[Arp220]

[ QUOTE ]
[ QUOTE ]
I don't know the actual numbers, they're not conveniently published in a summary table anywhere. But click the link. Much of the reasoning and understanding and quantification of the effects on different sectors is solid, and reports have been done on this very thing, including by relatively disinterested parties such as insurance companies and the world bank There's even a pretty graph with errors bars for you.

[/ QUOTE ]


Phil and wacki,

Note that I have not been asking for the complete analysis and how the various parts of the equation were derived. In fact I'm not even asking for any number except the degree of certainty as to the accuracy of the models. And as I said, I don't follow this closely so I haven't read and am not going to read reams of studies and data. However since you two and others are speaking authoritatively on same, surely one of you *should* be able to answer my question.

I mean how can any layperson who is being told something by scientists along with recommended actions, not ask for a degree of certainty number as to the scientific models that led to the predictions.

If none of you can answer this question, or refuse to knowing the answer, then I find that very telling.




Doctor: I did some tests and although they aren't totally reliable, they indicate you *might* have something serious. I recommend an expensive operation not covered by insurance followed by a long and painful recovery.

Patient: What do you mean by "not totally reliable/accurate"?

Doctor: I can't quantify it for you but I can refer you to several medical journal articles on the methodology of those tests and statistical analysis of the data.

Patient: Well if you've read all of them and are using that information as the basis of your diagnosis, why can't you just tell me how accurate those testing methods are?

Doctor: Just trust me you stupid layperson! Get that operation!

Patient: I think I'll just go either look for a second opinion or check out alternative therapies.

[/ QUOTE ]

I'll take a shot at it, though be warned I have a hangover and am about to go to bed.

Your question is not a straightforward one. The way the uncertainty is expressed in (say) the IPCC report is as a simple error bar on the predicted global average temperature rise. I cant remember the actual figure (2 degrees or so over 100 years?), so lets call it X degrees plus or minus Y degrees.

Here, the error is given as just one number, Y, but this is actually a compound of many MANY sources of error, some random, and some systematic. It can be split into multiple parts, where some come from climate models, some come from actual measurements of CO2 levels at monitoring stations, some come from the atmospheric composition 'baseline' that is established from historical records such as ice cores... and so on.

Y is typically given as a an error bar of a given number of 'sigma'. A one sigma error bar denotes the 68% confidence region, a two sigma error bar the 90% confidence region, 3 sigma is 95%, and so on. I may not have these figures exactly right, but they serve to illustrate the idea.

So, for example, if the IPCC report says the predicted temperature change over 100 years is +3 degrees plus or minus one degree at one sigma confidence, then they are saying that, based on all sources of error in models, data, etc etc, they are 68% certain the change is between +2 and +4 degrees, 90% certain its between +1 and +5 degrees, 95% certain its between 0 and +6 degrees. These confidence intervals include the sources of error, and uncertainties, arising from the known simplifications in the models, and can be thought of as a statement of how reliable the models are, though only a percentage of the magnitude of the error bar comes from the climate models.