“There is this mismatch between what the climate models are producing and what the observations are showing...We can’t ignore it.”
Susan Solomon, climatologist at the Massachusetts Institute of Technology in Cambridge
This statement nicely encapsulates the problem. The modeling isn't ab initio, simplifying assumptions go into the model-the nonlinear fluid dynamic partial differential equations aren't being solved overall of 3-space and time in these calculations. Any valid mathematical model must a) show agreement with known present and past data and b) have predictive ability. If a) isn't met any attempt at b) is meaningless. If a) is met, it does not immediately imply that any predictions will be accurate. Airflow over a wing is also a nonlinear fluid dynamic problem that is extensively modeled, yet only a true fool would climb into an air frame designed on modeling alone without any wind tunnel validation of the model.
It may also be true that a dynamic model shows short term agreement with data but then diverges. The time evolution of a dynamic system may move outside its domain of validity, even of the solution is mathematically convergent. Classic examples of this are what happened with the modeling of the financial markets circa 2008 or with the predictions made of nuclear explosions when the moratorium on testing was eventually lifted and real testing began again. Nonlinear systems are notorious for such behavior even if you could manage to convince yourself that you very accurately accounted for all variables.
To nafod's point, a zeroth order evaluation would suggest that *qualitatively* one reasonably expects that man-made pollutants have a deleterious effect on the climate, but it's the quantitative part that is suspect-how bad, how quickly, and what are the relative contributions of the signal (man-made effects) and the noise (all other effects influencing the climate). The hardest part of that is understanding the cross terms; how does each effect the other. Simple in a linear system by definition, not so nonlinear systems.
Further exacerbating the problem to my mind is that I have no idea what exactly a climate scientist is. Checking the websites of some universities it's not clear what you have to study and master to be a climate scientist. To be sure, the modeling portion of it is more applied mathematics, computer science and physics, than cloud formations and rain patterns. Like any scientific discipline, only a rather small subset of the field are people who have the mathematical wherewithal to do the modeling and, more importantly, understand it and know its limitations.
While agreement is nice, my favorite whipping boys, social scientists, agree on all sorts of stuff, even when proven hopelessly wrong. I'm loath to lump climate scientists into this category as there is some very good science here, and it's damn tough to do reductionist experiments and data collection in the actual environment with all the obfuscating influences. Clearly there are a few hucksters that undermine the whole enterprise and muddy the waters of what's known, what's suspected, and what's still out to the jury. The upshot is that those who make bold assertions based on a modeled prediction that proves to be wrong but then engage in legal action to defend themselves need to embrace the scientific method. Go back and understand what went wrong, refine your model, and iterate until it's right. Failing to do so leads to threads like this one where the arguments don't converge, to include people who fail to make the important distinction between weather and climate.
