Our current understanding of turbulence in fluids imposes that turbulence is governed by perturbations within a certain field of physical quantities. Therefore, quantifying turbulence in the atmosphere requires defining these perturbations. To define the turbulence perturbations, it is important to determine a suitable averaging time scale which can vary largely with changing atmospheric and surface conditions but oftentimes is kept constant. This averaging time scale is then used to separate the total signal into the mean and the perturbations which are used to calculate turbulent fluxes. If an inappropriate averaging time scale is used for a specific situation, this can lead to severe under- or over estimation of the turbulent fluxes.

The goal of this work is to define and quantify the averaging time scale and its temporal and spatial variability in the near surface atmospheric boundary layer over complex terrain. We use data from three 30 m towers with 6 levels of ultrasonic anemometers collected during the Terrain-induced Rotor Experiment (T-REX) conducted in Owens Valley, California, in spring 2006. We analyze an 8 hour period characterized by quiescent convective conditions early in the morning followed by the intensification of the along valley winds later afternoon. We use two different methods for the calculation of the averaging time scale: Fourier spectral analysis and multiresolution flux decomposition. Results from applying the two methods compare well and indicate a large spatial and temporal variability in the averaging time scale. We explore the causes of this variability using observations of the local and mesoscale flow structure obtained from a variety of in-situ and remote sensors.