Impact of the cumulus-scale effects in a mesoscale convection resolving model

At the last Conference in January 1999 were presented results from a new model as an alternative of Yamasaki(1986)'s model that intended to resolve mesoscale organized convection with incorporation of the effect of the cumulus-scale as the subgrid-scale for a 5-20km grid. One of the most significant modifications was that Kuo(1965)'s formulation was used as the cumulus-scale effect with several improvements. Since the model performance was studied with an axi-symmetric model, a three-dimensional model is used in the present study, and results from numerical experiments are presented in comparison with those from Yamasaki(1986), with the aim of better understanding the tropical cyclone (including mesoscale organized convection) and developing a better MCRM.

Our particular concern is how the behavior of mesoscale convection varies depending on the different formulation of the cumulus-scale. As was emphasized in Yamasaki(1986), heat release at low-levels is essential to the growth of mesoscale convection, and vertical distribution of heat is important for the structure and intensity of it. Since the vertical profile of heating in Yamasaki(1986) is assumed to strongly related to that of the buoyancy of the rising air parcel; temperature difference between the implicit cloud and the environment, as in Kuo(1965), both models may give similar vertical profiles of heating. However, the amount of heating may be significantly different under some conditions mainly because it is determined from the cloud mass flux that is assumed to depend on the buoyancy in Yamasaki(1986). Numerical experiments indicate that the general features of mesoscale convection and the TC can be simulated in both models if we set entrainment rate and the ratio of moisture to be used for the heating (b parameter of Kuo) properly. However, the dependence on the buoyancy and the relative humidity shows significant differences. Under Yamasaki(1986)'s assumption, convection can grow even when low-levels are not very moist if the buoyancy at low-levels are large enough, whereas heat release tends to be insufficient for the growth of convection in the new model. In both models small entrainment leads to higher peak level of heating. However, under Yamasaki(1986)'s assumption a small entrainment rate does not reduce the low-level heating so that convection may grow, while in the new model larger heating at upper-levels causes a significant reduction of low-level heating, and convection is often suppressed. An improvement in the new model is that the cumulus-scale cloud water is determined prognostically (diagnostically in Yamasaki,1986). The impact of this is also presented.