Correlation and structure function analysis in the quality assessment of long-term eddy-covariance measurements over a deciduous forest

摘要

Lagged covariance (correlation) functions and structure functions, along with spectral analysis are very useful in the quality assessments of eddy-covariance measurements. For example, certain features of these functions under homogeneous and stationary conditions can be used to examine whether the theoretical requirements of the eddy-covariance technique are met. Both functions can provide important information about the turbulence (e.g., characteristics length scales, intermittency) as well as the effects of stratification (Mahrt and Gamage, 1987). If there is any similarity in such functions for different scalars, it could also be used to correct for sensor separation. One advantage of the structure and lagged correlation functions over the spectral analysis is that they do not require complicated numerical schemes such as the fast Fourier transform. Structure function and lagged correlation function are generally smooth and stable compared to spectra and co-spectra, especially at smaller scales, and thus do not require smoothing, as often needed with spectral analysis. Structure functions are less vulnerable to trends, in particular at smaller scales and to linear trend, than lagged covariance functions.