Paper Reviewed
Woldemeskel, F.M., Sharma, A., Sivakumar, B. and Mehrotra, R. 2016. Quantification of precipitation and temperature uncertainties simulated by CMIP3 and CMIP5 models. Geophysical Research: Atmospheres 121: 3-17.

Woldemeskel et al. (2016) describe how they quantified and compared uncertainty in precipitation and temperature from 6 CMIP3 and 13 CMIP5 climate models, which feat they accomplished by "using the square root of error variance, which specifies uncertainty as a function of time and space." And what did they learn from this exercise?

The four researchers report that "uncertainty is significantly large in regions that receive [1] large amounts of rainfall and in [2] mountainous and [3] coastal areas," all of which findings suggest, as they describe it, that (4) "global climate models fail to represent these features appropriately." And they also note in this regard that (5) "despite reduction in horizontal resolution and other improvements in CMIP5 models, no visible reduction or increase in uncertainties is obtained."

Once again we have an example of the ever learning but never coming to a knowledge of the truth syndrome.