All natural forest ecosystems exhibit spatial variability at some range of scales. Micrometeorological measurements of CO₂ exchange are thus only expected to be representative of the forest to the extent that the biophysical forcings in the flux footprint reflect average forest conditions. Here, we examine the influence of spatial variability in biomass (as expressed by the normalized difference vegetation index, NDVI), and radiative exposure (through variations in terrain slope angle and orientation) on net ecosystem exchange of CO₂.

A high resolution IKONOS satellite scene is used to determine the distribution of NDVI in the vicinity of an AmeriFlux tower site at Morgan-Monroe State Forest in Indiana, USA. Slope angle information is obtained from a digital terrain model of the area. With regard to biomass variability, the spatial representativeness of CO₂ flux measurements at 46 m (1.8 canopy heights) above the forest is assessed by comparing the weighted distribution of NDVI contained in the flux footprint to the average NDVI distribution of the forest. An analytical, three-dimensional footprint model is used for this procedure. The source weight distribution, estimated by the footprint model, is overlaid and multiplied with the NDVI distribution, leading to an estimate of the sensor location bias for a given footprint configuration.

The sensor location bias is evaluated for a range of wind directions and stability conditions, with widely varying sizes and orientations of the flux footprints. The expected sensor location bias is determined by a footprint climatology, where the distributions of wind direction and stability are mached to measured records. The result is a quantitative assessment of the degree to which CO₂ flux measurements are representative of the forest ecosystem in which they are conducted.