Terrestrial gross primary production: Using NIR_V to scale from site to globe

Terrestrial photosynthesis is the largest and one of the most uncertain fluxes in the global carbon cycle. We find that near-infrared reflectance of vegetation (NIR_V ), a remotely sensed measure of canopy structure, accurately predicts photosynthesis at FLUXNET validation sites at monthly to annual timescales (R²  = 0.68), without the need for difficult to acquire information about environmental factors that constrain photosynthesis at short timescales. Scaling the relationship between gross primary production (GPP) and NIR_V from FLUXNET eddy covariance sites, we estimate global annual terrestrial photosynthesis to be 147 Pg C/year (95% credible interval 131-163 Pg C/year), which falls between bottom-up GPP estimates and the top-down global constraint on GPP from oxygen isotopes. NIR_V -derived estimates of GPP are systematically higher than existing bottom-up estimates, especially throughout the midlatitudes. Progress in improving estimated GPP from NIR_V can come from improved cloud screening in satellite data and increased resolution of vegetation characteristics, especially details about plant photosynthetic pathway.