It Has A Gigantic Supercomputer, 1,500 Staff And A £170M-A-Year Budget

[Mark Bayley]

The problem with the forecasts is there to vague. Why can't they do a link to a separate page showing the science behind the forecast with some detailed explanations. I find it odd how GP (and others) can predict this winter pretty much spot on with data that we the public have access to yet the met office have all these super computers and can't! Perhaps they shouldn't rely to much on there computers? Perhaps the super computer was a warming bias built into it? Who knows. It certainly needs improvements! As already mentioned the forecasts are to vague. I'm not expecting paragraphs however a bit more detail then 35%below Average, 65% Above average or 1/7 chance of a colder than average winter etc... would be nicer. The short range forecasts are very good! Lets just see if they can improve the long range ones

[gmoran]

Are you saying that it's IMPOSSIBLE to predict changes in global temperature a priori?

Firstly, what we are talking about are skilful predictions; if I predict that the sun will still be functioning next year, then I'd probably be right, but it isn't particularly skilful.

From a pure commonsense perspective, and a scientific one as well, it is obvious that future climate states cannot be skilfully predicted: climate system is non-linear; climate patterns (PDO, AMO, ENSO, AO) are poorly understood; future solar irradiance cannot be predicted; future volcanic stratospheric eruptions cannot be predicted - hence future climate cannot be predicted. Climate modellers are not attempting to predict future climate, but future boundary conditions given changes in forcing. Therefore yes, it is impossible to predict future climate, including GMST, using our knowledge of the climate system.

It is however possible to make naive predictions, e.g. the long term warming trend is .7'C per century, and it would be a reasonable bet that this would continue (however it is far from certain). As an example, if at the start of the 21C you were to predict the rate of change for the next decade using the naive method (.07'/D), you would have done far better than the GCM method (.2'/D), the actual change (Hadcrut3 dataset for the 120 months to Nov 2009) was .04'/D.

I agree with you, that 'local' climate is impossible to predict, but global temperature-prediction should be a lot easier to compute...That said, I'm not claiming that any particular prediction is right (or wrong). I can't possibly know that...

Don't you think this is an interesting point? If you take the synoptic definition of climate, then any prediction of future climate must involve skilful predictions of regional climate. Yet you think this is impossible to predict (a concept that this paper agrees with Koutsoyiannis et al 2008)? Why do you think predicting future global climate, using a synoptic model or a GCM, should be any easier?

[Mondy]

I'll refer you to climate scientist and modeller Gavin Schmidt of NASA GISS: "there is so much unforced variability in the system which we can’t predict — the chaotic component of the climate system — which is not predictable beyond two weeks, even theoretically.", interview available here: Gavin Schmidt Interview. This is what I mean when I say it is physically impossible to predict future climate; and is subtly different to whether GCM's can skilfully project future boundary conditions in response to a forcing - though so far they do not appear to be skilfull in even this regard.

http://blogs.telegraph.co.uk/news/jamesdelingpole/100022474/climategate-goes-american-noaa-giss-and-the-mystery-of-the-vanishing-weather-stations/

[Methuselah]

Don't you think this is an interesting point? If you take the synoptic definition of climate, then any prediction of future climate must involve skilful predictions of regional climate. Yet you think this is impossible to predict (a concept that this paper agrees with Koutsoyiannis et al 2008)? Why do you think predicting future global climate, using a synoptic model or a GCM, should be any easier?

Good morning mate...

Very interesting points you raise, there. I think I agree absolutely:

If I turn-on my heater, I can predict (quite accurately) what my room-temperatue will level-out at, based on very simple mathematics - if were nerdy enough?? But, what I can't do, is know where each individual molecule of gas will go; I'd have to have nearly-infinite information, not to mention computing power!

I guess that all I'm trying to say is that the heat-budget equations required to predict a degree of global temperature-change are far simpler than those (chaotic?) ones needed to forecast weather...Somewhere in between (I think?) lies the mathematics required to predict climate? At the moment, however, models are a longway short of the complexity required to be accurate...

Sorry for the waffle. It's at least 15 years since I last tried to get my head around any of this stuff! :D