Mike Poole

An idea of the uncertainty on snow for Sunday and Monday from the GEFS probability plots on Meteociel: Probability of snow 24hrs of Sunday: Probability of snow 24hrs of Monday:

For me, the most interesting new tool in our model analysing armoury this year are these Icelandic ECM cluster plots, as often posted (and well explained) by MWB. Although the pictures for each cluster cover rather small area of the globe, at the risk of stating the obvious, more information is provided by the colour of the border. These correlate to the four broad patterns at the bottom of the graphic. The first two are obvious – the others needed Google translate: Blue: +NAO Green: –NAO Red: Fyrirstada translates as ‘obstruction’ – I’m guessing ‘Scandinavian Blocking’? Mauve: Hryggur initially translates as ‘sorry’! But if you get the other meetings, includes ‘spine’ and ridge’ – ‘Atlantic ridge’? The percentages next to these, vetrarmisseri = Winter Term, likur = probability, the middle word doesn’t translate – I guess these are the seasonal likelihoods from climatology.

Worse, many people throw in the word "zonality" when it doesn't even exist.

Who needs model output? Dr. Cohen’s latest AO blog is hilarious. In it he presents the winter forecasts from various models (NMME, CFS, ECMWF and UKMO), which are all screaming mild – junks the lot and in his AER forecast goes for cold right across Eurasia, to quote: ‘multiple weeks of severe winter weather across the NH’. It’s quite a call – either he’s wrong or the models are. Well worth a read after the rather mixed model output this morning: http://www.aer.com/science-research/climate-weather/arctic-oscillation

At this stage I am optimistic that we may get a more interesting winter this year than the last four. I reason this as follows. Although the long range models should be taken largely with a pinch of salt this far out, there is some good news in that whereas they were pretty much screaming +NAO full stop a couple of months ago, CFS is now starting to move towards some high-latitude blocking. Will be interested to see if GloSea5 does so too when the October runs are out. I also think that with the easterly QBO and reducing solar activity, it is unlikely we’ll go the full three months without some blocked periods. Once concern I have is with the warm temperature anomaly that exists pretty much across the globe at the moment, will half decent synoptic patterns which would normally deliver snow, lead only to marginal 850 temperatures? A concern for the south of England anyway.

Well, this looks extraordinary for the UK. We appear to be off the scale into 'black hole' territory on the GFS 06z CAPE chart. Pretty close to my location too. Suspect nothing will happen but you never know.

Something about the chart below bothers me. It is the tercile probability plots from GloSea5 for the summer months for 500 hPa heights: I accept that these probabilities are derived from a large number of model runs and averaged over a 3 month period, but without being able to see the individual evolutions, I find the plot is hard to interpret. It seems to depict a rather extreme position, with higher probability of higher than average heights over pretty much the entire globe. Does anyone know what might be causing this – is the averaging process masking everything but significant seasonal drivers? – is this rather extreme-looking position a hangover from last year’s El Nino, for example? Finally, a few posters have expressed concern that we may be headed for a 2007/12 type summer – the plots aren’t available for 2007, but for 2012 the corresponding plot from GloSea4 (as I think it was in those days) is as follows: It couldn’t look more different to this year’s. GloSea5 clearly sees this summer very differently to how it saw 2012 at the same time 5 years ago, so I see a repeat of 2012 as rather unlikely.

Things continuing to fall in place for a decent summer - here are the May GloSea5 predictions for the summer months from the Met Office (2m Temp and 500 hPa height):

For the first time in perhaps a decade, I’m looking forward to the summer weather with a sense of optimism. It has seemed that since 2007, for some reason, the dice have been rather loaded against prolonged hot spells in summer in the UK – that’s not to say there haven’t been any (July 2013 springs to mind) – it’s just that they have been less frequent than previously. This year, the Met Office contingency planners forecasts have been suggesting warmer than average, the long range models that I have looked at also suggesting high pressure more likely. In fact, high pressure close to the UK seems to have been a factor in the weather for much of many months now, and while Sod’s law would suggest this will change in the run up to summer but there’s nothing scientific about Sods law! So my money’s on a hotter than average one, and if that transpires I think I’m about 60% hot, humid and thundery, 40% hot and dry. Time will, as ever, tell.

