The characteristic averaging time for the surface-layer fluxes

Abstract:

The choice of the representative averaging time to compute surface-layer fluxes remains a source of discrepancy between studies. While the Ogive function (Oncley et al. 1996) has become the standard approach to determine the ‘physically sound’ averaging time, uncertainty remains on how to precisely select the minimum-necessary averaging time. Alternatives based on a multiresolution analysis exist and they can further provide a characteristic time-scale separating turbulence from mesoscale motions (Vickers and Mahrt, 2003, 2005). Yet little is learned from the inherent turbulent time scales and their corresponding contribution to the overall surface-layer fluxes.

Here a new approach based on a Proper Orthogonal Decomposition (POD) will be presented, where the extracted characteristic averaging times are energy optimal. Consequently one can now select from these POD-provided characteristic times, the most relevant time scales depending on the desired application and based on the overall energy contribution to the surface-layer fluxes. One of the main advantages of the POD-technique compared to traditional Fourier analysis or wavelet decompositions is that the shape of the eigenfunctions is directly dictated by the input data and that the resultant eigenfunctions are energy ordered in the most optimal sense. Thus, one can really determine a-priori the most ‘physically relevant’ averaging time without source of ambiguity. Examples of the proposed approach in varied complex scenarios will be presented, spanning multiple atmospheric stratifications and topography driven flows. Further, results exploring the potential of the POD technique to determine the proper averaging times for tilt corrections will also be presented.