Jump to content
Will it snow? 7 day forecast Weather Radar Will it be cold?
Snow?
Local
Radar
Cold?

No Balls Like Snow Balls

Members
 View Profile  See their activity

  • Posts

    2,978

  • Joined

  • Last visited

  • Days Won

    3

 Content Type 

Posts posted by No Balls Like Snow Balls

    Winter Model Discussion - Cold Hunting - 20th Dec

    in Spring Weather Discussion

    Posted · Edited by No balls like Snow Balls

    Blimey, that really is quite a statement nea.gif

    With greater understanding of the physics involved - notwithstanding the ever-increasing power of super-computing - the 5-day forecast today, is as accurate as the 1-day forecast, 20 years ago.

    That's with the same 'weather' - it's not some new, easier-to-read weather.

    So science has evolved; our understanding has evolved, and as has the technology at our disposal. There are some processes and phenomena which are still in the infancy of understanding, and as this improves further with more research, it's these advances which correlate through to greater accuracy.

    Personally, I think some people expect too much; viewing these charts out in the 5-10 day range, wonder why they hardly ever verify, then bash the models. Laughably, some even look beyond 10 days. The verification scatter plots for all models post 5/6 days are hilarious. The goal for this decade, will be to extend that confidence out to the 7 to 10-day range - at least.

    Look at the Japanese. This year they on-boarded the worlds' most powerful meteorological super-computer: 847 TFlops! That'll deliver to the JMA capability for higher-res, more layers, increased range and greater scope to calculate new variables - all resulting in more accurate, detailed forecasting. That power doesn't come cheap though: $1Bn investment by JMA, hence the respect Exeter has for them. I've actually heard that UKMO and JMA will be working together to develop a brand new experimental platform by 2014, with a view to roll-out onto the UKMO's newest super-computer in 2017. I think it's going to be something along the lines of a massively scaled-up, continental UKV, with multi-member ensemble support. That's what I've heard anyway.

    The demand and importance for accurate weather and climate forecasts has never been higher; not merely to provide a good public service, but because so much of business - the success and failure - rests on solid accuracy. The Met Office make a tidy sum from licensing their UKMO base platform to meteorological agencies worldwide, and they also supply bespoke services to private companies and institutions. Beyond that though, or more - beyond borders - there's been a suggestion that worldwide meteorological super-computing could be harnessed, therefore delivering a game-changing array of forecasting suites. I very much doubt that will happen any time soon though, for a number of reasons - even though the benefits are admittedly exciting.

    A lot of people knock The Met Office, but the proof is in the pudding: in this tough financial climate, 'the Brits' are the preeminent world leaders in meteorology. Other meteorological agencies procure their services, and - if you read as many papers as I do - you'll see that they're frequently referenced, when it comes to peer review and due respect. They're one government agency that the British people can be thoroughly proud of, and I can't imagine how many thousands of lives their work must've saved across the world.

    The weather has and always will do something unexpected and go against all the rules, a computer model will never be able to expect the un-expected, there is nothing greater than mother nature. There has been numerous occasions where a forecast 24 hours prior to an event has been wrong, the weather "owns the models" not the other way round so to speak. I stand by my statement.

    Winter Model Discussion - Cold Hunting - 20th Dec

    in Spring Weather Discussion

    Posted · Edited by No balls like Snow Balls

    So, it's not what the EC32 is showing today for January, but - as we've seen this season - there's just as solid a chance that 'what the EC32 is showing for January' will change again next week.

    I quite understand Matt posting updates, but - ironically - he's just passing evidence over, which suggests it's a model well worth applying a heavy degree of caution to. It simply isn't performing well enough to be afforded such respect, or referenced to so frequently.

    The fact of the matter, is that no model has been that impressive this season; all have made errors - some more serious than others; all continue to propose different ideas and all continue to make relatively large calculation corrections. So, it's a word of caution to all models here: not just the EC32.

    So why is this happening? Well, from my understanding of mathematics, you'll develop a logical algorithm which - in the broad sense - will be governed by several different scientific principles, which are then converted into mathematical equations (the calculations themselves) and then the boundaries (the thresholds) So, in the main, this will be your base (and probably is for the GM)

    To localise the model, you then also need to programme prevailing patterns, which would then give a degree of artificial guidance to calculations. This would probably be several layers which adjusts the boundary fields, therefore giving calculations some sense of 'where to go next'.

