Background
The authors write that the accuracy of any GCM or global climate model "should be established through validation studies before using it to predict future climate scenarios," while adding that "although accurate simulation of the present climate does not guarantee that forecasts of future climate will be reliable, it is generally accepted that the agreement of model predictions with present observations is a necessary prerequisite in order to have confidence in the quality of a model."

What was done
Working within this conceptual framework, Miao et al. assessed the performance of the AR4 GCMs, otherwise known as the CMIP3 models, in simulating precipitation and temperature in China from 1960 to 1999 by comparing the model simulations with observed data, using "system bias (B), root-mean-square error (RMSE), Pearson correlation coefficient (R) and Nash-Sutcliffe model efficiency (E)" as evaluation metrics.

What was learned
The four researchers state that their results demonstrate that certain of the CMIP3 models "are unsuitable for application to China, with little capacity to simulate the spatial variations in climate across the country," adding that all of them "give unsatisfactory simulations of the inter-annual temporal variability." In addition, they find that "each AR4 GCM performs differently in different regions of China."

What it means
In light of their several findings, Miao et al. conclude that "the inter-annual simulations (temperature and precipitation) by AR4 GCMs are not suitable for direct application," and that "caution should be applied when using outputs from the AR4 GCMs in hydrological and ecological assessments" due to their "poor performance."