Sensitivity of Flux Footprint Modelling to Geographic Resolutions and Autocorrelation

Abstract

The standard practice of using moderate resolution satellite imagery for carbon flux studies at Ameriflux sites is called into question. In this study, a flux footprint model was implemented for modelling the carbon source areas for an Ameriflux site in Oklahoma using one year of carbon flux and meteorological measurements. A sensitivity analysis was conducted using cell sizes from 1 m to 30 m for the study site. The results indicate that over 95% of the source locations are within 93 m of the carbon flux receptor. Knowing the spatial accuracy is typically no better than ½ cell size, suggests that cell sizes of less than 45 m introduces far greater uncertainty than is appropriate.