Recent observational studies show that forest canopies strongly affect wind and scalar profiles in the roughness sublayer above canopy top. Turbulence fluxes found using the covariance technique are usually reported as if measurement sites were horizontally homogeneous. Though this is recognized to be an approximation, it has proved difficult to quantify how the presence of canopy gaps, a regular feature in many forests, might lead to biased fluxes. We report here on the results of large eddy simulations (LES), which have been used to infer the spatial distributions of mean quantities and higher order moments. This work was done in preparation for field observations planned in the rain fores as part of LBA-ECO.

Our aims are to determine the distortion produced by the gap to the flow and to locate where to place an observational tower that would be the most representative of the entire flow. Using Smolarkiewicz LES, five different simulations were performed, varying features as gap size, mean wind speed and heat flux at the gap surface. In all of them a horizontally uniform scalar flux was imposed at the ground to serve as an analogy to respiratory CO2, soil evaporation, radon or any scalar emitted primarily at the surface.

Results show that a roll circulation is started near the ground at the downwind side of the gap, in resemblance to the classic flow over a bacward facing step. The circulation only happens if either gap size, mean wind speed or surface heating are large enough. The gap is shown to anchor the scalar fluxes, acting as a “natural chimney”. Local fluxes over the gap can be as much as 3 times larger than the horizontal average. While it is clear that towers should be located away from the gaps to avoid these effects, this is not always practical. Determination of area-averaged, representative fluxes may need to account for gaps and consider the fact that fluxes are higher near them.