Recretos

Since the SSW has peaked out, its time to look at model performance of GFS and ECMWF in the last 31 days (ending on 14.1, so its fresh). Might be interesting for some, and boring for others. The area of the verification is 20N-80N, basically where all the main action was. First we have the 10mb temperature bias. We can see that GFS had a warm bias which increased with the forecast time, while ECMWF had a cold bias, also increasing with the forecast time. This means that GFS has slightly too warm forecasts with time, and ECM slightly too cold. The 10mb GPH bias. Both GFS and ECM had a negative bias for heights, increasing with forecast time. Basically meaning, they were promoting a stronger vortex with the increased forecast hour, especially the ECM. The "die off" for 10mb GPH. Skill score reducing with time (GFS still very good at 240h). Basically GFS outperformed the ECM in the stratosphere, especially on the actual SSW event, as I will show below. 10mb 20N-80N, 10-day zonal wind forecast. At the time of the SSW event, GFS outperformed ECMWF (higher the score, better the forecast). This is the accuracy of the forecast issued 10 days before the forecasted date. 50mb Temperature forecast. GFS outperformed the ECM in 10day forecasts, in the SSW period. 10mb Temperature. GFS outperformed the ECM in the 10 day forecasts, except for the last 2 days. 10mb GPH, 10day forecast skill. GFS nicely outperformed the ECM in 10 day forecasts of the SSW. GFS basically had better performance with the 8-10 day forecasts of the SSW. But notice the very high skill score of the 10 day forecasts, prior to the SSW, where there is much less complex dynamics. And that is what I was telling about the power of the model stratosphere forecasts in the 240h-300h period, compared to tropospheric reliability. Best regards. p.s.: Hello @Daniel, thank you for your input. I love your Accuweather astronomy Facebook page.

Regarding downwelling and its lag/time frame: First, lets look at this: Than at this: And this: Now since this probably wont be a classical "SSW hammer" type downwelling, there are certain small issues with it. First, if the troposphere dynamics play out the way they look, we will have to first differ what "went up" and what "came down". And if they meet half way. Not literally of course, but in the similar system that I was looking at in my previous post. So I think that the classic "10-20 day" response is a bit more complicated that just fixating on those 10-20 days. Looking at dynamics along the way and searching for those "connection points", which basically are already or soon to be present, with the split all the way from troposphere into stratosphere. Those type of things just can't be overlooked in my opinion. So looking for a classical downwelling might be slightly trickier. Or, perhaps I am just talking nonsense, because this is just my first real SSW season, and I have actually just learned in the past 2 months what its all about, so maybe the practical lacking is starting to show?

