Evaluating the East Asian monsoon simulation in climate models

Abstract

[1] Metrics are widely used as a tool for model evaluation to assess both the performance of and changes between different generations of models. However, often the choice of metrics is limited to simple root-mean-square statistics, and it can be difficult to interpret the quality of the models in representing important physical processes. In this study, metrics have been gathered from the available literature and have been refined and augmented to include the climatology, the evolution of the rainy season, and the interannual variability of the East Asian monsoon. We investigate how these process-based metrics may be used to evaluate the simulation of the characteristics of the East Asian monsoon in climate models. The metrics confirm previous findings that climate models tend to exhibit skill in simulating the climatology and variability of temperature and winds, with lower skill in simulating precipitation distribution, seasonal cycle, and interannual variability. However, this work illustrates that a wide variety of metrics is required to make a comprehensive evaluation of East Asian climate in global circulation models. These must include consideration of both the local conditions and the large-scale circulation and measures of the seasonal cycle and interannual variability. It is also apparent that careful choice of analyzed regions must be made to avoid cancellation of biases. Such comprehensive evaluation of regional climate can be useful in estimation of current climate model performance and model development.