Detection and analysis of coherent structures in urban turbulence

The continuous wavelet transform provides a suitable tool to visualize the vertical structure of turbulence and to detect coherent structures in turbulent time series. This is demonstrated with a simple example of an artificially ramp structured time series. In this study turbulence data, i.e. the fluctuations of the horizontal wind components u′ and v′, the vertical component w′ and temperature T′, sampled with 20.83 Hz and measured simultaneously at three levels (z/h=1.5, 2.1 and 3.2, with z as the sensor height and h the height of the roughness elements) over an urban canopy in the inner city of Basel, Switzerland, are analyzed. The detection of the coherent structures was performed using the Mexican hat wavelet and the zero-crossing method. The analysis for unstable conditions shows that organized structures (ejection-sweep cycles) cover about 45% of the total run time. A conditional average from a total of 116 detected ejection-sweep sequences during 7 hours was calculated over a time window of 100 seconds. This dominating time scale was derived from peak frequencies of the wavelet spectra as well as from the Fourier spectra. It is shown that the normalized amplitudes of fluctuations of temperature and longitudinal wind speed during the events are largest at the lowest measurement level just above the canopy and decrease with increasing distance from the roughness elements. A comparison of related studies over different non-urban surfaces (mainly forests) shows that the shape of conditionally averaged ejection-sweep sequences is very similar for all canopies, however, the dominating time scale in general increases the rougher the surface is and the higher the roughness elements are.