Robustness and Uncertainties of the “Temperature and Greenness” Model for Estimating Terrestrial Gross Primary Production

Terrestrial gross primary production (GPP) plays a vital role in offsetting anthropogenic CO2 emission and regulating global carbon cycle. Various remote sensing approaches for estimating GPP have attracted considerable scientific attentions, yet their robustness and uncertainties remain unclear. Here we evaluate the performance of the "temperature and greenness" (TG) model, a representative remote sensing model in estimating GPP, using the global FLUXNET GPP based on parameter sensitive analysis and optimization strategies. The results show that the minimum (xn) and optimum (xo) temperatures for photosynthesis are sensitive parameters but maximum temperature (xm) not. Optimized xn and xo differ largely from their defaults for more than half of 12 plant functional types (PFTs). Parameter optimization significantly improves the TG's performance in forest ecosystems where temperature or solar radiation has significant contribution to GPP. For water-limited ecosystems where GPP are strongly dependent of EVI and EVI are sensitive to precipitation, parameter optimization has limited effects. These results imply that the TG model, and most likely for other kind of GPP models using same methodology, can't be significantly improved for all PFTs through parameter optimization only, and other key climatic variables should be incorporated into the model for better predicting terrestrial ecosystem GPP.