Tropical intraseasonal oscillation simulated in an AMIP-type experiment by NICAM

It is the first time for the non-hydrostatic icosahedral atmospheric model (NICAM), at a horizontal mesh size of approximately 14-km, to conduct a continuous long-term Atmospheric Model Intercomparison Project-type simulation. This study examines the performance of NICAM in simulating the tropical intraseasonal oscillation (ISO) from a statistical point of view using 30-year data (1979-2008) in the context of the bimodal ISO representation concept proposed by Kikuchi et al., which allows us to examine the seasonally varying behavior of the ISO in great detail, in addition to the MJO working group level 2 diagnostics. It is found that many of the fundamental features of the ISO are well captured by NICAM. The evolution of the ISO convection as well as large-scale circulation over the course of its life cycle is reasonably well reproduced throughout the year. As in the observation, the Madden-Julian oscillation (MJO) mode, characterized by prominent eastward propagation of convection, is predominant during boreal winter, whereas the boreal summer ISO (BSISO) mode, by a combination of pronounced eastward and northward propagation, during summer. The overall shape of the seasonal cycle as measured by the numbers of significant MJO and BSISO days in a month is relatively well captured. Two major biases, however, are also identified. The amplitude of the simulated ISO is weaker by a factor of similar to 2. Significant BSISO events sometimes appear even during winter (December-April), amounting to similar to 30 % of the total significant ISO days as opposed to similar to 2 % in the observation. The results here warrant further studies using the simulation dataset to understand not only many aspects of the dynamics and physics of the ISO but also its role in weather and climate. It is also demonstrated that the concept of the bimodal ISO representation provides a useful framework for assessing model's capability to simulate, and illuminating model's deficiencies in reproducing, the ISO. The nature and causes of the two major biases are also discussed.