Give the Met Office £10m, says Transport Committee The Transport Committee is backing calls to hand £10m to the Met Office to spend on hardware to improve its long-term forecasting ability. The Committee reported yesterday on ways to reduce the impact of severe weather on transport links. Last year's winter in the UK arrived early and snow-filled and came as quite a shock to …
And it still won't stop the problem. Weather is very difficult to predict in the long term, but even if the Met Office could give an accurate forecast more than 24 hours in advance you would still need to have the infrastructure to cope.
90% of the problem for Rail south of London had nothing to do with the lack of snow ploughs and the fact that snow settles on third rails, and everything to do with the fact that in BR days they would keep trains running all night to keep the track clear using normal drivers, now Network Rail has no train drivers of their own, and has to contract them in, or persuade the train companies to run trains for which they have no financial incentive as they can write off days when the trains can't run with no penalty.
With air and road, try and convince an accountant to pay for a snow plough that might be used once in ten years. BAA have least excuse as they probably own theirs which is more than local authorities do now, used to be that my local authority's Bedford trucks all had plough frames on the front, the outsources service provider certainly doesn't. It's all down to money and being prepared.
The so-called "free" market (the only actually .free markets I know of are illegal) is only any good for discretionary purchases where choice is meaningful. Basic infrastructure is what the state is for. I still can't believe the utilities & railways were privatised - what massive cockups/ripoffs they were
Medium-range, in this context, means up to 10 days. What the Met Office want to be able to do is forecast a 'barbecue summer' in March or a 'White Christmas' in September. Some of us might wish that they could accurately forecast the weather within the next 12 hours, but let that pass ...
Computers have gotten exponentially faster and cheaper over the past 10 years, and the MET office's predictions exponentially shitter.
So firing the old guard who thought about the weather and used intelligence and instinct, and moving to a reliance on unproven software models didn't work, so adding more GHZ will solve things?
The Met office is all about short term forecasts. They would be better off asking a firm which specializes in long range forecasts, like Weather Action. The Met's long range forecasts are shit, since they are based around their accurate short range models, where as Weather Action only do long range forecasts.
...you have to collect all the data required for long term forecasts to make short term forecasts. Seems a pity not to make the most of it.
I notice Weather Action's website includes a rant about global warning (not just questioning the extent to which human activity affects it, but rejecting the basic science involved.)
It also contains claims that they can predict earthquakes (try asking a relevantly qualified scientist about that one), a cherry picked selection of 'successful' forecasts, and references to crank websites eg: "Emergentwisdom.com" (which appears to rubbish ALL science indiscriminately!) and individual cranks, eg: Alex Jones of 'every fruitcake conspiracy theory you can think of' fame.
I would not be in a hurry to do business with them, but then *I* wouldn't let creationists run the Natural History Museum : )
Until Piers Corbyn publishes the details of his forecasting methods, and submits *all* his forecasts (not just a cherry-picked selection) for proper, independent, analysis, I'll not trust his forecasts any more than I'd trust some crazy guy reading the patterns in chicken entrails.
...is not surprising. The amateur weather message boards were predicting a marked cold snap too, based on the ECMWF charts and Met Office data. That chart shown on the "not a forecast" page is only one tiny piece of the jigsaw. The Met Office also use the ECM 30-day ensemble data heavily (which for weeks was pointing to an exceptionally cold spell - those of you who read the MetO 30-day forecasts would have seen it) as well as predictions for the NAO (based on SSTs in May). Low solar activity and the trend for blocking due to southerly jets also enhanced the chances of cold - all factors the Met Office would have used in their forecast to the Government.
Sadly a lot of people moan without knowing the facts.
Why they don't just do a nice SETI like distributed client that all the weather nuts in the World would happily run and get some nice forcasts at higher resolution modeling than they can do now. Then they would just need to send and recieve the work units out. Sure they can use there onsite computers as well to add to the workload, and I'm fairly sure that various govermental departments have overspecced desktops CPU wise that sit idle alot given 5 year buying cycles on IT hardware(worst case) and that even a 5year old CPU is overpowered for most word users (though windows does try to help out there :-).
Also look at carrying on allowing weather model data for free as long as it's not for profit, we all know a few weather forcast sites that just scrape there data from xxx govermental agency in xxx country and mash it up and sell it on, if there making money then surely they should contribute somewhat.
But as we stand currently I believe they run the models at a resolution of 1 mile, they could be reduced with more comute power.
