I first introduce an improvement on historical techniques to decompose the wavenumber-frequency spectrum into contributions from standing and zonally travelling waves. Past methods directly separate the power spectrum—that is, the amplitudes squared—into standing and travelling parts. This, incorrectly, does not allow for any covariance between these waves. I propose a simple decomposition based on the 2D Fourier transform that allows one to directly compute the variance of the standing and travelling waves, as well as the covariance between them.

Variations in the relative zonal phasing between a wave anomaly and the climatological wave pattern—the “linear interference” effect—are known to explain a large part of the planetary wave driving of the polar stratosphere. While the linear interference effect is robust across observations and models of varying degrees of complexity, it is not well understood dynamically. Using the above-described decomposition, I find that the linear part of the wave activity flux is primarily driven by standing waves, at all vertical levels. This can be understood by noting that the longitudinal positions of the antinodes of the standing waves are typically close to being aligned with the maximum and minimum of the background climatology.