On the Spectral Distribution of Large-Scale Atmospheric Kinetic Energy

Abstract

Current interest in the limits of deterministic prediction in the atmosphere has focused attention on the longitudinal spectrum of kinetic energy at synoptic wavelengths. A number of additions and improvements on past determinations of kinetic energy are presented. We define the kinetic energy spectrum of interest as the portion of the spectrum arising from stochastic transient motion on all scales. This spectrum is estimated from four sets of data: the first two are balanced winds derived from objective analysis routines, the third streamline-isotach analyses for the tropics and sub-tropics, and the fourth untreated observed winds. For comparison of the effect of computing kinetic energy spectra in data-rich and data-poor regions, the first data act comprises balanced winds from 360° of longitude and the second from a 180° segment including North America and Europe. Spectra computed by using observed winds allow a comparison between the distribution of actual kinetic energy and that assumed by objective analysis schemes for numerical weather prediction. All estimates of the spectrum suggest that the spectral slope in the 3500–1500 km wavelength region can be represented by a power law with an exponent in the range −2.7 to −3.0. The range of uncertainty is due to the effects of smoothing and aliasing which are inherent in the synoptic wind data. The spectra presented show the variations in spectral shape as a function of season, pressure level, latitude, data density, and method of obtaining the basic wind component data.