Selected Spectral Characteristics of Turbulence over an Urbanized Area in the Centre of Łódź, Poland

Abstract

We present the turbulence spectra and cospectra derived frommore than five years of eddy-covariance measurements at two urban sites inŁód´z, central Poland. The fast response wind velocity components were obtained using sonic anemometers placed on narrow masts at 37 and 42 m above ground level. The analysis follows Kaimal et al. (Q J R Meteorol Soc 98:563–589, 1972) who established the spectral and cospectral properties of turbulent flow in atmospheric surface layer on the basis of the Kansas experiment. Our results illustrate many features similar to those of Kaimal et al., but some differences are also observed. The velocity (co)spectra from Łód´z show a clear inertial subrange with −2/3 slope for spectra and −4/3 slope for cospectra. We found that an appropriate stability function for the non-dimensional dissipation of turbulent kinetic energy calculated from spectra in the inertial subrange differs from that of Kaimal et al., and it can be satisfactorily estimated with the assumption of local equilibrium using standard functions for the non-dimensional shear production. A similar function for the cospectrum corresponds well to Kaimal et al. for unstable and weakly stable conditions. The (co)spectra normalized by their spectral values in the inertial subrange are in general similar to those of Kaimal et al., but they peak at lower frequencies in strongly stable conditions. Moreover, our results do not confirm the existence of a clear “excluded region” at low frequencies for the transition from stable to unstable conditions, for longitudinal and lateral wind components. The empirical models of Kaimal et al. with adjusted parameters fit well to the vertical velocity spectrum and the vertical momentum flux cospectrum. The same type of function should be used for longitudinal and lateral wind spectra because of their sharper peak than occurs for the Kansas data. Finally, it should be stressed that the above relationships are well-defined for averaged values. The results for individual 1-h periods are very scattered and can be significantly different from the generalized functions.