Sensiticity Of Global Climate And East Asian Summer Monsoon To Different Dynamic Cores With CAM3.1

This paper examined the sensitivity of CAM3.1 simulations of East Asian summer monsoon (EASM) to the choice of dynamic cores using three long-term simulations, one with each of the following cores:the Eulerian spectral transform method (EUL), semi-Lagrangian scheme (SLD), and Finite-Volume approach (FV). Our results indicate that the dynamic cores significantly influence the simulated fields not only through dynamics, such as wind, but also through physical processes, such as precipitation.Monthly mean circulation and precipitation data are used for analysis of sensitivity of global climate to different dynamic cores with CAM3.1. It found that dynamic cores significantly affect the simulation. The largest deviations between the three dynamical cores are the temperature on tropopause and lower stratosphere, precipitation in the tropical west Pacific, the intensity of subtropical high, Icelandic low, cold high pressure over land and Aleutian low in winter, and westerly jets and tropical easterly jet. SLD has better ability in most aspects of above simulation.Dynamical cores have considerable affection to the simulations of East Asian summer monsoon. Generally speaking, SLD is superior to EUL and FV in simulating the climatological features of EASM and its interannual variability. The SLD version of the CAM model partially reduces its known deficiency in simulating the climatological features of East Asian summer precipitation. The strength and position of simulated western Pacific subtropical high (WPSH) and its ridge line compare more favourably with observations in SLD and FV than in EUL. They contribute to the intensification of the south-easterly along the south of WPSH and the vertical motion through the troposphere around 30°N, where the subtropical rain belt exists. Additionally, SLD reduced the biases of zonal and meridinal winds even if the three experiments are similar with each other. Considerable systematic errors of the seasonal migration of monsoon rain belt and water vapour flux exist in all of the three versions of CAM3.1 model. The interannual variation of EASM is found to be more accurate in SLD simulation, which reasonably reproduces the leading combined patterns of precipitation and 850-hPa winds in East Asia, as well as the 2.5-year and 10-year periods of Li-Zeng EASM index. These results emphasise the importance of dynamic cores for the EASM simulation, as distinct from the simulation's sensitivity to the physical parameterisations.The physical parameterizations used by SLD and FV are developed for and more suitable to EUL version. They should be further adjusted carefully to optimize the SLD and FV ones. Additionally, CAM3.1 with SLD version has much higher computational efficiency and beneficial to the platform parallel computing.