Improved knowledge of the diffusion processes inside tall vegetation, such as forests, is fundamental to study source-sink relation. For example, it is still a research question how to properly quantify the source area and/or footprint of forest-ecosystem exchanges of heat, water, carbon and other important biogenic gases. It also provides important assistance in the scaling-up processes from biological measurements at the leave and stand level to canopy and ecosystem scale. Present Lagrangian and analytical diffusion models that may be used to address such questions have not yet been rigorously tested for diffusion processes inside forests, in part due to the practical difficulty and high cost for an extensive and meaningful field diffusion experiment in a forest environment.

Large-eddy simulation has been shown to be able to faithfully reproduce observed turbulent statistics and the 3-D details and time evolutions of important coherent flow structures (scalar microfront, sweep/ejection cycles), which dominates the transport and diffusion processes in a range of plant canopies. The scales of such coherent eddies are important in defining the so-called near-field and far-field diffusion characteristics.

In this study, a large-eddy simulation is performed to simulate diffusion from/to line/point sources/sinks inside a horizontally homogeneous forest with an observed vertical profile of plant area density and under neutral stratification. The source (water vapor) and sink (carbon dioxide) strengths are prescribed from a coupled plant model with values close to midday observations in the field during the growing season. Both concentrations and fluxes distributions (footprint) will be presented for each source/sink configuration (line, point), as well as important diffusion properties such as eddy-diffusivity and Lagrangian time scales. While the lateral line source/sink allows assessment of diffusion properties in the streamwise and vertical directions, the point source/sink permits study of diffusion in the spanwise direction. Even though the domain size in this study is greater than in previous simulations, our emphasis here is on the near-field diffusion processes.