Figure 1. Schematic of the experimental design and the development of the predictive framework for optical properties. The upper- and lower-layer leaves from four plant species (maize, rice, cotton, and poplar), categorized into monocots and dicots, were (IMAGE)
Caption
(A). Light was absorbed by a leaf and reflected and transmitted from the leaf. The reflect light includes specular and diffuse portion, and this reflect light distribution can be modeled with BRDF (B). Leaf section microscopy images were analyzed to obtain surface roughness data (G), which, along with other phenotypic traits (H), were fed into a predictive model. The DSDI platform was developed for measuring leaf reflect light distribution (C), calibrated for data accuracy with white board standard (D). Data of anatomical and physiological traits and the reflect light distribution data were used to develop ensemble learning (EL) model, including Support Vector Regression (SVR), Random Forest Regression (RFR), and Gradient Boosting Regression Tree (GBRT), for accurate prediction of BRDF parameters, roughness (σ(λ)), diffuse reflection coefficient (k(λ)) and refractive index (n(λ)). This study develops the BRDF parameter acquisition tools and its prediction model based on the data of leaf anatomical and physiological traits, which supports canopy light-use efficiency modeling.
Credit
The authors
Usage Restrictions
Credit must be given to the creator.
License
CC BY