The formation of tropical storms in a low-resolution Atmospheric General Circulation model is studied over the western North Pacific region during the June-October season. The model simulates the mean annual cycle of storm number in this basin quite well. Time-dependent composites of the storms are formed and analyzed, with a focus on the temporal evolution of quantities averaged in space around the storm centers. Day zero of each composite corresponds to the time at which the disturbance passes criteria for detection. The composites depict the model storms as convectively coupled, synoptic-scale vortices whose degree of coupling to convection increases at some point, leading to intensification. Variables related to disturbance intensity have significant anomalies at day -7, indicating a finite amplitude disturbance prior to "genesis". Many of these variables show similar temporal evolution, with a local minimum two or three days before day zero, and a strong increase after that for several days, followed by an eventual decrease. The precipitation reaches its maximum on day 2, the net moist static energy forcing (surface fluxes minus net tropospheric radiative cooling, each of which has an anomaly of 20-30 W m(-2) in the sense of warming the atmosphere) a day later, and dynamical variables such as vorticity and temperature still later, with broad plateaus centered around day 4 or 5. The vorticity increases at the surface at the same time as at midlevels, unlike in observed storms. The mean composite environmental vertical wind shear has a maximum amplitude on day -2 and then decreases. This could indicate a causal role of shear in limiting development, but would also be consistent with a coincidental storm motion to regions of lower shear, with development controlled by other factors. A signal in the skewness of the lower-level relative humidity distribution over the ensemble suggests that a dry lower troposphere can prevent development of a model tropical disturbance.
773RVTimes Cited:10Cited References Count:44