Numerical Simulations Initialized with Radar-Derived Winds. Part I: Simulated Data Experiments

Abstract

Techniques to initialize the boundary layer in a numerical model with radar-derived wind fields are tested with simulated data of a gust front. Experiments show that Newtonian relaxation or ?nudging? applied to the velocity fields is not capable of retrieving the buoyancy in the gust front. The traditional thermodynamic retrieval method is then applied to the simulated velocity data. Tests are performed to determine the sensitivity to the tendency terms, time smoothing, length of the data assimilation window, random error in the velocity fields, sloping terrain, and the nonhydrostatic terms. Two methods to initialize the flow in the data-void region above the boundary layer are also examined. The first solves the linear gravity wave equation, given the forcing at the top of the boundary layer. The second assumes irrotational flow and a divergence profile that exponentially damps the motion in the vertical. Both methods result in less than 10% arms error after 40 min compared with over 40% error if no initialization is performed. The retrieval and data-filling methods examined in this paper are applied to observed gust-front cases in Part II of this study.