Exploring vegetation chlorophyll fluorescence with leaf dorsiventrality

Abstract

Plant adaxial (upper) and abaxial (lower) leaf sides have different reflectances and transmittances, and this r-t dorsiventrality has been explored and modeled in several studies. The emitted chlorophyll fluorescence is also not identical when different leaf sides are illuminated; although this f dorsiventrality has already been discovered, its effects on canopy-level fluorescence have not been explored, and there is a lack of models for explaining this phenomenon. This study develops two radiative transfer models, LFD (Leaf chlorophyll Fluorescence with Dorsiventrality) and CFD (Canopy chlorophyll Fluorescence with leaf Dorsiventrality), to model leaf dorsiventrality (including r-t and f dorsiventrality) at the leaf and canopy scales, respectively, and to explore its influence on canopy fluorescence. Evaluation of LFD against the measured spectra revealed that the proposed model accurately simulated leaf fluorescence with an RMSE of 0.056 W/m2/µm/sr, an NRMSE of 3.3%, and an R2 of 0.963. Evaluation of CFD under various conditions via comparison with the ray-tracing discrete anisotropic radiative transfer (DART) model revealed that CFD is consistent with DART, with an RMSE of 0.031 W/m2/µm/sr, an NRMSE of 1.0%, and an R2 of 0.998. The influences of leaf dorsiventrality are then analyzed based on the developed models and measured spectra. The results show that neglecting both r-t and f dorsiventrality induces a maximum relative error above 30% and a maximum NRMSE of 15.7%. In contrast, neglecting only the f dorsiventrality induces a maximum relative error of more than 30% and a maximum NRMSE of 9.3%. It is concluded that leaf dorsiventrality is an important influencing factor in canopy chlorophyll fluorescence, and the proposed models can accurately and efficiently simulate this dorsiventrality.