The limits of climate models

The formation of cloud is very complicated and depends upon the presence of small particles upon which the drops can form. But most clouds don’t rain, and rain is necessary to remove the moisture from the atmosphere. A cloud will start to rain when the drops coalesce to form bigger drops. But even when the drops have become large enough to fall out of the cloud, it is hard to quantitatively predict how much water will be lost from the atmosphere in as rain.

It may well be a fairly hopeless ambition to model these processes in any way but the crudest manner.

In reality the parameters in the model controlling rainfall, (and there are many of them) can be tweaked to give roughly the right humidity around the world and about the right rainfall (although this is sometimes disputed). This is fine provided one is not then going to operate outside the range of conditions that the models have been tuned against. Unfortunately this is what we are forced to do in the case of C0_2. Because of the parameterisation schemes, the models inevitable create more water vapour due to an initial C0₂ warming. They hardwire positive feedback of a significant magnitude. The question is whether they get the magnitude correct.

From basic physical reasons it is plausible that the water vapour feedback is real and possibly powerful. But is assumes that no other powerful negative feedback mechanisms exist in the water cycle. For example, will all the extra water vapour produce more cloud which could have the effect of reflecting more incoming solar radiation? And will the extra cloud be at high level (where it would likely heat the earth) or at low level and cool the earth. I have no idea and it is a hotly debated topic (see hereand here).If we understood these processes better, we could have more faith.

In the end we are relying upon models which are based upon poorly understood physics and operating for conditions outside the range for which they have been tuned. They have been demonstrated to be not useful in making predictions on any timescale longer than a couple of months. Although the GCMs are of considerable scientific value in pushing ahead our understanding of climate physics, it is difficult to tell if they have any value in making predictions.