An improved instantaneous gross primary productivity model considering the difference in contributions of sunlit and shaded leaves to canopy sun-induced chlorophyll fluorescence

Abstract

Sun-induced chlorophyll fluorescence (SIF) from plants offers an effective proxy for estimating gross primary productivity (GPP) by modeling SIF-GPP relationships, a widely used method to evaluate the global carbon sink. However, most SIF-GPP models ignore SIF differences between shaded and sunlit leaves, resulting in GPP underestimation, particularly in dense vegetation. This study aims to partition the contributions of sunlit and shaded leaves to canopy SIF and GPP to refine the SIF-GPP estimation model. Data from 40 eddy covariance (EC) sites representing eight major biomes and TROPOMI SIF satellite data were used for site-specific and global-scale analyses. Our results showed that the contributions of sunlit and shaded leaves to canopy SIF were 80 % and 20 %, and to canopy GPP were 55 % and 45 %, respectively. For site-specific or satellite data, the SIF-GPP relationships were the strongest for sunlit leaves (R² > 0.51, RMSE = 4.03 μmol m⁻² s⁻¹, p < 0.001). The new SIF-GPP model, including sunlit-shaded SIF separation, can improve the accuracy of GPP estimation (R² = 0.53, RMSE = 4.38 μmol m⁻² s⁻¹, p < 0.001). Compared with the model established with observed data, R² was increased by 0.1, and RMSE decreased by 13.26 μmol m⁻² s⁻¹, indicating that the ‘two-leaf’ model could notably improve the SIF-GPP model. This study confirms the different contributions of sunlit and shaded leaves to canopy SIF and GPP, and ignoring this disparity would induce systematic bias in GPP estimation. Our methods and findings on sunlit-shaded SIF separation can be referenced by other studies to enhance GPP estimation accuracy.