Realistic and spatially explicit monitoring of gross primary productivity (GPP) and crop yield is of great significance in ensuring regional food security and understanding the terrestrial carbon budget. Remote sensing techniques that provide near real-time monitoring over terrestrial vegetation at high spatial and temporal resolutions have become increasingly important in assessing carbon balance in agricultural ecosystems in the context of global change. On the other hand, process-based ecosystem models are capable of quantifying transient responses of terrestrial carbon fluxes to multiple environmental changes. In this study, we used a process-based model (Dynamic Land Ecosystem Model) and sensing-based vegetation indices to estimate gross primary productivity and crop yield at > 30 eddy covariance sites globally, evaluate model performances, investigate underlying mechanisms, and identify model uncertainties.
Results/Conclusions
Our primary results suggest that the DLEM driven by remote sensing phenological indices can provide a more adequate quantification of GPP and crop yield. The model simulated results can capture seasonality and interannual variations of GPP and crop yield at most of sites. Our results highlight the need to further improve the representation of management practices, phenology, and allocation of carbon to different plant organs in process-based ecosystem model for carbon balance assessment in agricultural ecosystems at broad scales.