Evaluation of the interdecadal variability of sea surface temperature and sea level in the Pacific in CMIP3 and CMIP5 models

With significant impacts on the regional weather and climate, interdecadal climate variability is of great importance in understanding historical observations and predicting the climate in the near future. Using the currently available observation‐based products, this study evaluates the ability of climate models participating in phases 3 and 5 of the Coupled Model Intercomparison Project ( CMIP3 and CMIP5 ) to simulate the dominant interdecadal variability in the Pacific, i.e. the Interdecadal Pacific Oscillation ( IPO ). Our results indicate that compared with the CMIP3 models, the CMIP5 models exhibit slightly better performance in reproducing the observed interdecadal variability patterns of both sea surface temperature ( SST ) and sea level, and also exhibit smaller inter‐model spread. Climate models tend to simulate more realistic interdecadal variability patterns for SST than for sea level. A prominent deficiency among CMIP3 and CMIP5 models lies in the northwestern tropical Pacific, where the observations show large sea level variations associated with the IPO , but the simulations are usually much weaker or even of the wrong sign. These biases can be generally associated with the inaccurate representation of wind forcing patterns at the corresponding latitudes (∼12°N). We further suggest that the air–sea coupling may play a role in the bias of interdecadal wind patterns in coupled climate models, and the representation of climatological mean states in climate models could influence the simulated interdecadal variability patterns. This study provides necessary skill information on climate models for further studying the Pacific interdecadal variability, as well as for better distinguishing the climate change signal from internal climate variability.