The spatial variability of energy and carbon dioxide fluxes at the floor of a deciduous forest

Fluxes of carbon dioxide, water and sensible heat were measured using three different eddy covariance systems above the forest floor of a closed deciduous forest (leaf area index approximate to 6). The primary objective was to examine the representativeness of a single eddy covariance system in estimating soil respiration for time scales ranging from one-half hour to more than one week. Experiments were conducted in which the eddy covariance sensors were in one of three configurations: i) collocated, ii) separated horizontally or iii) separated vertically. A measure of the variation between the three systems (CV',related to the coefficient of variation) for half-hour carbon dioxide fluxes was 0.14 (collocated systems), 0.34 (vertically separated systems at 1, 2 and 4 m above the surface), and 0.57 (systems horizontally separated by 30 m). A similar variation was found for other scalar fluxes (sensible and latent heat). Variability between systems decreased as the number of half-hour sampling periods used to obtain mean fluxes was increased. After forty-eight hours (means from ninety-six half-hour samples), CV' values for carbon dioxide fluxes were 0.07, 0.09 and 0.16 in the collocated, vertically separated and horizontally separated experiments, respectively. The time dependence of variability has implications on the appropriateness of using short-term measurements in modelling validation studies. There are also implications concerning the appropriate number of half-hour samples necessary to obtain reliable causal relationships between flux data and environmental parameters. Based on the longer-term measurements, we also discuss the representativeness of a single eddy covariance system in long-term monitoring of soil respiration and evaporation beneath forest canopies using the eddy covariance method.