Abstract: Gross primary productivity is a key parameter for the research of global carbon cycle and global change. The remote sensing-based method is the mainstream approach to estimate GPP of terrestrial ecosystems.Solar-induced chlorophyll fluorescence is directly related to plant photosynthesis, and it is a signal emitted by the photosystem after plants absorb sunlight energy. Solar-induced chlorophyll fluorescence remote sensing can obtain vegetation growth status information in time. On the basis of the GPP-SIF empirical linear estimation model, some factors affecting the photosynthetic capacity and canopy SIF emission were introduced to construct a theoretical model of GPP estimation based on near-infrared fluorescence. The model is a goodremote sensing tool to monitor vegetation without severe long-term external stress.Verification analysis was carried out in different types of vegetation with the GOME-2 SIF products, FLUXNET2015 GPP products and MODIS GPP products.The research results showed that the estimation accuracy of the model on all vegetation types was greatly improved compared with the empirical linear estimation model. At the same time, the model can better reflect the seasonal change characteristics of the different vegetation types represented by each site. The application on the scale also achieved good results.