Power spectral artifacts in published balloon data and implications regarding saturated gravity wave theories

Abstract

There are several theories of atmospheric gravity wave power spectral densities (PSDs) which have been published. These, in turn, have inspired numerous experimental tests. The spectra involved are in the class denoted “stochastic, red noise spectra.” This means that most of the power is at the low‐frequency end and obeys a power law falloff in going to higher frequency. The present paper describes how some published experimental spectra are flawed by an artifact of spectral analysis which has not heretofore been recognized in the literature. It involves both an amplitude fluctuation enhancement and a coupling between spectral amplitude and slope, and it can be avoided only by stringent control of spectral leakage. Because of “trade‐off” considerations every data set, depending on its length and signal characteristics, requires a different method of analysis. It is therefore required that PSD analysis programs must be adjusted and tested to fit each situation. For this purpose a simple method is described to simulate data of known general characteristics for test purposes (to avoid the pitfalls). Since the papers by Nastrom et al. [1997] and de la Torre et al. [1997] have the unfortunate artifact in their analyses, their conclusions regarding saturated gravity wave theories should be reexamined.