A comparison of the carbon and water vapor exchange contributions of mean advection, eddy-covariance, and storage in a tall forest

An intensive experiment was carried out in a 65 m tall, 400-600 year old coniferous temperate rain forest in Southern Washington, a long-term AmeriFlux measurement site, to determine the micrometeorological components contributing to Net Ecosystem Exchange (NEE). Three-dimensional, sonic wind velocities were measured at 10 Hz 70 m and 3 m above the forest floor, as well as the concentrations of water vapor and carbon dioxide at 10 Hz. Two-dimensional wind velocities were measured at six heights between 2 m and 70 m above the forest floor. Mean concentrations of water vapor and carbon dioxide were measured at eight heights in the vertical, and in several horizontal locations at the 2 m and 30 m heights, to assist in estimation of the mean advection and storage terms in the NEE equation. Additional mean microclimatic data were measured at the same six heights as the 2-D wind measurements. Energy budget and mass exchange terms were also estimated using measured micrometeorological data and the Advanced Canopy-Atmosphere Simulation Algorithm (ACASA), which consists of a layered canopy radiation model, a physiological model for photosynthesis and stomatal control, a higher-order diabatic turbulence transport model, and a soil mass and energy transport model. The relative contributions from the eddy-covariance, mean horizontal and vertical advection components, and storage terms were compared.

It is shown that at times each of the components can be important components of the mass budget. The 'burp' in eddy-covariance exchange can follow storage of trace gases in the morning, as reported in the literature. Mean horizontal advection can be a significant portion of the under-story exchange, but appears less important at the canopy level. Mean vertical advection is significant under certain conditions. Both day-time and night-time occurrences of substantial mean advection are found, although the nocturnal occurrences appear more significant. Our results imply that mean advection components may have to be estimated at many FluxNet and AmeriFlux sites.