Reduced-Order Models for the Large-Scale Atmospheric Turbulence

Identifying the key spatio-temporal characteristics of the large-scale atmospheric turbulence plays an important role in understanding the multi-scale dynamics of the large-scale circulation. In this study, we aim at extracting these key characteristics using data-driven reduced-order modeling techniques. We apply methods that are based on the Fluctuation-Dissipation Theorem, Koopman spectral analysis, and Green’s functions to data from idealized global circulation models and reanalysis to develop a robust and accurate reduced-order modeling framework for the large-scale atmospheric turbulence. We discuss the connections of the extracted spatio-temporal characteristics to those obtained using the conventional empirical orthogonal function analysis. We further demonstrate how the calculated reduced-order models can be used to quantify eddy-mean flow interactions in the large-scale atmospheric circulation by computing the eddy-jet feedback in the extratropical low-frequency variability.