Mike Poole

The zonal winds chart is interesting, in as much as there seems to be an increasing cluster going for a second, deeper, dip (increase in easterly winds) about a week after the initial SSW. Will make for an interesting March, I think, and a delayed start to spring.

It does usually, when it’s on the back of a strong strat vortex, but there aren’t any significant westerlies there to flush down in this case, because of the weak strat vortex all winter, and of course the previous SSW in January. For example, GFS 12z zonal winds: If there were strong westerlies to flush you’d see a strong pink anomaly (right chart) in the lower strat moving down with time with the blue anomaly from the reversal above it. It isn’t there.

ECM clusters T192-T240: Cluster 2 (contains the op) looks interesting more as to what comes after, with low Euro heights and a ridge building west. T264+: Once again, promise in the extended. Cluster 2 the puck with south Greenland high, cluster 1 with Atlantic ridge also cold. Clusters 3 and 4 northwesterly wind for the UK, cluster 5 high retrogressing a bit. All very jam tomorrow, but there’s wasps in the jar by now.

We had this discussion in the run up to winter. The problem is, these days, it is very rare to get colder winter weather after any September (cold, average, warm, the lot). And with climate change, warmer Septembers than historic averages are more common so it is possible to perceive a link that is not there. Correlation is possible, therefore, but it doesn’t imply causation. There is no logical reason I can see why a warm September in the UK of all places should affect weather patterns three months later. (There are other places in the world where Autumn weather may affect the coming winter because it affects the developing polar vortex - the UK is not one of them!) It may just be one of those statistical anomalies (like the 13th June used to be), but it is a theory that is rolled out every year. Problem is, after this year, its nose will grow a few inches!

ECM clusters T192-T240: Cluster 2 and 5 have the ridge in the Atlantic west of the UK, cluster 5 a bit pathetic really as it only has 2 members. Block’s not up to much on clusters 3 or 4 and cluster 1 it remains mid-latitudes during this timeframe. T264+: Clusters 3 and 4 develop a strong block over Greenland, although on cluster 4 it looks to be weakening at day 15. Both are well orientated. Cluster 1 and 2 have more of a ridge in the Atlantic, but both pull in air from the north by day 15. So, again, in this timeframe, there is potential, not sure if it is related to the SSW, but it will need to get its skates on if it is to turn into anything meaningful.

ECM clusters tonight show some movement, T192-T240: The op is in cluster 2, which has the ridge go up west of the UK - it has 14 members. Cluster 1 similar but doesn’t get as far north. Cluster 3 pushes the ridge over the UK and it ends up to the east. Clusters 4 and 5 don’t really push the ridge up much at all. T264+: Roughly even split between the two clusters. Cluster 1 has a strong block which is also well positioned. Cluster 2 with a weaker ridge that eventually moves west of the UK, but maintains heights to our south.

Run of the day is the KMA 12z, here at T216: The key to this, as I’ve said before, is getting the initial push of heights north to take place west of the UK and not over the UK. Will be interesting to see what the EPS show in terms of the likelihood of this.

ECM showing a much more positive evolution out to T192:

