Time Series Analysis With Application to Velocity Fluctuation Data Measured in a Wind Tunnel, Atmospheric Boundary Layer

Abstract

The techniques used to estimate the statistical parameters describing the turbulent velocity fluctuations in an urban atmospheric boundary layer simulated in a long test section wind tunnel are described. An extensive review of the theory underlying the selection of autocovariance and spectral density function estimators and of the equations needed for the digital evaluation of sample spectra using Fast Fourier Transforms is presented. A computer program for evaluation of sample spectra and autocovariances, developed on the basis of these equations and including the determination of confidence intervals for the spectral estimates, is given. The selection of series and subseries lengths and sampling intervals is illustrated through application of the program to autoregressive processes and to the atmospheric boundary layer velocity fluctuation data. The determination of integral length scales and the scales in the inertial subrange of the spectrum is discussed in particular for the velocity fluctuation data.