Improved estimations of gross primary production using satellite-derived photosynthetically active radiation

Terrestrial vegetation gross primary production (GPP) is an important variable in determining the global carbon cycle as well as the interannual variability of the atmospheric CO2 concentration. The accuracy of GPP simulation is substantially affected by several critical model drivers, one of the most important of which is photosynthetically active radiation (PAR) which directly determines the photosynthesis processes of plants. In this study, we examined the impacts of uncertainties in radiation products on GPP estimates in China. Two satellite-based radiation products (GLASS and ISCCP), three reanalysis products (MERRA, ECMWF, and NCEP), and a blended product of reanalysis and observations (Princeton) were evaluated based on observations at hundreds of sites. The results revealed the highest accuracy for two satellite-based products over various temporal and spatial scales. The three reanalysis products and the Princeton product tended to overestimate radiation. The GPP simulation driven by the GLASS product exhibited the highest consistency with those derived from site observations. Model validation at 11 eddy covariance sites suggested the highest model performance when utilizing the GLASS product. Annual GPP in China driven by GLASS was 5.55 Pg C yr(-1), which was 68.85%-94.87% of those derived from the other products. The results implied that the high spatial resolution, satellite-derived GLASS PAR significantly decreased the uncertainty of the GPP estimates at the regional scale.