Spectral Analysis of Detailed Vertical Wind Speed Profiles

Abstract

Vertical wind speed profiles of unusually high accuracy have been measured at Cape Kennedy, Fla., using the FPS-16 radar/Jimsphere technique. Wind speeds averaged over 50 m layers were obtained at 25 m height intervals from near the surface to approximately 16 km. Sequences of these profiles show the detailed structure of the flow and the short-period changes in time. The present study investigates these data by spectral analysis, using the fast Fourier transform method of computation. The spectra of winds have the general characteristic that spectral density is approximately proportional to wavenumber to the −3 power. There are no clear minima in the spectra that indicate natural separations between predominant scales of motion as suggested by terminology such as large scale, mesoscale, and microscale. If these terms are to be used in reference to speed variations along a vertical axis, they must be defined on some other basis than the existence of clearly demarked spectral regions. Spectra were also computed for speed deviations from mean profiles, and for speed changes between successive wind profiles. These deviations and changes show predominantly the smaller, transitory features of the flow. To a large extent, these features apparently change in a rather random fashion from one profile to the next. At wavenumbers ≳0.3 cycle km⁻¹, their spectra have slopes near −5/2. These relatively steep slopes are in marked contrast to the −5/3 law generally found for wind measurements along horizontal axes, but are approximately in accord with theoretical analyses (beginning with Bolgiano) of the effects on spectra produced by a stable density stratification in the vertical. The spectra of vertical profiles are compared with previous ones computed from winds measured by aircraft in horizontal flights across jet streams. Horizontal and vertical wavelengths having equal spectral density are shown.