Filtering of Gravity Modes in Atmospheric Models

Abstract

The impact of gravity modes in atmospheric model predictions is assessed quantitatively by comparing integrations with a normal mode initialized primitive equation model and its corresponding pseudogeostrophic form to document some generally accepted presumptions. Analysis with a linear system yields horizontally scale-dependent differences in the Rossby frequencies derived from the two conditions (initialized and geostrophic) as well as differences in the initial divergence required. Nonlinear calculations indicate that initializing gravity modes in the primitive system does not affect the forecast of the Rossby modes. However, integration with the initialized primitive equation model shows differences in both the Rossby and gravity modes after five days when compared to the corresponding pseudogeostrophic model results, differences which depend both on horizontal scale and vertical mode. Comparison results become more similar if the geostrophic model is converted to balanced form. As the integration time is extended, the modal amplitudes predicted by the initialized primitive equation and geostrophic model rapidly become different. Yet the statistics of energy in the shorter scales for both these experiments during days 20?30 of the integrations are remarkably similar. Inclusion of forcing in the model showed changes in details of the response but not its fundamental character.