A Numerical Case Study of the Implications of Secondary Circulations to the Interpretation of Eddy-Covariance Measurements Over Small Lakes

We use a large-eddy simulation (LES) to study the airflow patterns associated with a small inland lake surrounded by a forest of height one-tenth the radius of the lake. We combine LES results with scalar dispersion simulations to model potential biases in eddy-covariance measurements due to the heterogeneity of surface fluxes and vertical advection. The lake-to-forest transition can induce a non-zero vertical velocity component, affecting the interpretation of flux measurements. Significant horizontal gradients of mean concentration are generated by the forest carbon sink and lake carbon source, which are transported by local roughness-induced circulation. We simulate six hypothetical locations for flux towers along a downwind gradient at various heights, and calculate at each location the effects of both the average vertical advection and average turbulent-flux divergence of . We compare our model results with an analytical footprint model to find that the footprint predicted by the analytical model is inaccurate due to the complexities of advection for our test case. Similar small lakes surrounded by forests are likely affected by these phenomena as well. We recommend specialized levelling of sonic anemometers to reduce the effects of non-zero wind components. Flux towers over small water bodies should be constructed at a distance 0.5-0.67 times the diameter of the lake to provide ample separation from the areas affected by the rotor effect of the upwind forest/lake transition and the updraft at the downwind edge. Finally, we also suggest the filtering of wind directions based on the Higgins ratio.