<p>Photochemical reflectance index (PRI) as a proxy for light use efficiency (LUE) has the potential to improve the estimates of vegetation gross primary productivity (GPP) using LUE model. Solar-induced fluorescence (SIF) has increasingly been shown to be a promising approach for directly estimating GPP. However, a number of factors including the view-geometry and environmental variables, which may disassociate PRI and SIF products from photosynthesis, are important for the estimation of GPP, but rarely investigated. In this study, we observed multi-angle SIF and PRI in a maize field during the 2018 growing season, and compared the PRI-based LUE model and SIF-based linear model in estimating GPP. Our results showed that the averaged PRI and SIF using the multi-angle observations performed better than the single angle observed PRI and SIF in estimating LUE and, GPP respectively. We also found that the seasonal GPP dynamics were better captured by the SIF-based linear model (R<sup>2</sup>=0.50) than the PRI-based LUE model (R<sup>2</sup>=0.45), while the PRI-based LUE model performed better in estimating the diurnal variations of GPP (R<sup>2</sup>=0.71). Random forest analysis demonstrated that PAR and RH were of the most importance in the estimation of diurnal GPP variations using the SIF-based and the PRI-based models, respectively. The PRI-based LUE model performed better than the SIF-based model under most environmental conditions, while SIF should be a preference under clear days (Q>2). Thus, our study confirmed the importance of multi-angle observation of SIF and PRI in estimating GPP and LUE, and suggested that the environmental effects should be considered for accurately estimating GPP using SIF and PRI.</p>
Environment-sensitivity functions for gross primary productivity in light use efficiency models