    You then collate data, throw it into the model and see what Aladdin's lamp does with it.

    I think the prime suspect for 'error', would be where an algorithm is given 'intelligent data' which artificially influences what it would otherwise mathematically wish to do. But you do need this, otherwise chaos theory would - by definition - ruin all of your hard work. So, it is a necessary evil.

    However, if you think about it, some of that 'intelligent data' - historical tendencies; recent, mid, long-term, etc - isn't exactly set in stone; just because, for instance, UK winters have been relatively mild if you analyse the last 25yrs or so, doesn't ergo mean that pattern is fixed and will continue. But there will be an element of a false bias towards this being so, and - in statistical terms - I would totally agree with having that.

    Personally (but what do I know!?) I do think our winters are changing. I think they're becoming colder - in relative terms you'll appreciate, and milder patterns (equally relative) look less commonplace. Aside from good old cyclical change, there are so many reasons as to why that might be happening; but I wonder how quickly a NWP model with bias towards a 'milder' evolution can react to that?

    I honestly don't know the answer to that question, but I can hypothesise that - without human correction - such bias would continue, even in a pattern change, as that's what the algorithm is programmed to do. Perhaps you can overlay a further 'bias correction' layer, in order to realign? Immediately though, I think doing that would increase risk, as you don't really know how strong +/- to modify; you can't, it would be a matter of subjectivity, and necessarily that scientific.

    I've a good friend who works down in Exeter, and apparently they'd like me to do some work for them; just bits and bobs which they just don't have the resource to research - but more from a climatological sense. So this opportunity should - at the very least - allow me greater insight, and I'm really looking forward to it. I start next July, so unfortunately you'll have to listen to prattle on 'till then laugh.png

    Anyway, those are my musings on NWP model performance; if they add something great - if they don't, I apologise! Such exciting things I talk about crazy.gif

    I would say quite simply that the error is caused by erm!!, wait for it........... THE WEATHER BEING THE WEATHER good.gif

    The weather will always be one step ahead of the models and that will never change no matter how much science evolves.

    METO UK Further Outlook 16 Day To 30 Forecast

    in Spring Weather Discussion

    Posted

    This is what the UKMO were predicting on 6th December for the period we are just entering, just about the opposite of what we are getting and about to get I would say. So hopefully their current predictions will be wrong too! I know it doesn't always work like that but you can but hope....

    UK Outlook for Friday 21 Dec 2012 to Friday 4 Jan 2013:

    Temperatures by both day and night are most likely to be below average during this period, with a greater incidence of overnight frost and, at times, fog or freezing fog. The signal for rain and snowfall amounts during this period shows a high degree of uncertainty, however, on balance, slightly drier conditions compared to average are favoured for most of the UK. Even though there is a drier-than-average signal for much of the UK, where precipitation does occur, there is a greater chance that it will fall as snow compared to normal.

    Updated: 1136 on Thu 6 Dec 2012

    I wish

    Winter Model Discussion - The Hunt For Cold -18z 17/12/12

    in Spring Weather Discussion

    Posted

    It's a surface high; Greenland is so cold that whenever there is not a low pressure system moving across, pressure builds rapidly at the surface due to the dense cold air. Any low pressure apporaching that high will break through easily. You need to see more green and orange on that chart which would indicate a high pressure with more depth which would be able to withstand the approach of a low pressure.

    Let's hope for some yellows and oranges then :-)

    Winter Model Discussion - The Hunt For Cold -18z 17/12/12

    in Spring Weather Discussion

    Posted

    I find the De Bilt ensembles pretty useless to be honest when De Bilt is at least 200 miles East of the most Easterly point in the UK.

    As we all know we are quite often the last stopping point for cold air and even more so we are in the middle of the battle of the mild atlantic air and cold European air, and it's quite apparent that De Bilt is more often than not likely to be in the cold european air, while us in the UK miss out.

    Winter Model Discussion - The Hunt For Cold -18z 17/12/12

    in Spring Weather Discussion

    Posted · Edited by No balls like Snow Balls

    Exactly, the pattern change over the Xmas period has changed alot. The upgrades for the reliable are good the AH is moving south as you say to allow the low pressure to move east. I wouldnt be suprised by the time the 18z roles out there will be a few suprises for Xmas day!

    Do you mean we will find out it wasn't santa drinking the sherry and eating the mince pies after all cray.gif

×
×
  • Create New...