Methods for uncertainty assessment of climate models and model predictions over East Asia

ABSTRACT Model performance and uncertainty have been assessed using simulations of the climate in the 20th century based on the 21 models of the Intergovernmental Panel on Climate Change's Fourth Assessment Report ( IPCC AR4 ) and NCEP / NCAR reanalysis data. To evaluate the fidelity and reliability of the simulations of East Asian climate change, the following approaches are compared to assess the uncertainty of East Asian monsoon and climate projection in conjunction with global warming: Taylor diagrams using correlation and standard deviation of model results over East Asia (100°E–150°E, 20°N–45°N), signal‐to‐noise ratio ( SNR ), and principal‐mode comparison identified by empirical orthogonal function ( EOF ) analysis. On the basis of the Taylor diagram and SNR results, good performance models with statistically high fidelity produce higher values of warming over East Asia and an enhancement of the northwest–southeast temperature gradient between the land and ocean. This enhanced temperature gradient may strengthen the East Asian summer monsoon flow, resulting in a greater increase in precipitation along the East Asian summer rain band on the continental side of East Asia. However, the good performance models as determined by the principal‐mode comparison produce lower values of warming over the East Asia region during winter; these values are clearly different from the corresponding values obtained from the Taylor diagram and SNR approaches. These results suggest that the models that give priority to the signal associated with the first leading mode of EOF or the principal mode may predict less warming than other models. The models that predict an El Niño‐like state in response to greenhouse warming produce less warming over East Asia, corresponding to the results of the principal‐mode comparison.