If I can refer to this with some of my thoughts. Basically as you can see on that graph, there are some propagation preferences. But only time will tell if that is more of an actual propagation, or more of a reflection of the magnitude of the event. I'm guessing it is more both than more one in this time frame. A good way to filter the SSW out from the Z anom charts, is by looking at this temp chart or the zonal anomaly chart. Another important thing to consider, is the interaction between trop and strat. If we look at the Z anomaly, or the "height" anomaly, we can see the strong dynamics are both sided, and soon to kick up a notch or two, around this presented region (65-90). So if I present a look of a "classic" downwelling event, it would look something like this below. In this case you can see the upward "propagation" starting in late Nov., with stronger upper wave 1 acivity following in mid Dec. as a precursor to the SSW. Textbook stuff right there. Added are the corresponding EP-Flux, Wave1 and entire SSW displacement animation. Now the reason i am showing you this, is because this is how a classical SSW downwelling looks like by my standards. Now this year, as I mentioned before, there is more dynamic activity from both sides. I was writing about it before, recently in here: http://forum.netweat...40#entry2465902 So basically this downward propagation, what ever and how ever it might be or is, it is basically going to "meet halfway" with the raging (so to speak) troposphere. Looking at 58 and 85 as examples. Of course leading to almost instant connection between both dynamics and amplification or continuation of the pattern. As example, 1958 again. Pre SSW ///////////// Post SSW A similar case being 1985. Pre SSW ////// Before the post SSW activities began, the main high-latitude block moved closer to Greenland. //////// And Post SSW I am not trying to emphasise there will be an exact same pattern (or it might, we will see ), but I am trying to emphasise the "special" type of propagation and tropospheric connection. Also worth pointing out, is that these types of responses are kinda more "in tune" with split Polar vortex. Basically by my understanding, the height rises connect much faster throughout the atmosphere, by the troposphere and mainly the Atlantic/North Atlantic area already having ridging tendency with the precursing wave 2 for the split in the stratosphere and the heights rising in the gap. So the closest to this downwelling idea are for now the CFSv2 weekly ensembles. I know CFSv2 has some problems with stratosphere coupling, but since the SSW is already under way, and the downwelling process and effect probably wont be really that textbook like 2004, it might pick up on something. The "evolution" kinda fits the 58/85 idea, with week 2 and 3 looking more in the "strat/trop connected" state. With week 3&4, the signal strength is of course falling, due to the period length and a bigger deviation. But at least the idea is definitely there. A big factor will of course be how the polar vortex is going to behave, meaning especially the Canadian vorticity. Some ominous ideas in the GFS as I have read, regarding its movement at repositioning attempts. We shall see how thing will fall into place. GEFS kinda struggles, but that is understandable, given the deviation increase with time. Still not really that wrong with the idea or the trend. For now at least. And MJO still not looking bad, with "the big 3" riding into phase 7 and kinda stagnating there. My main point is of course to just point out some interesting things about the downwelling of this SSW, so it might not confuse a person or two. And I have to add my "signature" sentence: I apologise if all of this has already been pointed out by someone, or pointed at some papers regarding this, because then I obviously missed it. But I think that in a certain way this is kinda unique, because I am referring to our specific downwelling situation, or at least my perspective of it. Best regards.

Its not really a top secret, but you do have to dig into it, and pull some resources to find it. After the last upgrade in Feb. 2012, GEFS now has T254 spectral resolution from 0h-192h, which is around 55km, and T190 resolution (around 70km) from 192h-384h. The vertical levels were increased from 28 to 42 levels. So the current GEFS vertical top is at or around 2hpa (at or around, because it is a sigma pressure system). You can see GEFS vertical levels represented in height on the graphic below. The old GEFS before this upgrade, had 28 vertical levels with top at 10hpa. I hope I answered your question, BA.

Yes, the JMA and NOGAPS had the 10mb temp before, as far as I know. About the CMA and the Indian model, CMA has 60 vertical levels, with the top at 0.1hpa, and the Indian model is basically derived from GFS and UKMET, has 70 vertical levels, with the top around 1hpa, as far as I know. What I dont like about this particular model at Meteociel, is its ridiculous 2.5° horizontal resolution. The CMA is not really particularly skilful. I would say it is somewhere in the range of NOGAPS, with a better horizontal resolution. Imho of course. p.s.: About CMA: This year, they have in plan to "upgrade" the model, and one of the "upgrades" will be the reduced vertical resolution. The top will be only at 10mb. Well you know what they say about "Made in China".

The 10mb 65-90 graph is also on the rise. 65-25 is leveling off, and even showing a decrease, mainly due to the fact that the warm core is weakening in that area, and the cold core slightly moving into this region. it will be interesting to see, if it will climb to my representation of the 24.12 12z GFS, some time ago. http://forum.netweather.tv/topic/74587-stratosphere-temperature-watch-20122013/page__st__1420#entry2456034 Regards.

Yes Lorenzo, that could be the case. I mean, look at the current state, and the activity above 5mb. Amazing wave 1 activity propagating downwards, and also a little bit of wave 3 in there too. Tho it will weaken, it will leave its mark. The ECM EPS initial probably has the full scale representation, but the calculation regime of the EPS is still below 5mb. So I am just guessing that the ECM ensembles should correct themselves a bit, over time.