Also one thing the met office needs to look at is the ability to source other sources of measurement/metrics and adjust, so even though the 100 home weather data sets don't furfill the metoffice layout standard and not used there data can be over a short time calabrated in a way that would accomodate there location quirks and indeed factor that in giving more data that is not used currently. One example would be a weather station ontop of a block of flats say. Now that incures urban heat radiation effects so it's effected by the enviroment to the stage that it's not used by the met office as it falls foul of some of there station location criteria (which in many ways makes sence). But if everybody in that area/block of flats also feels that effect then isn't that the weather they experience, we all know London is hotter becasue of that heat radiation effect (all that concrete etc) and that effects the wether so lets use that and not exclude that. One hand saying we effect the climate and on the other saying ew that data is effected by the enviromental human effect and not use it is a little bit crazy.
This is why I for one would rather see this money spent on a project to do software that distrabutes the load and allows statistical analysis/smoothing to allow more data to be utilised.
Feeding back actualities back into the model would easily allow it to adjust for general unsuitable weather stations and as such afford mroe data and a better grasp on whats happening giving better models, better forcasts, better for all.
Another thought for a laugh if I was to have a computer that generated as much heat as the sun then it could forcast saftly everytime it runs that it will be hot :D.
lets do some smart software and look at other sources of data and do this smart and not the usual here is a wedge of cash please spend on wedge of metal approach.
Spend the money on think not fail.
STOP THINK
NWP needs fast interconnect, hence the expensive MMP machines. Without each cell knowing what's going on at the boundaries of its neighbours it's not going to work too well. SETI on the otherhand doles out completely independent work packages, and there's hardly a time limit - we need the T+24 forecast quickly, but we can wait (a bit) for the little green men to be found.
SETI and weather forecasting are different problems
SETI is perfect for distributed computing, as it only analyses sets of data that don't affect each other. For weather forecasting, the 'cells' into which the simulation is divided have to communicate with each other really fast. Pings in the Internet typically vary from a few milliseconds to a good fraction of a second. In the insides of these supercomputer the equivalent concept -the time needed for data from any given cell to reach other cells- is measured in nanoseconds or microseconds.
My 0,02 €
They would have to either get better at predictions or go bust and let someone else have go.
Most days they would be better of with a dozen old gits with pine cones guessing what the weather will be and phoning it in.
Plus sack the rail management who sent the de-icing trains of for their annual service in November & December! Is it that hard to do it in the summer when you know they will not be needed!
I also have my suspicions that the real reason we had no snow plows out the last two years (in Kent), is because some H&S twat thinks it not safe for the drivers.
Although we are really talking about long term forecasting, it seems like a few people can't resist the opportunity to trot out some old cliches about the Met Office never getting short term forecasts right.
It seems to me that short term forecasts (2-3 days) are very accurate these days (compared with a few years ago). The other day we had a clear blue sky in the morning. The forecast said that there would be a short shower at 7pm, and there was. I think that is almost a form of magic.
Is cynicism compulsory on this site?
it's political. The Met Office have either been infiltrated by the MMGW brigade, or have taken a strategic decision that their existence needs to agree with MMGW. Therefore, their "forecasts" need to be in line with the doom and gloom the MMGW brigade want to see.
*If* MMGW were a given, this might be easy. However, as far as I am concerned, the juries out on MMGW (ask me again in a few millenia), and more importantly, the *weather* just isn't playing ball.
So the Met Office daren't forecast anything which isn't the "hottest summer since records began" or "the mildest winter since records began".
Alternatively people who have been looking at weather and climate have some idea of the science behind it, and have analysed the patterns and therefore have the expertise and experience to make the rational decision that AGW is real and significant. As compared to people working outside of their experience and expertise. That is, the original person who posted about "infiltration by the MMGW brigade".
Climate change relates to "years and decades out" forecasting (i.e. climate).
Now, how this relates to "a few months out" and better predictions for this sort of timeframe is hard to understand, outside of the original post just being a knee-jerk post from a climate change denier.
Perhaps they should forecast as a percentage of likely-hood. Like "there is a 70% chance of precipitation in your area this afternoon", or there is a "60% chance of a severe winter", then they have some wiggle room and can never be wrong.
That's how they do it in the US..... Just to stir up you haters.
This is the "severe winter forecast" the Met Office supplied to the Cabinet Office.
Mild -0.1 to +1.3 Probability 30%
Average -0.5 to +0.6 Probability 30%
Cold -1.5 to +0.4 Probability 40%
You'll notice that it only barely favours a cold winter, and that the categories overlap: 0C is considered "mild" and "average" and "cold".
http://www.bishop-hill.net/blog/2011/1/31/everyones-a-winner.html
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
