Optimal Frequency-Response Corrections for Eddy Covariance Flux Measurements Using the Wiener Deconvolution Method

Abstract

We describe a new direct correction approach to accurately restore frequency attenuated eddy covariance (EC) measurements. The new approach utilizes the Wiener deconvolution method to optimally estimate the original signal from noisy atmospheric measurements. Key features over conventional EC spectral correction methods include (i) the use of physics-based response functions, (ii) the ability to account for the non-linear phase contributions, and (iii) the direct restoration of the original signal rather than simulating the effect on an ideal reference spectrum. The new correction approach is compared to conventional spectral correction methods in a numerical simulation where the magnitude of the key limitations of conventional methods is explored under conditions relevant to common EC set-ups. The simulation results showed that the spectral correction methods commonly used for calculating EC fluxes introduced systematic error up to 10% to the restored fluxes and substantially increased their random uncertainty. The errors are attributed to the effect of using inappropriate response functions, failing to account for the contribution of the non-linear phase, and due to the assumption of spectral similarity on the scale of averaging intervals. The Wiener deconvolution method is versatile, can be applied under non-ideal conditions, and provides an opportunity to unify analytical and “in-situ” spectral correction methods by applying existing transfer functions to directly restore attenuated spectra. Furthermore, the Wiener deconvolution approach is adaptable for use with various micrometeorological measurement techniques such as eddy accumulation and flux profile measurements.