About the vertical resolution, ECM operational is going to get a huge upgrade of its vertical resolution, from 91 levels to 137 levels, with the ceiling remaining at 0.01hpa. How about that. http://www.ecmwf.int/products/changes/ifs_cycle_38r2/level_comp_91_137.html

Thank you Shuttler for your updates on the ECM EPS. Just one question: Do you see only the ensemble means, or can you see individual perturbations, or plotted all at once? The deviation is really increasing from 240h on, most of the time, so it would be interesting to see individual members. Also, given the fact that the ceiling for ECM ensembles is at 5mb (can't understand why), maybe, just maybe that is why it is losing its "momentum" regarding the more intense dynamics, and trying to reform the vortex.

About the propagation forecasts. I plotted some GFS meteograms. I think it should be interesting. Let me point out that this is not really that "straightforward" as it might seem, but I think these meteograms can show at least some idea of the trop/strat coupled activity. And let me say at the start, that the colour of the lines ( I drew in MS Paint) is basically random and doesn't have a specific meaning. A problem with these meteograms is, that it is a point forecast. So basically the activity in the atmosphere is not really straight vertical, so we would have to make meteograms through more points to see more of a 3D picture where certain activity "comes from". I am not really going to present anything new here, but I think it might be interesting for some at least. And these first meteograms go only till 192h, where there is not really that much going on up there, when compared to the 200-284h period. Not to mention that the tropospheric responses and signals start to lose reliability after 144h, and especially after 192h. This is a GPH meteogram for Greenland (more E Greenland). I chose Greenland, because it is a "point of interest" regarding to post-SSW activity. There is a response from mid down to lower stratosphere with one part of the split vortex hanging over Canada, and heights rising in over Greenland. Heights start to drop off again, with the Canadian vortex moving slowly N/NE. What is interesting, is the activity in the mid and upper troposphere. On a time scale, it kinda fits in with the wave 2 forecast, touching down into mid troposphere, and peaking around 8-10th. Now the ECM chart is a zonal mean and averaged out on a daily scale. So we don't really have a spatial presentation where exactly is the max. But we have the mid-troposphere "touch" between 70N and 80N, which is around the location of the meteogram plot. But of course the meteogram is only one single point. As we see on the chart, the activity kinda retrogrades with height. So I made a meteogram of a point more south, in north Atlantic. It is more SE of the Greenland point. And basically, while the "tropospheric wave" on the meteogram is not so strong as more north over Greenland, it is stronger in the lower and mid stratosphere. The problem is of course, that we dont really have a good enough spatial presentation from these cross-sections, where exactly the main activity lies. So basically that's where the good old charts come into play. Especially the isentropic "heights" prove useful in this case. Looking at the pole, the dynamics are not so intense, and there is some propagation from both sides, basically showing that the strat-trop interaction is not really that "straightforward" as it might seem. But of course nothing new here. Looking over to north of Alaska, thing are a bit more straightforward with less dynamics. A tropospheric response "look alike" of a stratospheric "height wave", with troposphere having a shorter wave response. Looking at the 192-384 period is a bit more tricky, because the tropospheric dynamics are much less reliable do to all the know issues with NWP's. Some propagation look alike over the pole, while nothing reliable over Greenland. I combined all the periods in MS Paint, for Greenland. There is propagation, but the tropospheric signals have a low reliability. Basically the main problem is that this is vertical representation of a single point. With all the strong and intense dynamics, we should be looking in 3D. Because when looking at a single point, you can easily overshoot or undershoot important trends. For example, where it looks like upward propagation, it might actually be downward, but in a tilted manner. So these meteograms do show a certain degree of dynamics, but cant represent dynamics which are tilted in nature. Lets shift to 