The Asian Summer Monsoon: An Intercomparison of CMIP5 vs. CMIP3 Simulations of the Late 20th Century
We have compared two vintages coupled ocean-atmosphere simulations of the late 20th century (CMIP3 vs. CMIP5) to test if improvements to model physics have resulted in better representation of the Asian summer monsoon. We find that the more recently constructed CMIP5 models marginally outperform the CMIP3 models in terms of the time-mean state of precipitation and 850hPa winds, the annual cycle and interannual varaibility of precipitation, and the intraseasonal variability of outgong longwave radiation.
While model improvemennt has been found, the pattern of systemaitc model error is largely unchanged between CMIP5 and CMIP5. In terms of the time-mean precipitation, the model error looks like the observational uncertainty, suggesting that lack of adequate observations is an impediment to further model improvement.
Historical simulations of the late 20th Century have been analyzed to evaluate the ability of the CMIP5 and CMIP3 models to simulate the Asian summer monsoon. Skill metrics indicate that the CMIP5 models have improved relative to the CMIP3 models in terms of the mean state, annual cycle, interannual, and intraseasonal variability. The improvement is typically reflected by larger pattern correlations and more realistic amplitudes of rainfall and winds with respect to observations. The improved skill may lead to a better understanding of monsoon processes and more robust estimates of climate change projections over the monsoon domain.