Met Office to build £97m supercomputer

[knocker]

 

Funding has been confirmed for a £97m supercomputer to improve the Met Office's weather forecasting and climate modelling.

 

The facility will work 13 times faster than the current system, enabling detailed, UK-wide forecast models with a resolution of 1.5km to be run every single hour, rather than every three.

 

It will be built in Exeter during 2015 and become operational next September.

 

The Met Office said it would deliver a "step change" in forecast accuracy.

 

http://www.bbc.co.uk/news/science-environment-29789208

 

http://investors.cray.com/phoenix.zhtml?c=98390&p=irol-newsArticle&ID=1982080

Edited October 28, 2014 by knocker

[Hocus Pocus]

How much did they spend recently on the other one from a few years back? It's an awful lot of  the taxpayers money which would be better spent elsewhere IMHO.

How much did they spend recently on the other one from a few years back? It's an awful lot of  the taxpayers money which would be better spent elsewhere IMHO.

I've should have read the full article first, so apologies for that.

[johnholmes]

I'm saying nowt!

[knocker]

I'm saying does this wasted money also apply to weather radars and the maintenance thereof?

[feb1991blizzard]

Would it not have been better to keep the regularity of runs the same but try to decrease the size of the grids by spending more money on manned weather stations thus increasing the accuracy and amount of the data fed into the computers.

[Neilsouth]

Improve it? HAH! They've been trying to improve it for years. 

[Turnedoutniceagain]

£97 million !!! You can build a hospital for that. What a colossal waste of taxpayers' money

[Summer Sun]

Sky news saying "it is set to give the UK its most accurate weather forecasts ever and can perform more than 16,000 trillion calculations per second"

 

"The first phase of the system will be operational in September 2015 and it will reach full capacity in 2017"

 

http://news.sky.com/story/1361615/huge-supercomputer-to-boost-weather-forecasts

[knocker]

Would it not have been better to keep the regularity of runs the same but try to decrease the size of the grids by spending more money on manned weather stations thus increasing the accuracy and amount of the data fed into the computers.

 

Manned weather stations are not more accurate than AWS. Back in the day the cost of manning and running weather stations was a huge chunk of the METO budget and these days is unnecessary. An upgrade is essential to vastly improve warnings to public services and to maintain their position as one of the leading and respected organisations in the world.

 

The other pluses are they will be more competitive vis a vis the 'professionals' on NW and the Yanks are eating their heart out.

 

 

NOAA left in the dust again. UK MetOffice joins ECMWF in buying monster, American made Cray Supercomputers

Edited October 28, 2014 by knocker

[feb1991blizzard]

^^

 

Its no good having models that are the same grid size but run every hour, those snow / freezing rain events in 2012 surely taught people that, better to reduce the grid size else they will still be wrong but this time wrong every hour should a similar event happen again.

[Mapantz]

I'm all for it

[Dorsetbred]

480,000 processors, nice.... lets hope their code is good for that and scales well.

[carinthian]

Improve it? HAH! They've been trying to improve it for years. 

I remember being shown around the new super computers installed in Bracknell some 40 year ago. Could not really understand what it was all about then, only Dr Mason knew ! Yes, as the Met Office say another step. Still when I look back at all those excellent weather forecasters and observers I worked with, who poured over plotted charts, upper air data, surface weather plots , land and out in the Atlantic, they were never far off the mark with predictions. So what I am saying , there is not a massive advance in forecasting with the increased computer technology over the skilled eye for detail from doing it the old way. Long term weather forecasting seems to be the spur for investment but are the results satisfactory in relation to this cost ?

 C

[Formula_1_Fan]

I don't think 97 million is too bad when you compare it to the ridiculous amount of money being spent in say football or something these days. I know that's not tax payers money but ultimately it is the public who have to pay with increased ticket prices etc.

I mean 97 million is probably Gareth Bale plus his wages for a year, if it's moving science forward it can only be a good thing.

[feb1991blizzard]

I don't think 97 million is too bad when you compare it to the ridiculous amount of money being spent in say football or something these days. I know that's not tax payers money but ultimately it is the public who have to pay with increased ticket prices etc.

I mean 97 million is probably Gareth Bale plus his wages for a year, if it's moving science forward it can only be a good thing.

 

What annoys me about it though is they then censor the data on the grounds that they are selling it to other companies, they should decide whether they want to be commercial and play by the same rules as everyone else like Netweather etc or they want to be public funded.

[Hocus Pocus]

I remember being shown around the new super computers installed in Bracknell some 40 year ago. Could not really understand what it was all about then, only Dr Mason knew ! Yes, as the Met Office say another step. Still when I look back at all those excellent weather forecasters and observers I worked with, who poured over plotted charts, upper air data, surface weather plots , land and out in the Atlantic, they were never far off the mark with predictions. So what I am saying , there is not a massive advance in forecasting with the increased computer technology over the skilled eye for detail from doing it the old way. Long term weather forecasting seems to be the spur for investment but are the results satisfactory in relation to this cost ?

 C

Sums up my views nicely carinthian. I see an over reliance on computers and not enough old school metrology where  human intuition and know how still surpasses technology. I think it's an awful amount to spend when the economy is already in dire straits, but that's another kettle of fish altogether. 

 

Will it be able to predict if it will snow IMBY though?

[knocker]

I remember being shown around the new super computers installed in Bracknell some 40 year ago. Could not really understand what it was all about then, only Dr Mason knew ! Yes, as the Met Office say another step. Still when I look back at all those excellent weather forecasters and observers I worked with, who poured over plotted charts, upper air data, surface weather plots , land and out in the Atlantic, they were never far off the mark with predictions. So what I am saying , there is not a massive advance in forecasting with the increased computer technology over the skilled eye for detail from doing it the old way. Long term weather forecasting seems to be the spur for investment but are the results satisfactory in relation to this cost ?

