Kelvin Waves, they create a noise signature in the models which mimics OLR and hence we get an MJO forecast. Might need to start looking more closely at OLR - Outgoing Longwave Radiation plots in conjunction with whatever is forecast for verification.

olr.anom.90day.gif

 

Me being no expert on equatorial waves, but I think what actually happens is that the CCKW's do not mimic an OLR signal, they create it. OLR anomalies (outgoing longwave radiation - not sure if that's appearing on the new mouse-over?) is a good proxy for convection in the tropics. CCKW's are a fastish, eastwards moving wave which come with a convective signature that shows up in the OLR anomalies. So the OLR signature is true, and real.

I think the issue is that models find it hard to distinguish this from the MJO wave.

 

(Broadscale) Convection in the tropics can be due to CCKW's, mixed Rossby-Gravity Waves, tropical cyclones, cold fronts, convergence zones, equatorial Rossby Waves, favoured zones of convection due to ENSO state.....and probably a lot more. The above operate on different time-scales and can interfere with each other. 

To isolate the contributions of different waves, they run their signals analysis (which I think is a fourier analysis) and try to identify which waves are contributing to the current convection. This can be difficult. It may be especially so when you're dealing with CCKW's and the MJO, because the two behave quite similarly. 

The structure of the MJO is often similar to a CCKW, but it moves more slowly. 

 

The models might perk up a bit when the CCKW and MJO are no longer in phase with each other and causing confusion of the signal. 

[lorenzo]

Me being no expert on equatorial waves, but I think what actually happens is that the CCKW's do not mimic an OLR signal, they create it. OLR anomalies (outgoing longwave radiation - not sure if that's appearing on the new mouse-over?) is a good proxy for convection in the tropics. CCKW's are a fastish, eastwards moving wave which come with a convective signature that shows up in the OLR anomalies. So the OLR signature is true, and real.I think the issue is that models find it hard to distinguish this from the MJO wave.

 

(Broadscale) Convection in the tropics can be due to CCKW's, mixed Rossby-Gravity Waves, tropical cyclones, cold fronts, convergence zones, equatorial Rossby Waves, favoured zones of convection due to ENSO state.....and probably a lot more. The above operate on different time-scales and can interfere with each other. 

To isolate the contributions of different waves, they run their signals analysis (which I think is a fourier analysis) and try to identify which waves are contributing to the current convection. This can be difficult. It may be especially so when you're dealing with CCKW's and the MJO, because the two behave quite similarly. 

The structure of the MJO is often similar to a CCKW, but it moves more slowly. 

 

The models might perk up a bit when the CCKW and MJO are no longer in phase with each other and causing confusion of the signal. 

 

Thanks for the further explanation, it is something I was beginning to look at before being distracted by the Northerly, the above bit in bold was generally what I was trying to say.

 

The chart I am looking at and getting my head round for this is located here http://www.atmos.albany.edu/facstaff/roundy/waves/hovsdet/0_7.5N/

 

Latest forecast example 

 

The MJO CPC update comes out each Monday evening. Mike Ventrice site is also an amazing resource of everything in the tropics, can't begin to describe everything here http://mikeventrice.weebly.com/tropical-waves.html

 

MJO shown in red here with Kelvin Waves in Black

 

Plotted a couple of composites to try and fathom whether MJO activity would add bearing to the first half of December. Looking at the phase of the GWO, phase 2 Northward momentum transport, a low AAM state and taking into account the neutral ENSO state. These are all dates since 74, filtered for phase 2/3 of GWO and ENSO -1 > 1 values.  Then again with MJO filtered for phases 2/3/4.

 

Not withstanding the west atlantic ridging am unsure about these plots, what is clear from the dates when filtered that there is a healthy split between developing positive NAO regime and the Scandi Block. At present, the positive NAO I believe has the edge. Not great news for folks hunting cold, and hope to be wrong !

 

Having talked myself out of the Scandi high verifying, for reference this morning 00z normalised Ensemble mean, advertising the Scandi High, this for 14th, which helps change my mind back the other way ! ECM will have to do 'very' well with this to help everyone get over 'that ECM'..

 

Edited December 2, 2013 by lorenzo