Reference
Lau, K.M., Shen, S.S.P., Kim, K.-M. and Wang, H. 2006. A multimodel study of the twentieth-century simulations of Sahel drought from the 1970s to 1990s. Journal of Geophysical Research 111: 10.1029/2005JD006281.
Background
The Sahel drought, which the authors describe as "one of the most pronounced signals of climate change with devastating impacts on society," is said by them to provide "an ideal test bed for evaluating the capability of CGCMs [coupled general circulation models] in simulating long-term drought, and the veracity of the models' representation of coupled atmosphere-ocean-land processes and their interactions."
What was done
Lau et al. "explore the roles of sea surface temperature coupling and land surface processes in producing the Sahel drought in CGCMs that participated in the twentieth-century coupled climate simulations of the Intergovernmental Panel for Climate Change [IPCC] Assessment Report 4," in which the 19 CGCMs "are driven by combinations of realistic prescribed external forcing, including anthropogenic increase in greenhouse gases and sulfate aerosols, long-term variation in solar radiation, and volcanic eruptions."
What was learned
The climate scientists found, in their words, that "only eight models produce a reasonable Sahel drought signal, seven models produce excessive rainfall over [the] Sahel during the observed drought period, and four models show no significant deviation from normal." In addition, they report that "even the model with the highest skill for the Sahel drought could only simulate the increasing trend of severe drought events but not the magnitude, nor the beginning time and duration of the events."
What it means
Since all 19 of the CGCMs employed in the IPCC's Fourth Assessment Report failed to adequately simulate the basic characteristics of "one of the most pronounced signals of climate change" of the past century (the Sahel drought of the 1970s-90s, as defined by its start date, severity and duration), the results of this "ideal test" for evaluating the models' capacity for accurately simulating "long-term drought" and "coupled atmosphere-ocean-land processes and their interactions" would almost mandate that it would not be wise to rely on their output as a guide to the future, especially when the models were "driven by combinations of realistic prescribed external forcing" and they still could not properly simulate the past.