Multi-site evaluation of an improved SIF-based model for estimating canopy GPP

Yanping Liu, Genxu Wang, Zhaoyong Hu, Arthur Gessler, Shouqin Sun

Sun-induced chlorophyll fluorescence (SIF) offers a promising path to quantify gross primary productivity (GPP) across spatiotemporal scales. Recent studies have demonstrated the applicability of the SIF-based mechanistic light response (qMLR-SIF) model for estimating canopy GPP. However, the inadequate simulation of the key parameter qL (the fraction of open photosystem II reaction centers) restricts the applicability of the qMLR-SIF model. Therefore, we developed a SIF-based qL function to improve the simulation of qL, and thereby to enhance the performance of the qMLR-SIF model. Tower-based datasets collected from 16 sites were utilized to evaluate the improved qMLR-SIF model. Multi-site validation indicated that the improved qMLR-SIF model enhanced the accuracy of canopy GPP estimation for different plant functional types, photosynthetic pathways, and temporal scales. In particular, the improved qMLR-SIF model significantly reduced the underestimation and overestimation of GPP (y-slope = 0.94 to 1.05; y-intercept = -0.94 to 0.74 μmol CO2 m-2 s-1). Threshold analysis showed that environmental stresses decreased model performance, while even in this situation, the improved qMLR-SIF model still outperformed the original model (except for the boreal evergreen needleleaf forests). The advantages of the SIF-based qL function stem from the mechanistic link between SIF and qL and several unique features of SIF. Our findings demonstrate the enhanced applicability of the improved qMLR-SIF model, highlighting its potential for estimating photosynthesis at regional and global scales.