A study of the recursive filter for computing scalar eddy covariances

Values of turbulent fluxes in the atmospheric surface layer obtained with block averaging and with recursive filtering showed consistent differences during a 4-day experiment conducted over grassland in central Kansas. Analysis of the direct and indirect impacts of recursive filtering on the calculated fluxes showed that flux values can be affected significantly by mismatched filtering in different channels, unsuitable filter initialization, a poor choice of coordinate rotation technique, or a combination of these.