Establishing a Gross Primary Productivity Model by SIF and PRI on the Rice Canopy

Solar-induced chlorophyll fluorescence (SIF) has shown remarkable results in estimating vegetation carbon cycles, and combining it with the photochemical reflectance index (PRI) has great potential for estimating gross primary productivity (GPP). However, few studies have used SIF combined with PRI to estimate crop canopy GPP. Large temporal and spatial variability between SIF, PRI, and GPP has also been found in remote sensing observations, and the observed PRI and SIF are influenced by the ratio of different observed information (e.g., background, direct sunlit, and shaded leaves) and the physiological state of the vegetation. In this study, the PRI and SIF from a multi-angle spectrometer and the GPP from an eddy covariance system were used to assess the ability of the PRI to enhance the SIF-GPP estimation model. A semi-empirical kernel-driven Bidirectional Reflectance Distribution Function (BRDF) model was used to describe the hotspot PRI/SIF (PRI_hs/SIF_hs), and a modified two-leaf model was used to calculate the total canopy PRI/SIF (PRI_tot/SIF_tot). We compared the accuracies of PRI_hs/SIF_hs and PRI_tot/SIF_tot in estimating GPP. The results indicated that the PRI_tot+SIF_tot-GPP model performed the best, with a correlation coefficient (R²) of the validation dataset of 0.88, a root mean square error (RMSE) of 3.74, and relative prediction deviation (RPD) of 2.71. The leaf area index (LAI) had a linear effect on the PRI/SIF estimation of GPP, but the temperature and vapor pressure differences had nonlinear effects. Compared with hotspot PRI_hs/SIF_hs, PRI_tot/SIF_tot exhibited better consistency with GPP across different time series. Our research demonstrates that PRI is effective in enhancing SIF and PRI for estimating GPP on the rice canopy and also suggests that the two-leaf model would contribute to the vegetation index tracking the real-time crop productivity.