Slab‐ and height‐resolving models of the tropical cyclone boundary layer. Part II: Why the simulations differ

Depth‐averaged, or slab, models of the tropical cyclone boundary layer have had important practical uses in engineering design and climatological risk‐assessment studies and as components of tropical cyclone potential‐intensity models. In Part I of this article, such models were compared with a fully height‐resolving model, and it was shown that there are substantial differences in the simulations. This second part determines the reasons for these differences. The main tool used is a new model, which parametrizes the vertical structure of the boundary layer and thereby allows more accurate calculation of the surface drag and nonlinear terms. It is more accurate than the slab model, but of similar computational cost. This model, and hybrids of it and the slab model, are used to determine the consequences of the approximations in slab models. It was indicated in Part I and is confirmed here that the excessively strong inflow and too great departure of the boundary‐layer mean winds from gradient balance in the slab model are due to excessive surface drag. The tendency of slab models to produce quasi‐inertial oscillations is shown to be due to the inaccurate treatment of the vertically averaged nonlinear terms in the momentum equations. In particular, these oscillations are shown not to be due to the improper specification of the upper boundary condition as has recently been argued by Smith and Vogl. The considerable deficiencies of slab models of the tropical cyclone boundary layer are therefore inherent to their formulation. It is further shown that the dynamics leading to supergradient flow are different in the slab model and in the height‐resolving model. The possible uses of linearized versions of the slab model are briefly considered. Copyright © 2010 Royal Meteorological Society