Accurate remote sensing inversion of mangrove canopy chlorophyll (Cab) is vital for dynamically monitoring ecosystem health. Yet canopy structure induces strong uncertainty in Cab inversion, complicating model development, parameter optimization, and inversion accuracy. To explore the influence of the canopy on Cab inversion, this paper combines PROSAIL with machine learning to analyze mangrove canopy spectral characteristics, focusing on the leaf area index (LAI) and average leaf angle (ALA), and quantifies their individual/interactive effects on Cab. Results: (1) Mangroves exhibit distinct spectral characteristics, with reflectance in the green and red-to-near-infrared range consistently intermediate between those of Phragmites australis and Spartina alterniflora. (2) Leaf structural parameter (N), LAI, and ALA modulated canopy reflectance across the spectrum, with the LAI exerting influence on overall spectral region and other traits acting locally. (3) Inverting Cab using LAI and ALA separately, support vector regression outperforms random forest, improving accuracy by 31% (R² = 0.83) and 13% (R² = 0.88) and reducing RMSE by 69.51% and 73.75%, respectively. (4) With LAI–ALA synergy, a synergistic physically constrained model achieved R² of 0.74 and RMSE of 3.07. The model showed superior performance in independent validation (R² = 0.75, RMSE = 1.42), surpassing a machine learning based on vegetation indices (R² = 0.49, RMSE = 11.24). Its practical utility has been confirmed in the Aojiang Estuary mangroves.
Zhang et al. (Thu,) studied this question.