2D from the top now. Now matter how I look at it, the tropospheric and stratospheric vortex split are basically simultaneous, with troposphere slightly in the lead, as shown up on the meteograms and the wave 2 graph. And what a split it is! It is in the same "spirit" as the late Nov./early Dec. split. ECM operational is really pushing the power of the Greenland blocking up to the max! The ensembles are following nicely, having a 350m anomaly on a 240h range. That tells a lot about the consistency of the ensembles and a not so bad deviation. With the ECM operational forcing such a strong block into Greenland, that of course means a strong WAA into the polar region and strong CAA from the pole, down the southern latitudes. http://shrani.si/f/3F/ql/4naemtaj/ecmepst850aarctic11.png GEFS following, or walking shoulder to shoulder with ECM. But there are some differences, as it was already pointed out a few posts back. http://shrani.si/f/N/At/2XCFxSZZ/gefsz500anomnh41.png http://shrani.si/f/1d/ed/JQUvqn7/gefsao18.png http://shrani.si/f/3W/WH/2A7ceCby/gefsnao18.png I couldn't agree more. Now the way I see it, is that this split would not have happened in this magnitude if it weren't for the SSW and the severe vortex weakening. With the vortex getting displaced and starting to weaken, the not really so strong wave 2 will begin to take it apart. And with the simultaneous or precursing tropospheric split and height rises connecting from troposphere and stratosphere, the polar vortex vortex is looking at its doom, and tropospheric higher latitudes are looking at blockings. Shall the subsequent warmings continue, the vortex will not be able to reform or restrengthen to its previous state, before the final warming onsets. http://shrani.si/f/V/Nn/131WSN8l/nhhgt30mb360.gif http://shrani.si/f/3R/rA/2wbsgTn6/nhhgt10mb360.gif Another thing worth pointing out is the EP-Flux. Just saw the reanalysis on the JMA site, and I just have to add it. Given the look on the meteograms, the ECM forecast for EP-Flux should eventually start to became more favourable. http://shrani.si/f/Y/13R/3LEnTiV0/fluxes.gif http://shrani.si/f/1i/RM/3xmFG511/ep12zstnh.gif Maybe I should also add MJO, to make this post a bit more colourful. http://shrani.si/f/B/142/17hQrnHE/allemeanphasefull.gif We are in a phase 4 which is promotive for Atlantic and Pacific ridging. http://raleighwx.americanwx.com/MJO/JanuaryPhase4.gif http://shrani.si/f/47/Bg/3KsIFWaQ/gefsz500anomnh5.png Now basically phases 5 and 6 are also promotive mainly for Atlantic and Aleutian ridging, with lower pressure being promoted over Scandinavia. And phase 7 is more or less the best looking one for January. http://raleighwx.americanwx.com/MJO/JanuaryPhase5500mbAnomalies.gif http://raleighwx.americanwx.com/MJO/JanuaryPhase6500mb.gif http://raleighwx.americanwx.com/MJO/JanuaryPhase7500mb.gif http://shrani.si/f/1E/jC/4XfnyvyA/ecmepsz500anh8.png http://shrani.si/f/2I/OC/2MYfuuXT/ecmepsz500anh11-1.png So as far as MJO goes, it might have looked unfavourable with its recent and current phases, but when looking at the bigger picture and stratospherically, Atlantic and Pacific ridging is all we could hope fore at this stage. http://forum.nwstatic.co.uk//public/style_emoticons/default/smile.png A cold shot is in the "model works", and I think we could easily link it with the stratosphere and its activities, of course coupled with troposphere. http://shrani.si/f/2o/J7/29HXOLjd/ecmt850anomeur8.png http://shrani.si/f/1U/NU/4peY4Sgw/ecmt850uveur9.png This post is just as dynamic and versatile as the current atmospheric activities, but I wasn't really writing much recently, so I had to cover more topic in this one. http://forum.nwstatic.co.uk//public/style_emoticons/default/smile.png And I admit that I still haven't read the papers on stratosphere and SSW's, especially the ones where MJO linkage is being mentioned. So I apologise if I wrote some stuff that have already been mentioned here, or in some other thread. I only follow this thread, so I apologise for that too.. And I just had to add some model chars, not to make this thread look like "model discussion", but to make sure that when we look back at this thread in some time, there will be some troposphere model output in here too, not just stratosphere charts. http://forum.nwstatic.co.uk//public/style_emoticons/default/smile.png Barely awake right now, but really glad I finally finished the post and got all of this of my chest. Just couldn't find enough time earlier. http://forum.nwstatic.co.uk//public/style_emoticons/default/sorry.gif Best regards.