Thanks, that question being my 16th post seems to have solved the problem then, as if by magic I now seem to have permission.

Hi folks, I don't seem to be able to post in the 'serious discussion' section, although I can most everywhere else, is there any reason for this? I'm only an occasional poster but wanted to reply to something posted about the election and found I couldn't. Thanks Mike

Unfortunately the Met Office only quote the probabilities for the categories for the Dec-Feb period in their Contingency Planners update, which are fairly average masking a forecast cold start and mild end to winter. So I overlayed a grid of squares to calculate the probabilities for the December only probability distribution graph on Figure 2T on the link Matt quotes (presumably this information comes from GloSea5 runs primarily), and got the following: Warmest category - 7% Second-warmest category - 11% Middle category - 16% Second-coldest category - 22% Coldest category - 44% As someone who uses probabilistic modeling in another (non-weather context), this looks to me a very strong signal compared to climatology (each category 20%) for December.

Yes, Here it is (for France - they didn't do UK charts in those days). http://www.meteociel.fr/modeles/archives-gfs/28/6/2012/0/cape-et-lifted-index/3h.htm

My take on Summer '15 from Oxfordshire. Plus point - it was largely dry - I organise a big rounders competition over the summer months and we've had little in the way of problems with the rain. But other than that - where was the heat???? Few hot days around 1 July and another on 22 August - apart from that it might have been late spring or early autumn. Still, I was staying in Bournemouth the night of THAT storm!!! Oh well, roll on winter...

I'm waiting for: Atlantic Storm Nathan - ramped as Storm of the Century by the Daily Express, making landfall as a slight breeze and a bit of drizzle.

A word on the model verification statistics, here's the 5 day graph: There's clarification on what the PR... models are: PRX: Experimental T1534 Semi-Lagrangian GFS run by NCEP NCO PRHW14: Experimental T1534 Semi-Lagrangian GFS run by NCEP EMC New GFS run in two different places maybe? I'm rather dissappointed if so, I've had my eye on the PRHW14 most of the year and it's a bit better than GFS, close on UKMO but still short of ECM - if this is now the highest resolution model, it is a bit underwhelming.

Steve, thanks for the forecast, and all the analysis in the previous thread. It's well reasoned and I hope that in 5 years time long range forecasts of this nature are the norm. Good luck, love to see some snow in January!

This is an utterly compelling thread! The OPI evolution is simultaneously both fascinating and maddening! I've been giving some thought to the wild swings we've seen - on Friday and again today. Some thoughts: I'm taking as read that the OPI is as stated in Riccardo's post i.e. that today's value is average of 12 days actual data + 10 days GFS forecast data. The swing today from the 06Z to the 12Z was from -0.64 to -3.15. Given the GFS can forecast pretty well up to 4 days out, the change must largely come from 6 days out of 22. What could explain this? I think the answer must be that this is a highly non-linear index - the translation by Interitus a few posts above seems to suggest this regarding the angular position of this axis parameter. So I conclude that a few days contribute massively to the negative values we see and most others much less so. Given at day 12/22 we are still seeing massive swings suggests that the values south of -3 have probably arisen mostly from the forecast part rather than the 'banked part' of the index, and the forecast ones from earlier in the month didn't verify. On that basis I'd guess we'd finish up about -1.5. Which is nice.

I presume its based on GFS because the data is freely available. Whether to use an ensemble mean is more interesting - its the mean of a large number of entirely physical realisations, but the mean itself is not physically reasonable - it could not have occurred in nature - so I don't think it would be useful for this OPI index which looks to be highly dependent on detail.

