The motivation of this study arises from some uncertainties about interannual variations of tropical cyclone (TC) activity, such as the mechanism of the variability, contributory factors, and impacts of anthropogenic climate change. Although a general circulation model is becoming a promising tool to study climatological aspects of TCs, it is often pointed out that credibility of modeling studies is lost due to large model-to-model differences. Thus, we're conducting this study with the following two objectives using several versions of the NCAR Community Atmosphere Model. The first one is to verify the model's performance regarding TC simulations focusing on the sensitivity to moist convection schemes and spatial resolutions; the second one is to investigate SST-forced impacts on TC activity with an emphasis on differences between two major TC basins, the western North Pacific and the North Atlantic. So far, I performed some sensitivity experiments regarding a convection inhibition scheme, and long-term ensemble simulations using the NCAR model with two different resolutions, T42 and T170, forced by observed SST during a period from 1979 to 2000.

Although simulated TC frequencies do not much depend on resolution, more similarities to observed TCs are obtained with increased resolution. In either resolution model, simulated frequencies are very sensitive to degree of the convection inhibition. SST-forced interannual variations are more evident in seasonal frequencies, and relatively well simulated in the North Atlantic. The long-term simulation period includes two strong El Nino events, 1982-1983 and 1997-1998. These two events are similar regarding strong warm phase and rapid transition to cold phase in the second year. During these events, suppressed activity was both observed and simulated in the late season in the first year in the North Atlantic, and in the early season in the second year in the western North Pacific. These suppressed activity is consistent with large-scale fields, such as upper-level divergence for the western North Pacific and vertical wind shear for the North Atlantic. However, many observed TC variations are partly reproduced by the model. This partial capability implies some factors other than SST-forcing as well as model errors.