University of Zagreb
Statistical theory of turbulence relies on ensemble averaging of multiple realizations of the same turbulent process which can be achieved quite successfully in laboratory conditions, but never in the atmosphere. In the atmosphere, one can only assume that the average of a single realization represents an ensemble average using the ergodic hypothesis. A necessary condition for the application of this hypothesis is that the data are stationary. If the atmospheric turbulence data are nonstationary, the time average of a single realization is a weak (unreliable) estimate of an ensemble average which may enhance scatter in estimated turbulence statistics and uncertainty in similarity functions used in turbulence parameterization schemes. To assess this uncertainty, it is important to understand the physical mechanisms and dynamical processes which cause this nonstationarity. It is also unknown how nonstationary is modified by complex terrain in the presence of multi-scale flows.
In this work, nonstationarity in the near-surface atmospheric turbulence time series over complex terrain is investigated using data from the Terrain-induced Rotor Experiment (T-REX) conducted in Owens Valley, California, in March and April 2006. The data were obtained by ultrasonic anemometers mounted at six levels on three 30 m towers on the valley floor and on the western slope. A time series of certain atmospheric variable is considered to be nonstationary within an observed time interval only if both first (mean) and second (variance) order statistical moments of this variable are simultaneously nonstationary. To detect nonstationarity of these moments, the reverse arrangement test is applied to 30-min intervals of the time series of all three wind speed components and the sonic temperature.
The results indicate, in agreement with previous studies, that the observed time series are most nonstationary during the transition periods (from unstable to stable conditions and vice versa) and are more nonstationary during nighttime (stable conditions) than daytime (unstable conditions). Nonstationarity generally decreases with height above the surface for all locations, especially during transition periods when there are rapid changes in surface forcing. Also, there are indications that nonstationarity is less pronounced over the slope than on the valley floor. Underlying physical mechanisms that cause this behavior, including intermittency and slope flows, are discussed.
Friday, August 17 2012, 10:30 AM
Refreshments 10:15 AM
3450 Mitchell Lane
Bldg 2, Small Seminar Room (Rm1001)