@GP: how do you mean that? I thought IF was referring to Jan. 87? edit: oh, Now I understand. I wasn't 67, but 87.

And off she goes!! MJO forecasts looking better. Not really that awesome, but slowly trending to a better situation. Looking at "nearby" future: (this maybe belongs in the model thread, but since it is stratosphere connected, I think it should stay in here, so when we look back at this thread in a year or two, all the info will be here, including model output during the event. ) Beautiful "tropospheric wave2". Around that time, stratospheric wave2 slowly picking up. Basically troposphere split down the middle. And not just troposphere, but also low and mid stratosphere. Some (nice) trends: Call me crazy, but the way I see all of this, is more or less in the same light as the dynamics back in late Nov/early Dec. The big difference is this time we have an SSW in the "game". Maybe this was already pointed out and I missed it. There are some differences between these events. Most interesting being 87, with that SSW being barely a downweller. Interesting dynamics ahead, my friends. A great opportunity for all of us to learn about these dynamics in "real-time". An offtopic question: When comparing this thread to the SSW threads in the previous years, how "good" or informative is it? Regards.

Regardless of the downwelling, what we shall see (or what I see at the moment, and maybe most of you too) in the "near" future is the split connection between troposphere and stratosphere. I think that should be obvious by now. Of course given the wave2 "positioning", the tropospheric "split" dynamics are just as much tropospheric forced as they are stratospheric forced. Regards.

First just a little info about ECMWF: After all the recent upgrades, the operational run has horizontal resolution 0.125°/16km, till 240h. Control run has horizontal resolution of 0.25°/30km all the way till 384h. Ensembles have resolution of 0.25°/30km till 240h, and 0.5°/60km from 240h to 384h. Vertical resolution of the operational and the control run is 91 levels, with the top at 0.01hpa, while ensembles have 62 levels, with the top at 5hpa. And just a side info, some recent model anomaly correlation skill comparison. Ok, down the the real business. Looking at the ECM forecast, it is obvious that the split will be evident in the mid and lower stratosphere, slightly before in the mid/mid-upper strat. Looking at wave 2, it is evident lower in the strat, so it is kinda understandable. And it picks up toward the end of the period. Now looking at the end of the period, the split is well evident in the lower stratosphere, all the way into upper troposphere, and I think even in the mid troposphere. ECM operational, and the ensembles, are picking up the tropospheric response to the split with ease, with two ridge axis, one from Atlantic and one from Pacific, and well defined Polar Vortex split vortices. And GEFS isn't far behind either. CFSv2 weeklies are now starting to pick up on these signals too, on the 500mb anomaly, with week 3 possibly having a better SSW response. Had to get it off my chest. Best regards.

Interesting, but 84/85 was a direct split SSW.

You also have to realize, that 100mb level is in between the tropospheric and stratospheric forcing (leaning almost more toward the stratosphere), while H500 is much more "in tune" with the tropospheric forcings. And besides that, I think you will find much better correlation between H500 and GPH100, than the 100mb temperature.

For Jayces: Edit: Probably the most awesome stratospheric contrast I have ever seen.

Just some latest GEFS trends. AO NAO

Well I was only trying to answer him. And you just made me look like I don't realize what at least the main global drivers are. Thanks for that. If I would want to show him correlations for every important piece of the puzzle, there is not enough space in the post. And I event cant make correlations for all, and the problem is that some if not almost all are successive "through" each other. But ok. Cheers.

@EML: Basically you should treat this chart the same way as any ECM/GFS geopotential height chart. The contours represent the height of the 30mb pressure level in Dam (Dekameter). So if you see a contour having a 2352 label for example, that would translate to roughly 23.520m height. About the "blob" over Iceland. I drew streamlines on you chart, for a better idea of the flow. Basically a complete split yes. I wouldn't put much focus on that blob in this range, but the way I see it, its just a small negative vorticity center, on the edge of two bigger positive vortices. Might have a future within next runs. edit: And as always, I was beaten to it by someone else. I need to speed up.

