Spatial representativeness and the location bias of flux footprints over inhomogeneous areas

Turbulent fluxes over natural vegetation are spatially inhomogeneous up to a level called the physical blending height. Measurements below this blending height suffer from a location bias that varies with sensor height, wind direction and stability. The location bias is the degree, to which a flux measured at a given location differs from the aggregated ecosystem-scale flux. This location bias is analyzed using the source area/footprint approach, where the aggregated surface conditions contained in the footprint are compared to average surface conditions in the ecosystem. When a single observation event at a given location is considered, this method leads to a quantitative measure of the sensor location bias. It provides an objective estimate of how a given flux observation compares to the ecosystem scale flux. The expected sensor location bias is expressed as the fraction of surface variability not accounted for by measurements at a given sensor height, independent of its location within the ecosystem. It results from a convolution of the footprint filter function with a detailed surface condition database. These tools to examine the spatial representativeness of flux measurements are applied to an inhomogeneous savannah surface (Tiger bush) in the Sahel region of Niger, and to measurements of evaporation collected during the HAPEX-Sahel field campaign.