Impact of non-local advection on flux footprints over a tall forest canopy: a tracer flux experiment

Since the early 1990s, in the planning and execution of their flux measurement campaigns, micrometeorologists have used the footprint concept to determine site quality with respect to both fetch and surface homogeneity. While the usefulness of these models has been demonstrated time and time again, there are cases which clearly call for caution. For instance, over tall forest canopies, footprint models have been applied without prior experimental validation, thus causing uncertainties in the interpretation of flux data particularly close to sources and sinks. Another example is that, despite recent advances aimed at identifying both the spatial extent of upwind sources and their individual weight to a point flux measurement, little attention has been given to the possible contribution of sources well outside the footprint region to flux measurements. This paper evaluates footprint models applied to fluxes above tall forest canopies with a tracer experiment. This work further identifies the contribution of sources outside the footprint region to an in situ flux measurement. A field campaign was performed over an 1 l-year old slash pine canopy and a passive tracer (SF6) was released from a line source deployed near the treetop. Vertical SF6 fluxes were measured using the eddy covariance technique at two positions downwind from the source. Measured fluxes during near-neutral, unstable and very unstable conditions were compared against a Lagrangian simulation and an analytical solution to the diffusion equation, two methods commonly used to predict the footprints. Both the formulations compare favorably with SF6 flux measurements except in cases characterized by wind direction-specific advection. Results suggest that, in addition to local source inhomogeneities within the footprint, in some conditions, non-local, larger scale forcings originating hundreds of meters outside the footprint envelope can contribute significantly to flux measurements. This finding is supported by sodar measurements of the three-dimensional velocity field. This finding further suggests caution to experimentalists involved with flux measurements and illustrates the fact that flux measurements must be made with an awareness of landscape-wide surface properties.