Oh the dailies. Yea I think they have some wild ideas for some time now. Dailies are basically just a time downscaled version of the monthlies if I'm not mistaken. A little something about CFSv2 and stratosphere. I showed this CFSv2 generated blocking forecast some time ago. I will test it in our case, to see just how legitimate it is, and if it will pick up any possible tropospheric response and how far in advance. But about CFSv2 and stratosphere in general, here are a few facts, by Shaw&Perlwitz: - In CFSv2 polar vortex is too weak in early winter and too strong in late winter. - Variability of stratospheric polar vortex is similar between CFSv2 and Reanalysis. - Dynamic coupling between the stratosphere and troposphere is not well represented in the CFSv2. - CFSv2 has a serious spin up problem that is mostly pronounced in the stratosphere but can also be seen in the troposphere - will degrade any assimilated stratospheric information with potential benefit for improved tropospheric forecast. - Doesn't generate QBO very well and has a lower number of vertical levels, which will need to be upgraded. I am not saying CFSv2 wont pick up any signals, but it has some known problems with stratosphere coupling. It will still be useful as a test, to see how good or bad it really is. P.s.: If I can add my personal thought on CFSv2 monthlies and the seasonal runs: I really really don't like it too much. No offence to the great people of NCEP , but I think that this CFSv2 operational system has to stop. Like right away. The old CFS had an ensemble system, but this new one, has an operational system, meaning it has 4 daily operational runs. Like the GFS system for example. The old one also had 4 runs, but was more like the GEFS system. Basically it had much less variability or the "jumping around". CFSv2 has some great data assimilation systems, but this operational forecasting system has to go away. CFSv2 assimilation system and an ensemble forecast system would be the bomb if you ask me. Of course I dont mean it would be as accurate as it gets, but it would pick up seasonal and sub-seasonal pattern changes much quicker. For example, here are the last 25 runs from CFSv2, 500mb anomaly forecast for January. That is only one month lead time, where it should show some consistency, but it is basically allover the place. P.s.2: Hello and welcome scellus.

We were all very busy looking at 300h-384h periods in the stratosphere. But now I will change my focus a little bit, to the same period, but to GEFS. Cant wait to see what its tropospheric response is going to be. And dont worry, I am more than aware of all the "tricks" and "dangers" of ensemble forecasting and "anomaly based forecasting" in this range. . Animation is a bit lower quality, due to the resize.

If you know how to interpret these correlations, I made some general AO and NAO correlations for you. If you want to see some correlations with the stratosphere (Tho very noisy), I made a post a few days ago on that. I see LomondSnowstorm beat me too it. Those stratosphere correlations are too superficial and noisy if you ask me. Check a few pages back (link below) where I tried to do it. It took a lot of imagination to finally filter it to get the SSW signals out, and its still not a clear signal or a clear enough correlation for someone to say "so that's where the SSW is going to strike". I would need more time to work with the correlations to maybe get a better picture. http://forum.netweat...00#entry2455573 But I think there were some papers written on this subject.

That is all because of the "Polar High". I am not sure how to exactly name that anticyclone, or what its technical term is. Maybe Polar Anti-Vortex? Here are a few graphics, that will help you understand. Here I drew the "streamlines".

Thanks Chiono for those charts. Last few runs were really over the top, to put it that way. And I agree with Chiono, that this could in fact be classified as an SSW. No doubt about it. The thing that bugs me at the moment is the classified type of the SSW. I have heard people saying its a split SSW on the GFS. Of course its a split after the vortex is getting beaten down and basically getting ripped apart. But the way I see it, this should be classified as a displacement SSW, because the SSW and the "lethal blow" is supposed to come with the strong wave 1 and the displacement. For me, a split type SSW is by definition and by an example, the Jan/Feb 2009 SSW. So feel free to correct me if I am wrong, but how would you personally classify this upcoming event? Or maybe I missed something here. The upper strat. looks like its going to get literally overpowered by the High. p.s.: Now we just need the ECM to kick things up a notch or two.