The landfall and structure of a tropical cyclone: The sensitivity of model predictions to soil moisture parameterizations

A regional mesoscale multi-level primitive equation model is used to predict the landfall and structure of a tropical cyclone. Three areas of model sensitivity are addressed in this paper; (1) the horizontal resolution, which includes the representation of orography; (2) the impact of an improved representation of the distribution of land surface soil moisture on the landfall problem; and (3) the sensitivity of the storm to lateral boundary conditions. A diagnostic part of this study describes a statistical regression approach to determining a ground wetness parameterization from moisture budget computations to derive estimates of surface fluxes, which are used to determine the parameterization. The model sensitivity analysis compares several versions of ground wetness parameterization. The experiment where perfect (i.e., based on analysis of observations) boundary conditions are used is defined as a bench-mark. At the highest horizontal resolution (=50km) using the ground wetness obtained from the regression, the best results were found for the structure and motion of the tropical cyclone. When the boundary conditions from a global model are used at a resolution T106 (roughly 100 km resolution for the transformed grid), the results degrade somewhat. The rain bands are predicted, but do not contain the same detail. Several other sensitivity experiments illustrate the degree of degradation of rain bands, precipitation distribution, hurricane structure, and phase speed errors as the lateral boundaries, resolution, and ground wetness parameterization are altered.