Statistical analysis of low level atmospheric turbulence

Statistical properties of wind turbulence data measured with Model 1080 total vector anemometer and Model 1296 dual split-film probe manufactured by Thermo-Systems Incorporated have been calculated. The time series was divided into a certain number of data blocks depending on the lowest frequency of interest and the storage capacity of the computer. A moving-average and differencing high-pass filter has been used to remove the possible trend in the time series. The nonstationarity of the time series was determined by applying a non-parametric statistical test on the statistical quantities of each block in the time series. The smoothed spectral density estimates have been obtained by using cosine tapering data window in the time domain and applying combined frequency smoothing technique in the frequency domain. A fast Fourier transform algorithm requiring a no-bit reversal procedure has been developed.

A time series using the semi-empirical van-Karman Spectrum was simulated. The spectrum calculated from the simulated time series agreed well with the original spectral function from which the data was obtained.

The data used for statistical analysis were measured with either the Model 1080 total vector anemometer or the Model 1296 dual split film probe manufactured by Thermo-Systems, Inc., and the Gill propeller anemometers. The results have proved to be promising.