All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 121 15 1
PDF Downloads 7 6 1

Contribution of Linear and Nonlinear Processes to the Long-Term Variability of Large-Scale Atmospheric Flows

Thomas BrunsMax-Planck-Institut für Meteorologie, Hamburg, F.R.G.

Search for other papers by Thomas Bruns in
Current site
Google Scholar
PubMed
Close
Restricted access

Abstract

From 10 years of Northern Hemisphere streamfunction data we estimate the contribution of linear and nonlinear processes to the variability of large-scale flow patterns at time scales up to 50 days. The dynamical model is baroclinic and quasi-geotrophic, For the horizontal and vertical representation we use spherical harmonies triangularly truncated at wavenumber 15 and the first two modes of an EOF expansion, respectively. The spectral tendency equation for each mode is formulated as a linear regression model in the frequency domain, the model quality being expressed in terms of coherence spectra. Generally two regions in wavenumber space can be distinguished one dominated by wave propagation and the other dominated by nonlinear advection. The variance explained by these processes depends strongly on frequency and decreases toward long periods. A fraction of almost 50% streamfunction variance, integrated over large scales up to wavenumber m=8, nm=7, can be explained for time scales between 2 and 14 days. For periods longer than 14 days linear propagation and nonlinear advection account for at most 30–50% observed variance for a few selected modes and about 27% of the integrated variance over all wavenumbers. For these long time periods the contributions of the internal dynamics to the long term behavior of the atmosphere cannot be meaningfully discussed without simultaneous consideration of other significant processes such as slowly changing boundary conditions.

Abstract

From 10 years of Northern Hemisphere streamfunction data we estimate the contribution of linear and nonlinear processes to the variability of large-scale flow patterns at time scales up to 50 days. The dynamical model is baroclinic and quasi-geotrophic, For the horizontal and vertical representation we use spherical harmonies triangularly truncated at wavenumber 15 and the first two modes of an EOF expansion, respectively. The spectral tendency equation for each mode is formulated as a linear regression model in the frequency domain, the model quality being expressed in terms of coherence spectra. Generally two regions in wavenumber space can be distinguished one dominated by wave propagation and the other dominated by nonlinear advection. The variance explained by these processes depends strongly on frequency and decreases toward long periods. A fraction of almost 50% streamfunction variance, integrated over large scales up to wavenumber m=8, nm=7, can be explained for time scales between 2 and 14 days. For periods longer than 14 days linear propagation and nonlinear advection account for at most 30–50% observed variance for a few selected modes and about 27% of the integrated variance over all wavenumbers. For these long time periods the contributions of the internal dynamics to the long term behavior of the atmosphere cannot be meaningfully discussed without simultaneous consideration of other significant processes such as slowly changing boundary conditions.

Save