Retrieval of purification ability of urban forest to SO2 stress based on the coupling of radiative transfer and AO-DELM models

• DOI: 10.1016/j.jag.2023.103644
• Publication Date: 2024-01-09T18:04:15Z
• Authors: Aru Han; Yongbin Bao; Zhijun Tong; Xingpeng Liu; Song Qing; Yuhai Bao; Jiquan Zhang
• Keywords: Red edge

Currently, hyperspectral remote sensing technology used for vegetation monitoring mainly uses empirical and semi-empirical statistical methods to calculate heavy metal content. Combining physical models machine learning algorithms is an effective method estimating biochemical parameters without much ground measurement data. However, a deep extreme (DELM) has faster training speed better generalization ability. By introducing the Aquila Optimizer (AO) algorithm, process of DELM can be accelerated. This study combined PROSAIL model, chlorophyll concentration information, SO2 purification ability, comprehensively applied analyze optical characteristics urban forest rate other factors. A coupled model (PROSAIL + AO DELM) was constructed simulate forests canopy ability then inversion. These results indicated that Syringa oblate Lindl. (S. oblate) Ulmus pumila "Jinye" (U. pumila) had moderate capacities, whereas Prunus cerasifera var. atropurpurea Jack. (P. cerasifera) were low. The highly sensitive in green, red, red-edge spectral ranges. In estimation by subset (Corresponding Sentinel-2 band), NDI, DI, RVI indices, best performance, with R2, root mean square error (RMSE), relative percent deviation (RPD) 0.73, 0.056, 1.61, 0.68, 0.096, 1.06 T2 (low concentration) T3 (high treatments, respectively. extended multispectral images (Sentinel-2), where NDI inversion closest those field monitoring. indicate this potential on large scale. addresses research gap regarding rapid, non-destructive, low-input evaluation plant capacity achieves non-destructive detection from point static dynamic. enables efficient, cost-effective air levels, provides basis predicting regions later stages.