PRHW14 is the new upgrade to GFS which is currently being tested, I believe - has been for a few months. I've been keeping my eye on the stats recently and while a couple of months ago it seemed to have similar performance to GFS, it seems to have upped it's game recently. The ECM's poor recent day 10 verification stats were commented on in here over the last couple of days, but it's rather gone under the radar that at day 10 PRHW14 has performed best by some distance recently for the northern hemisphere:

Interesting, so to be clear the upgrade to the GFS this year is not the FIM? I understand that on the model verification charts the GFS upgrade is denoted as PRHW14 (see below) is this the FIM or a more mundane upgrade to GFS?

Relatively new to Netweather and my first post on the MOD. Why? I've noticed a new model on Meteociel that looks as though it will be worth keeping an eye on, the FIM. Here's its take on Sunday: Here's what Wikipedia says about this model (quote): The Flow-following, finite-volume Icosahedral Model (FIM) is a numerical weather prediction model currently under development at the Earth System Research Laboratory. The FIM is being developed as a candidate to eventually supplant the Global Forecast System, the United States's current medium-range forecast model. The FIM was originally slated to become operational some time in 2014 (as of spring 2014 the model is still testing); the GFS will continue to be run and maintained for several years afterward, much in the same way the GFS and its predecessor, the Nested Grid Model, ran concurrently for several years. The model currently produces similar results to the GFS, but runs slower on the NWS's operational computers. Its three-part name derives from its key features: "flow-following" indicates that its vertical coordinates are based on both terrain and potential temperature (isentropic sigma coordinates, previously used in the now-discontinued rapid update cycle model), and "finite-volume" describes the method used for calculating horizontal transport. The "icosahedral" portion describes the model's most uncommon feature: whereas most grid-based forecast models have historically used rectangular grid points (a less than ideal arrangement for a planet that is a slightly oblate spheroid), the FIM instead fits Earth to a truncated icosahedron, with twelve evenly spaced pentagons (including two at the poles) anchoring a grid of hexagons. The FIM runs as a multiscale model, with a suffix number indicating the model's horizontal resolution. FIM7 operates at a spatial resolution of approximately 60 km, FIM8 at 30 km, FIM9 at 15km and FIM9.5 at 10km. The FIM7 and FIM8 both run twice daily (0z and 12z) with 6-hour temporal resolution out to 14 days. The FIM9 runs four times daily, also with 6-hour steps, out to 7 days. (FIM9.5 is not currently in operation.) Unquote! I think it will be interesting to see how this model performs on two levels as a possible successor to GFS. First, does the novel geometry of the model overcome any of GFS biases, second this model seems to run 14 days but without the drop in resolution at 8 days - will this be better at picking up trends than GFS? Time will tell no doubt.

I've always liked snow and hot weather, but what caused a keen interest in the weather was for many years I organised the fixtures for a large rounders competition on a campus in Oxfordshire - and in the summer months would try to predict what the weather was going to do for rearranging games etc when it rained. I gradually became more and more interested - I am a physicist and am generally interested in areas of science that are complex and with some depth - there's plenty going on with the weather to maintain the interest. Have learned a lot from posts on this forum - please keep them coming!

This is a good one for me, I think in Oxfordshire this was the moment we changed from having no snow, virtually ever, to the way the weather has been in the last few years. (new poster, so hope this pic appears!!) This channel low dumped Wantage with 5 inches, and our office was closed - trouble was we had an international meeting - one guy coming from Germany! So I braved the 7 mile journey (and abandoned my car) German guy having no problems with his journey naturally...health and safety meant we had to leave the office and adjourn to the pub! We're really not cut out for it, we british are we? And it snowed another 4 inches the next night!

Thanks ABNS! Now just tying to work out how to correct my dodgy maths - but maybe this year is going to be a 110% winter! I thought it was possible to edit posts though! Sussed it now!!