This dissertation research project aims to develop a low- order model of low-frequency atmospheric variability with the high-frequency components parameterized quasi-stochastically. A two-level spectral primitive equations model constitutes the framework of the low-order model. Output from an unfiltered, rhomboidal-15 truncation simulation provides the data necessary to select a low-pass cutoff frequency and truncated set of planetary-scale modes to characterize the low-order model. The output will be analyzed using global spectral energetics and Eulerian diagnostics to study the relationship between the low- frequency variability and high-frequency eddy forcing. Flux- gradient and purely stochastic eddy forcing parameterizations will be evaluated by means of statistical analysis of the resolved variables and eddy forcing time series. A quasi- stochastic parameterization that reflects salient features of the eddy/mean flow feedback will be developed. Several sets of Monte-Carlo simulations using the low-order model will be run to examine spatial and temporal characteristics of the intraseasonal predictability degradation under low and high zonal index states, and under cases of regionally persistent low and high geopotential height anomalies. This work is important because it will help delineate the high frequency synoptic scale events that have significant influence on low frequency planetary motions. Hence, it should contribute to improving our capability to make medium range forecasts.
