Hyperspectral leaf reflectance simulation considering internal structure.

Abstract

The mainstream radiation transport models represented by the PROSPECT model regard the internal substances of the leaf as uniformly distributed. The influence of the uneven distribution of substances inside the leaf on the spectra of the leaf was not considered. The PIOSL-3 model proposed in this study assumes that the leaf is composed of three layers of optical properties. The particle swarm optimization algorithm (PSO) was used to determine the distribution proportion of biochemical parameters including chlorophyll, water and dry matter in each layer of leaves. In this paper, the LOPEX and ANGERS datasets were used to verify the spectral simulation effect of the PIOSL-3 model, and it can be seen from the optimization results of the parameters that the structural parameters in the leaves of different plant types are higher in the upper layer, and the distribution of chlorophyll also shows similar characteristics, while water and dry matter mainly exist in the lower layer of the leaves, and some plants also show different characteristics. In terms of simulated spectra, the PIOSL-3 model reduced RMSE mean values by 1.78, 0.39, 6.12, and 0.9, and SAM mean values by 0.07, 0.0094, 0.2267, and 0.03 compared to the PROSPECT model on the LOPEX and ANGERS datasets, respectively. The hypothesis of layered simulation is feasible, and the proposal of PIOSL-3 model provides a new idea for modeling the leaf radiation transmission process.

Plant phenotyping relevance

葉のハイパースペクトル反射をシミュレーションする新規モデルを開発し、既存モデルおよびデータセットでスペクトル再現性を検証しており、植物形質取得に関わる方法が中心である。

Code and data availability

The paper uses the public LOPEX and ANGERS leaf spectra datasets, but these are pre-existing community datasets (cited prior resources from JRC and INRA), not assets produced by this study. The authors' own PIOSL-3 model code, PSO scripts, and derived parameter outputs have no explicit availability statement or public;

No evidence-backed public reproduction asset is currently recorded.