Downscaling method for red solar-induced chlorophyll fluorescence using modified kernel normalized difference vegetation index

Solar-induced chlorophyll fluorescence (SIF) is a direct proxy for photosynthetic activity. Red SIF (RSIF), primarily emitted from Photosystem II (PSII), offers closer physiological relevance than far-red SIF (FRSIF). However, canopy-level RSIF is substantially affected by chlorophyll reabsorption and multiple scattering, and thus represents only a small fraction of photosystem-level fluorescence (RSIF PS ). Existing methods for downscaling RSIF to RSIF PS , which often rely on empirical indices, exhibit limited robustness in accounting for canopy structural heterogeneity and background effects, particularly in sparse vegetation conditions. To address this limitation, we developed a Modified Kernel Normalized Difference Vegetation Index (MKNDVI) based on nonlinear mapping theory to improve the estimation of the fluorescence escape probability and RSIF PS . Based on MKNDVI, we derived the red reflectance of pure vegetation (Red MKNDVI ) and downscaled the canopy-RSIF to the photosystem level (RSIF PS-MKNDVI ). We evaluated the performance of our proposed downscaling approach using both SCOPE simulations and field measurements across two crop types (wheat and maize). By analyzing the correlations of RSIF PS-MKNDVI with the model-simulated RSIF PS (RSIF PS-model ) and absorbed photosynthetically active radiation by green leaves (APAR green ), we assessed the method's efficacy in representing total fluorescence emission across diverse canopy conditions. In comparison to existing downscaling methods, RSIF PS-MKNDVI demonstrated consistently stronger correlations with both RSIF PS-model and APAR green under a range of soil reflectance, leaf area index (LAI), and chlorophyll content (Cab), with notable improvements under low-to-medium vegetation density. Our method effectively mitigated soil background interference and chlorophyll reabsorption effects, thereby enhancing the accuracy of RSIF PS estimation. Field validation further confirmed that RSIF PS-MKNDVI maintained a stronger and more consistent relationship with APAR green across multiple years and crop types. Therefore, the MKNDVI provides a robust and effective tool for downscaling RSIF from canopy to photosystem level, which significantly improves the potential of RSIF for monitoring photosynthetic activity. • A novel Modified Kernel Normalized Difference Vegetation Index (MKNDVI) is proposed. • MKNDVI is efficient for estimating the pure red-band vegetation reflectance. • Red MKNDVI -based RSIF downscaling exhibits improved robustness across conditions. • RSIF PS-MKNDVI suppresses soil background effects under varying canopy conditions. • RSIF PS-MKNDVI maintains a stable, strong response to APAR green in C 3 and C 4 crops.