Sensitivity of Local Model Prediction to Large Scale Forcing.
AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH
Pagination or Media Count:
Limited area models LAMs are becoming increasingly complex in order to produce accurate and finely resolved numerical weather forecasts. This thesis describes a LAM developed and used at the University of Utah and examines the choices of numerical difference schemes, vertical coordinate, and grid arrangement. A Fourier filter is employed to control computational modes and this application is tested in a one-dimensional shallow water model. The central issue studied in this research is the impact of large scale forcing on the resulting LAM forecasts. An outer model is used to provide initial and boundary conditions to the LAM. Nesting schemes tested include 1 perimeter nudging, 2 Davies nudging, and 3 spectral nudging. The latter allows the outer model to influence the development and movement of the longest wave components within the LAM, whereas the first two schemes directly modify only the circulation near the LAM boundaries. These nesting schemes are first tested on the shallow water model to compare the impact that varying the nudging scheme has on the interior forecast points, under simple conditions. Next, the various methods of large scale forcing are applied to the LAM. Variables studied include the vector wind fields, temperature and precipitation. The large scale information is input effectively by the spectral nudging scheme, while allowing the LAM to interact with the detailed model terrain. The outer models used to test the sensitivity of the LAM to changing the data used for initial conditions and nudging are the Nested Grid Model or the ETA model of the National Meteorological Center. The rms differences of the LAM forecasts obtained from different outer models are approximately the same magnitude as those due to changes in the boundary nudging approach