Spectral characteristic analysis of vegetation in typical subalpine peatlands

Dajiuhu Wetland, a rare mid-latitude peatland ecosystem, harbors a diverse assemblage of rare and unique plant species. Given its vulnerability to intensifying climate extremes and anthropogenic disturbances, high-resolution monitoring of vegetation composition and spatial dynamics is essential to support effective conservation and functional integrity. Hyperspectral remote sensing, grounded in spectral feature analysis, is crucial for species identification and community dynamics monitoring. This study focused on the leaf-level hyperspectral data of 23 representative plant species in Dajiuhu Wetland. Comparative analysis of raw spectral profiles across multiple species, followed by application of Savitzky-Golay smoothing, first derivative transformation (FD), multiplicative scatter correction (MSC), and spectral feature parameter analysis, enabled the extraction of distinct spectral signatures and interspecies variation. Characteristic spectral parameters were computed and used as input for partial least squares discriminant analysis (PLS-DA) to evaluate species separability. Results showed that (1) raw spectra exhibited canonical features of wetland flora, with pronounced differences in the near-infrared region among most species, which was further enhanced by spectral transformations; (2) derived spectral parameters, particularly those associated with the “three-edge” regions, were closely related to the biochemical properties of vegetation and facilitated species-level discrimination; (3) PLS-DA models incorporating FD and MSC preprocessing achieved a mean coefficient of determination of 0.548, confirming the robustness of spectral-based classification despite variable predictive accuracy across taxa. In summary, this study elucidated critical linkages between wetland vegetation spectral parameters and physicochemical characteristics, validated the feasibility and effectiveness of species classification using hyperspectral remote sensing, and provided scientific evidence and support for dynamic monitoring and rational management of wetland vegetation.