It is important to express mesoscale convection appropriately in order to simulate the development of tropical cyclones. For this purpose a cumulus parameterization scheme is constructed using an axisymmetric nonhydrostatic(Yamasaki,1983) model for 10km grid. The results are evaluated in comparison with cloud resolving (1km grid without parameterization) cases.
In order to express mesoscale convection Yamasaki's(1986) scheme includes prediction of cloud water and rainwater. In the present scheme these are also predicted. For the latent heat release Kuo's(1965) method is applied. The most characteristic point of the present scheme is that the water vapor consumed as latent heat release by parameterization is converted to parameterized cloud water. Thus not only grid scale but also parameterized condensation produce cloud water(rainwater) and they are predicted separately.
Without parameterization two unrealistic characteristics are recognized. One is strong local circulation near the center because of large latent heat release at low levels. It causes a rapid intensification of tangential velocity, and an eye wall is formed only from this local circulation. The other is weakeness of convective activity in the outer region. Precipitation and cold pool are weak and lifetime of mesoscale convection is typically about 1 hour. It is different from the cloud resolving case, in which 2 to 3 hour is dominant and becomes longer than 10 hour as frictional inflow intesifies.
The main improvement by using the parameterizaion scheme is that convection can occur and maintain at various radius with more distinct cold pool. Each convective area accompanies vertical circulation and tangential wind maximum. Temperature rises and circulation intensifies in a large area, which causes gradual development of an eye wall. Parameterized cloud water and rainwater especially contribute to long lifetime of mesoscale convection in outer rainband. They work through drag force and prevent updraft from intensifying excessively. Without this effect updraft becomes too strong to keep warming against adiabatic cooling and convection dissipates in a few hour even under intensified frictional inflow. The other effect of parameterized cloud water and rainwater is moisture supply through evaporation around convection. It provides favorable condition for maintenance of convection and for growing of next one. In the cases without this effect convection tend to stay at the same radius and formation of new cloud or propagation of convective area can not be well expressed.