 C

 

I was working at Dunstable (then the headquarters) in 1959 and they had a Ferranti Mercury computer and it used punched tape and it had valves! That was when the great minds like Dr, Sawyer, Fred Bushby, et al were getting numerical forecasting off the ground.

 

Not to forget climate research which is hugely important.

Edited October 28, 2014 by knocker

[carinthian]

I was working at Dunstable (then the headquarters) in 1959 and they had a Ferranti Mercury computer and it used punched tape and it had valves! That was when the great minds like Dr, Sawyer, Fred Bushby, et al were getting numerical forecasting off the ground.

 

Not to forget climate research which is hugely important.

Knocker, you old boffin ! I was only in shorts in 1959, but remember it was a nice late summer. Would that be Basil Ferranti ?

C

[knocker]

What annoys me about it though is they then censor the data on the grounds that they are selling it to other companies, they should decide whether they want to be commercial and play by the same rules as everyone else like Netweather etc or they want to be public funded.

 

What data are they censoring? Perhaps they should censor the radar data as well.

Knocker, you old boffin ! I was only in shorts in 1959, but remember it was a nice late summer. Would that be Basil Ferranti ?

C

 

The whole summer was great and I think it was Basil Faulty.

[feb1991blizzard]

What data are they censoring? Perhaps they should censor the radar data as well.

 

The whole summer was great and I think it was Basil Faulty.

 

Where do you want me to start????   The mogreps, the 168 GM op, all ensembles, the UKV, the uk4 until recently!!!!

[knocker]

Where do you want me to start????   The mogreps, the 168 GM op, all ensembles, the UKV, the uk4 until recently!!!!

 

That isn't data they are products.  What i tend to find on here is that people who do all the shouting and denigrating the METO actually know zilch about the METO structure and financing  Companies like NW don't have the same responsibilties nor do they require the same infrastructure ( do they use their own radars?) so they are the ones who aren't playing on a level playing field.

 

Although the METO has had a commercial arm for many years now if you want a first class national met, office coupled with a world class research facility then it has to be financed much the same as satellites that everyone takes for granted.

Edited October 28, 2014 by knocker

[Yarmy]

It does seem a bit perverse that weather enthusiasts are decrying money being spent to improve weather forecasts.

 

As noted above, the MetO is now a commercial operation and it cannot afford to stand still if it wants to remain competitive.

[Hocus Pocus]

That isn't data they are products.  What i tend to find on here is that people who do all the shouting and denigrating the METO actually know zilch about the METO structure and financing  Companies like NW don't have the same responsibilties nor do they require the same infrastructure ( do they use their own radars?) so they are the ones who aren't playing on a level playing field.

 

Although the METO has had a commercial arm for many years now if you want a first class national met, office coupled with a world class research facility then it has to be financed much the same as satellites that everyone takes for granted.

Whilst their can be no arguments as to the MetO's credentials at being thee finest it does seem strange that only a couple of years ago they spent a fortune upgrading their computers. Maybe I'm old fashioned but I still believe as good as computers are there's no substitute for experience.

[johnholmes]

ah well, breaking my promise

This old chestnut about costs and not allowing everyone access to everything is not too dissimilar to the comments about the ODI. Folk will not carefully read how/what the ODI is about and what the team are trying to do. Likewise most on here are only interested in what they can get for their own back yard. Very few on here, perhaps 6 maybe 7 have the faintest idea what the Met O is allowed to do and not allowed to do. Most don't care, no problem there. But is it not a touch unfair to criticise when the facts have not been discovered by most of those chucking stones. Here I must admit I have not yet read the Met O press release about just what this latest Cray will do, grid lengths, number of layers, etc etc but I doubt many on here have really read it carefully. Do we want progress or not? If we do then trying to improve forecast not just for joe bloggs will it or not in my back garden, but for aviation, energy industry, commercial requirements of what to stock, how much and how long for, none of these users can get improvements without these huge sums. Read the Annual Reports of not just the Met office but cicil aviation and how much money they can save/lose if a 30-35000ft wind is +/- 10knots, or a predicted temperature 72-120 hours ahead is outside +/- 2C for Asda/Tesco, Gas/electric, the sums are enormous.

The Met O does two things at once, provides the MOD data for anywhere in the world which is largely the main omputer costs and then the forecasts for all the commercial systems, for which it must show a profit, then lastly, the media, for which it is paid, perhaps rightly, peanuts, be that data for Net Wx or the BBC.

When they do something wrong count on me to make a noise but it is fun for many to throw bricks with little knowledge of what is the reason for some of the very expensive decisions that governments of various political persuasions make about the Met Office.

Having broken my promise I will shut up.

[knocker]

Whilst their can be no arguments as to the MetO's credentials at being thee finest it does seem strange that only a couple of years ago they spent a fortune upgrading their computers. Maybe I'm old fashioned but I still believe as good as computers are there's no substitute for experience.

 

In the modern world of meteorology and climate then I think we will all agree that computer power is essential.

 

As to your second observation, yes you have a point. Not that many years ago the METO had weather centers littered around the country. (come in John all is forgiven) and there is no question that forecasters with years of experience and local knowledge were a distinct advantage. For example Plymouth Weather Center was ideal for my neck of the woods. But they were hugely expensive to operate and when commercial companies, operating from a tin shack in saffron Walden, started to spread their wings and the government demanded more for their money the METO couldn't compete.

 

This was part of the cycle of moving from one age to another. We used to have eight manned radiosonde stations but they cost £2m a year to run. Now we have two.

 

EDIT

I hadn't read John's post when I wrote this. Suffice it to say I totally agree.

Edited October 28, 2014 by knocker