This thread is very necessary! And, quite fitting, in a weird way, to have the post-mortem while this excuse for a winter is still alive! Before giving my thoughts on winter, I just wanted to make a comment on the GSDM - I don’t usually, because I don’t want to get caught in the crossfire between those who are passionate believers and very defensive when challenged, and those who use it as an easy target for criticism when it doesn’t snow. I don’t have access to the GSDM charts myself, but I do factor into my thoughts the excellent posts on here from those that do. I have read some papers on it, and @Tamara kindly provided me some useful help on aspects of it a few years ago - it seems sound and very useful, and given it is founded on the principle of conservation of angular momentum, it makes sense to me as a physicist. I don’t think it is infallible though, forecasts seem to be predicated on the assumption that AAM will rise/fall in a predictable way, and I also have a question mark (in winter) about how intuitions derived from the GSDM fit with things occurring in the vertical dimension i.e. the stratosphere, which I think has been relevant this winter. It also doesn’t inform on local UK-scale issues which can be so important. The backdrop to this winter was world weather patterns out of past precedence with extraordinary sea surface temperatures. My view in the run up to winter, was that any attempt to make predictions based on analogues was therefore pointless, and I put more weight in the seasonal models - because they simply simulate the laws of physics, without knowledge of past patterns, I figured they would be the best guide. They weren’t! At best, they could be said to have predicted a weak vortex, which there definitely was, but on pressure patterns relevant to the UK, they have not performed well. We have had a very weak strat vortex ever since the Canadian warming (first one for 30 odd years). The trop has been leading the dance most of winter, and the strat vortex has been behaving really quite oddly. I thought that surely with that in our favour we must be in with a shout of a decent winter. Wrong! But why? It seems to me that there are two primary possible reasons: Climate change has done something to global weather patterns that have made it next to impossible to get the kind of synoptic patterns which in the past would have delivered cold to the UK - the dreaded ‘Iberian high’ being a symptom (but not cause) of this. We were just very unlucky this year. Even unfettered by the strat vortex, the trop vortex just never got itself into a favourable position. I still think with a different roll of the dice we could have ended up with something memorable. One of the reasons I don’t (yet) believe reason 1, that decent UK winter patterns have become next to impossible, is because they have showed up in the models this year, not just the odd run, but substantial clusters of ensemble runs - we have just not managed to count them down into the reliable. They would not show up if they were no longer physically possible, for whatever reason, due to climate change. So, I put most of it down to bad luck. Which brings me to my final learning point which is disregarding the model output. It is OK to bin an individual run if the output gets so contorted you know (from looking at charts that have actually verified) that a run is garbage. But, I think we have been a little quick to dismiss larger swathes of model output on the assumption that ‘the models are misreading the progression of the MJO’, ‘the models always underestimate Iberian heights’, ‘the ECM is prone to over-amplification’, or ‘the GFS is off on one and will come into line’ etc. I hold my hand up - I’ve done this - and found later that the disgraced model was on to something after all. All in all, a very disappointing winter, but while we have had no meaningful snow (locally, again, 4 years on the trot now), it has been a (morbidly) fascinating learning experience.

ECM clusters T192-T240: Clusters 3 and 5 look good for a cold plunge by day 10, clusters 2 and 4 have things too far west, cluster 6 settles on a UK high, cluster 1 can’t lose heights to the south. T264+: Cluster 3 looks good with a nicely placed ridge into Greenland by day 15, cluster 2 has a weak Greenland block that goes a bit too far west, cluster 1 rather has the blocking collapse. There’s still potential here, and with the SSW to factor into the mix, cluster 1 looks to me unlikely. But getting anything at all into the reliable just seems problematic…

ECM clusters T192-T240: All four clusters look blocked, cluster 4 the strongest with a persistent block in Greenland. Cluster 3 has things a little too far east, cluster 2 the Atlantic low is too progressive, cluster 1 the build of heights too much through the UK. T264+: There does seem a genuine theme in these, that wasn’t there this morning or in yesterday evening’s output, for a continuation of a blocked theme, a block somewhere around Greenland in all 5 clusters at day 15. Pointless looking in any detail at that, and if it is fallout from the SSW, then the models will make a meal of it before deciding on anything useful in terms of a forecast, but the interest remains, even if the clock is ticking into the red zone…

feb1991blizzard Yes, would agree we need a good EPS suite to give even an entry pass to the last chance saloon. The SSW might still deliver, but the trouble is, SSW, MJO, GWO, all are backed by good science but I wouldn’t trust any of them to deliver cold to the UK. The opposite, though, if any signal is predicting mild, it’ll be bang on!

ECM extended clusters, T264+: These are not as good as the 0z by a margin. Cluster 4 looks decent, cluster 6 looks OK ish, cluster 3 the block is too far north and west, and the others the blocking is disintegrating - I think that is the change from this morning, this is no longer trending towards a blocked outlook.

