JoeP wrote:This is exactly what I was getting at in one of my comments in these threads. The complexity of the system is such that the models we have are insufficient. Interdependence between sets of variables and feedback strengths are in many ways unknown.
Years ago I did some work writing genetic algorithms that searched for good solutions out of huge solutions spaces involving dozens of variables. I also did some experiments in mapping optimized routing code -- again using GAs. This gave me some little perspective as to what a proper climate model would require...that kind of system at the scales required are probably not currently sufficient as of now IMO.
I too have used stochastic learning algorithms to find patterns in data (including GAs). There are all the issues of over-fitting etc. In fact if you look at the climate denier arguments that are clearly wrong about half are over-fitting to some imagined fictitious cycle, the other half are just not understanding the physics.
HOWEVER - climate models are not the same as stochastic models. They are dynamical.
What that means is that, at ;least in principle, they don't need to learn anything at all. They can be an accurate physical simulation will all parameters tied down to the physics from first principles or experimental observations on specific elements quite different from the overal gungy climate.
This is the thing that lots of critics don't understand.
Now, even with this, climate models are very complex, and it is worth checking that all parametrized relationships are properly validated by physics or observations independent of the global temperature.
This means, in the case of CO2, that we do not need to be able to model all the chaos in the climate (actually it is clear that models are getting quite good at doing this) but suppose there was unknown forcing X - would it stop models from giving accurate estimates of the effect of CO2?
No it would not. The forcing from CO2 can be tied down easily from theory and satellite IR spectrum observations which agree. The feedbacks, which are the difficult bit, are the same for CO2 as for anything else. So all we need is a known forcing - or several different known forcings - and we can over both historical and paleo temperature time sequences look to work out the value of ECS from the correlation between temperature and forcing.
That is just one parameter - climate sensitivity - from a lot of data. Even with massive noise its doable. Now actually the time sequences come on many different time scales, years, decades, centuries. To include all of these you need to know time constants which are complex, though partly constrained by possible physics mechanisms. So maybe 3 or 4 parameters including these.
It is very doable. The errors come from uncertainties in forcings, and uncertainties in global temperatures, and finite amount of data which means that internal variation and unknown forcings cannot be eliminated completely.
Since the warming has flat lined despite the predictions, obviously the models are too simplistic or tuned in such a way to reflect the experimenter's biases. I suppose changing the name from GW to Climate Change is the real answer to that. Spin. How transparent is that?
The projections are not predictive over more than 6 months. You get a lot of internal variation. In teh 1990s temperatures were going up faster than models expected. Recently they have been going up slower (but not flatlining unless you cherry-pick both end-points and temperature datasets).
So every year with lower temperature is some evidence for lower ECS, but it is folded in with lots of years and with estimates of climate sensitivity independent of the direct one as above. And the last 17 years, when looked at with the overall temperature graph and the model run variability, look expected. Finally, we can explain between 50% and 100% of the last 17 years relative slowdown on known internal variation that is not predictaively modelled (it is averaged in the models):
(1) variable TSI
(2) variable ENSO phase.
There very well may be warming going into the future, but it is obvious there are problems with the predictors -- a little bit of humility and less hubris would suit -- on all sides of the issue, but mainly the so called 97%-ers. LOL.
I think you are getting caught up on the politics. the science is saying that AGW exists but the magnitide of it is variable over a 3:1 range and not known. The political campaign against this is full of misinformation and obviously wrong science. I sympathise with the climate scientists who become defensive amd political themselves given such a campaign. Personally I would hate it.
I think the scientists who get politicised are wrong. I think the scientists who argue that presenting the science in its true uncertainty publicly will not work are right. Politicians seem to have to use spin - special ways of saying things - to win elections. Personally I hate that and won't do it. I think no scientist should do it. But I can understand why some do because people are just not good at making decisions based on probabilty PDFs where the systematic uncertainty in the problem introduces meta-PDFs!
But on the internet the lies and misrepresentations from the denier side far far outweigh the political spin from the climate scientists. The IPCC reports do a good job of summarising the science and while I can cavil at the edges overall I think they have got both the expected climate sensitivity and the uncertainties about right.
From what people say here I'm not convinced they even know what the IPCC reports say about the uncertainties.
The IPCC summary for policy makers I would personally ignore. It is vetted by governments which means it will have spin.
Even if there is more warming, the side effects and the damage of abandoning current energy sources will be large. I have yet to see a serious cost-benefit done on the side of those that want severe limitations. This without the hysteria and political "Green Marxism" that the GW movement seems to have morphed into.
The best I have seen is a report commissioned by the UK government. It is serious cost/benefit - but as with all such things there are uncertainties and someone wanting to argue could probably get different answers.
http://en.wikipedia.org/wiki/Stern_Review
My (uninformed) opinion so far is that the cost of measures taken thus far, though real, is very small as fraction of GDP especially because taxes on energy to fund carbon neutral energy only cost the economy for the distortion they introduce, which will always be less than the total tax/subsidy - money spent on renewables circulates round the economy. They have had the desired effect of massively boosting solar - which will soon be competitive (maybe already is) for non-base-load applications. I'm less sure about whether other renewables are going to be competitive.
Electric cars are more difficult with another 15 years at least of technological innovation before being competitive - but arguably we want them anyway when you factor in the costs of PM2 pollution - that is very difficult to get rid of from IC engines. One serendipity is that intelligent charging of electric cars can provide distributed energy storage if car ownership and renewable power get ramped up together.
And, of course, sensible people are pro-nuclear for base load - modern nuclear power stations - even conventional ones - are pretty good. Alas it does not look like public opinion goes that way anywhere except France and China.