Improved modeling of gross primary productivity (GPP) by better representation of plant phenological indicators from remote sensing using a process model

Phenology is a significant indicator of ecosystem functioning and is one of the most important controllers of gross primary productivity (GPP). The Integrated Terrestrial Ecosystem C-budget model (InTEC) predicts carbon cycling by modeling a number of ecosystem processes, and in particularly, phenology derived from a degree-day metric. However, empirical temperature thresholds may not well represent ecosystem growth at low latitudes. Here, using 30-year Advanced Very High Resolution Radiometer (AVHRR) normalized difference vegetation index 3rd generation (NDVI3g) data (1983-2012), we obtained the start (SOS), end (EOS) and length of growing season (LOS) with three algorithms from time series of NDVI for forests ecosystems of China. The phenology module was then incorporated into the InTEC model before validation using ground observations from eddy covariance measurements. Our results showed that compared with temperature-based phenology of the original model, using NDVI-based phenology improved modeling of GPP. The modified InTEC model was used to analyze the spatial and temporal patterns of GPP for forest ecosystems of China during 1983 to 2012. We found that remote sensing-based phenology was more reliable than temperature-based phenology for large-scale analysis. Using the modified InTEC model, we revealed that the GPP of China's forests ecosystems increased over 1983-2012 with high spatial heterogeneity, with a mean of 1.31 Pg Cyr(-1). Our results demonstrated the significance of remotely sensed phenology for improving the accuracy of GPP modeling with ecosystem models, which is enlightening for the large-scale evaluation of carbon sequestration.