The Role of Spatial Averaging Scale in Leaf‐To‐Canopy Scaling of Non‐Linear Processes in Homogeneous Canopies

Abstract In formulating canopy biophysical models of non‐linear processes like photosynthesis and stomatal conductance, it is necessary to average inputs over some representative spatial scale (sub‐leaf, leaf, sub‐crown, crown, whole‐canopy, etc.). The consequences of the choice of averaging scale is often unclear due to the inability of experiments to robustly quantify fluxes across the full range of relevant length scales. This study quantified errors in canopy‐level flux simulations due to input radiative flux averaging scale choice using a detailed leaf‐resolving biophysical model that represents three‐dimensional canopy geometry at the sub‐leaf level. Simulations were performed for an array of homogeneous canopies differing in leaf area index (LAI), canopy height, leaf inclination angle distribution, leaf size, leaf shape, and the size of leaf sub‐sections relative to total leaf size. Canopy fluxes based on a sub‐leaf averaging scale were compared with fluxes derived from averaging over single leaves, averaging sunlit and shaded portions of leaf area separately (i.e., two‐leaf model), and averaging multiple vertical layers and leaf angle classes separately (i.e., multi‐layer model). Canopy‐level fluxes were found to be influenced by the chosen averaging scale and other variables tested, especially LAI. Averaging approaches inherent in the two‐leaf and multi‐layer models currently used in most canopy models resulted in low or moderate over‐estimates of canopy‐level fluxes under many conditions tested. Surprisingly, averaging over entire leaves (i.e., leaf‐resolving model) had the potential to create the largest errors. This result underscored the need to sub‐divide leaves in leaf‐resolving models.