Impact of parameterized convection on a numerically simulated tropical Cyclone structure

Abstract

ABSTRACT. Condensational heating is a primary source of energy for disturbances like a tropical storm. The resolvable scale condensation and the parameterized convection, in many fine mesh numerical models, are evaluated at intervals greater than the time step, order of a minute, used for computing dynamical processes. The latent heating may depend on the model resolution and the interval at which the precipitation physics is evaluated. Numerical results from a series of short range forecasts are compared to study the impact of varying the horizontal resolution and the interval for evaluating condensation physics, and of excluding the parameterized convective heating. A horizontal grid spacing of 40 km (coarse mesh) or 20 km (fine mesh) in National Centers for Environmental Prediction's Quasi-Lagrangian Model (QLM), and the initial data for a tropical storm case, are utilized. Resolvable scale condensation is invoked only at supersaturated grid points, and a Kuo-type convective parameterization procedure is employed.

Significant structural differences are produced when the interval for computing both parameterized convection and resolvable scale heating is changed, and these differences broaden when the horizontal resolution is increased. The central warm cote structure and storm intensity are simulated better when both condensational processes are evaluated at an interval of twelve time steps than at each time step. Vertical columns in central storm area rapidly become convectively stable, and the maximum in vertical motion and strongest horizontal winds shift in the outer storm area, when both condensational processes are invoked at each time step. The central storm area remains conditionally unstable, and strongest winds develop close to the center, when both condensational processes are evaluated at intervals of twelve time steps.

The central storm area remains conditionally unstable also in the fine mesh experiment in which the parameterized convective heating is excluded and the resolvable scale heating is evaluated at each time step. Intense vertical motion and vigorous heating develop in deep vertical columns, indicating that the heating on the convective scale is simulated as the resolvable scale heating. The vertical distribution of heating and the storm structure, during the first six hours in this case, are similar to those in the fine mesh run in which both condensational processes are evaluated at intervals of twelve time steps. However, the storm intensifies more rapidly after 6 h in the former than in the later case. Numerical results from additional experiments are presented to show that predicted storm structure is modified with a change in interval for invoking either or both condensational processes, and these circulation differences are not due to the initial spin up.

Transfer of moisture and heat from low levels into the higher troposphere in cumulonimbus clouds takes place in several minutes. Above cited and other predictions from the QLM suggest that storm structure. intensity and motion in a mesoscale model are likely to, be improved when parameterized convective heating is included; however, a parameterization scheme that concurrently produces alterations in the entire model cloud depth should be invoked at intervals of several minutes.