Abstract: Gross primary productivity (GPP) is a key element of global carbon cycle. Accurate assessment of the spatiotemporal dynamics of GPP in terrestrial ecosystems is essential for global climate change and carbon cycle research. In this study, we simulated and analyzed the spatiotemporal patterns of GPP in Hebei Province from 2003 to 2020 using multi-source satellite remote sensing data of solar-induced chlorophyll fluorescence (SIF), photochemical reflectance index (PRI) and near-infrared reflectance of vegetation (NIRv) and observations from four flux sites in the North China Plain, and compared the differences of results from BEPS, MODIS and GOSIF GPP. The results showed that: (1) The multivariate linear GPP models constructed based on SIF, PRI, and NIRv, can effectively capture the GPP dynamics of flux observations and outperformed the traditional SIF-GPP linear model in grassland, shrubland, and forest (ΔR²=0.02, 0.04, 0.10), but not in cropland; (2) the GPP of terrestrial ecosystems in Hebei Province was 205.63±14.29 Tg C·a^-1 during 2003-2020, with a spatial pattern of low in the northwest and high in the southeast, as well as an overall upward trend with an average annual growth at 2.35 Tg C·a^-1; (3) The simulated GPP of cropland was significantly higher than the results of the other three models, indicating possible underestimations of crop productivity to varying degrees by these models. This study elucidated the potential of multi-source remote sensing data for accurate estimation of vegetation productivity in the North China Plain, and indicated the underestimation of BEPS, MODIS and GOSIF GPP products in croplands, providing directions for further improvement and optimization of GPP products.

Key words: gross primary production, solar-induced chlorophyll fluorescence, photochemical reflectance index, near-infrared reflectance of vegetation, model simulation