The Sensitivity of the Short-Range Forecast Skill of the NCAR Second-Generation General Circulation Model to Data Analysis and Initialization

Abstract

Initialized and short-range forecast fields derived from the NCAR second-generation general circulation model were diagnosed using kinetic energy budget analyses. Two sets of objectively analyzed data, one produced from the National Meteorological Center's Cressman successive approximation scheme and a second from the NCAR multivariate statistical objective analysis scheme, were available to produce forecasts for this study. By examination of kinetic energy budgets produced from forecast fields derived from these two objective analysis procedures, the sensitivity of the NCAR GCM to analysis was evaluated. Results indicate that even though substantial differences existed between the kinetic energy budgets derived from the. two analyzed atmospheres, subsequent forecast fields were very similar. Results also show a significant loss of kinetic energy during initialization and a continued decrease throughout the subsequent forecast periods. Much of this energy loss was in the geostrophic component of the flow. The ageostrophic motions remained near their analyzed magnitudes, but even in areas where the resulting kinetic energy generation represented a source, it was not sufficient to compensate for energy losses caused by other mechanisms. A new initialization method was developed which retained the analyzed kinetic energy levels. However, the new initialization procedure yielded no improvement in forecast skill. Failure of either analysis or initialization scheme to alter significantly the subsequent forecasts suggests that the model is insensitive to the scale of the changes being made, either in amplitude or wavelength.