Met4Cast Clusters 3 (with a westwards tweak) and 5 look decent also. Better than I was expecting at T240 to be honest. Clearly there is some signal still to build the ridge west of the UK.

Yes, another poor ECM - will wait to see what the ensembles say, but I would not be surprised to see a down turn after promising clusters this morning. You’re right about the PV - there has been an extremely weak strat vortex all winter (since the Canadian warming, in fact), which another time could have produced an epic winter, but as far as the trop patterns were concerned this year what can go wrong has gone wrong. For snow in my location, and barring a major turnaround in the next 20 days, it has been the 4th consecutive 0/10 winter. Never mind, next winter model watching will be much easier. My predictions are: Westerly QBO winter. After brief northerly tease, rampant PV sets up shop over Greenland in the first week of December. Safe to declare winter over on 7th December and we can go and do something more productive instead.

UKMO T168: Much better, more room between the low and the UK.

Models at T144: GFS and UKMO have the more negative tilt to the low to the west compared to the GEM. But all look promising, let’s see how they go. GFS has more weight to the small low near the UK, UKMO hardly has it at all - which will be better, I think. GEM somewhere in the middle.

Lift off at day 10 on the GFS 6z. And west of the UK, just.

ECM clusters looking much more promising this morning, T120-T168: The point of interest in this timeframe is the ridge starting to go up, notable on clusters 1,3,4 (not so much 2) at T168. It is crucial that this happens west of the UK in the Atlantic. T192-T240: Cluster 2 still has this ridge too close to the UK. Cluster 1 holds promise as regards building a well placed block to the NW, and cluster 3 is excellent - I think the JMA solution would fit well in this one - quite a turnaround seeing this within 10 days. T264+: Plenty of interest here, apart from cluster 1 where the block is too far west and the UK is in westerlies. Clusters 2 and 4 are exactly what we want to see as regards Greenland block placement and trough into Europe. Cluster 3 is OK, but dealing with the aftermath of the ridge through the UK earlier I suspect. Cluster 5 brings in the chance of an easterly with heights to the north more generally. Back in the game, this morning.

I reckon the JMA has this nailed at day 9 this morning:

I think we may see a lot of this sort of evolution over the next few suites, GFS T318: What I will be looking for is for the transfer of heights north to occur to our west in the Atlantic, rather than through Spain up through the UK. There is enough time for this shift to happen as the shenanigans associated with the SSW (if it occurs, still not sure) kicks in, I think.

Daniel* Yes, the way the strat and trop have been in step this year, it has been quite difficult to tease out what is cause and what is effect. Should the split vortex SSW happen, I would expect a very quick trop response, even to the extent of both impacts appearing to happen pretty much simultaneously.

Really interesting to see how AI forecasting develops. I’ve been impressed by what I’ve seen from the ones on the ECM website, astounding, actually, that they can even hold their own with the big NWP models. Presumably though, none of the pressure fields that they generate are actually constrained by the laws of physics? For the NWP, since they simulate the laws of physics (or their best approximation) all patterns generated are physically possible - that wouldn’t be the case with AI. Does this matter? If it gets the gist right, would people care? There must be a huge amount that AI could learn from the reanalysis data - but we wouldn’t get the understanding, just the output, who knows what teleconnections AI might discover but have no way of communicating to us. Fascinating new story of weather models, which we are only just at the beginning of.

Models at T144: Interesting differences in terms of the depth of the two upstream lows (marked 1 and 2 on the GFS chart). GFS has the first low more developed, therefore the WAA ahead of it, GEM the two are more equal so there is some WAA beginning ahead of the second one, and UKMO has the first one practically non-existent, which means the WAA will be further west ahead of the second one (we can check at T168). I think UKMO solution will suit UK cold prospects best.

Zonal winds down to -5m/s on the mean now: Reckon we probably need to see one of those bad boys plunging -20m/s or so, if a split like on GFS 12z is to verify. The SSW could still be a game changer, if we get a really favourable split, at the very least would take us into extra time, meaning that the exceptional synoptics needed to deliver in March, might actually be on the table for once…