Improving Global Gross Primary Productivity Estimates by Computing Optimum Light Use Efficiencies Using Flux Tower Data

In the light use efficiency (LUE) approach of estimating the gross primary productivity (GPP), plant productivity is linearly related to absorbed photosynthetically active radiation assuming that plants absorb and convert solar energy into biomass within a maximum LUE (LUE_max) rate, which is assumed to vary conservatively within a given biome type. However, it has been shown that photosynthetic efficiency can vary within biomes. In this study, we used 149 global CO₂ flux towers to derive the optimum LUE (LUE_opt) under prevailing climate conditions for each tower location, stratified according to model training and test sites. Unlike LUE_max, LUE_opt varies according to heterogeneous landscape characteristics and species traits. The LUE_opt data showed large spatial variability within and between biome types, so that a simple biome classification explained only 29% of LUE_opt variability over 95 global tower training sites. The use of explanatory variables in a mixed effect regression model explained 62.2% of the spatial variability in tower LUE_opt data. The resulting regression model was used for global extrapolation of the LUE_opt data and GPP estimation. The GPP estimated using the new LUE_opt map showed significant improvement relative to global tower data, including a 15% R² increase and 34% root-mean-square error reduction relative to baseline GPP calculations derived from biome-specific LUE_max constants. The new global LUE_opt map is expected to improve the performance of LUE-based GPP algorithms for better assessment and monitoring of global terrestrial productivity and carbon dynamics.