Identifying Degraded Grass Species in Inner Mongolia Based on Measured Hyperspectral Data

Grassland degradation is one of the most serious ecological crises today. Using hyperspectral data to identify grass species on a more fine scale will allow for a more comprehensive exploration of grassland degradation processes from the perspective of ecological changes in grassland structure, which will aid in the development of targeted measures to control grassland degradation. To this purpose, we selected representative dominant grass and degenerative indicator grass species in typical steppe landforms of Inner Mongolia to evaluate their spectral separability. Spectral characteristics of interspecies were determined using the confidence interval mean difference method (CIMDM) developed for this article and principal component analysis based on the original first-order derivative reflectance spectral and continuum removal data, whereas species identification was determined through stepwise discriminant analysis. Results showed a detectable difference between spectral data from both the dominant and degraded indicator grass species, wherein the spectral separability of the latter was generally better than the former, among which Caragana microphylla Lam. was best, whereas Leymus chinensis, Allium ramosum L., and Stipa grandis were generally poor. Based on CIMDM results, the distribution of most band characteristics was between 540 and 750 nm. Furthermore, the recognition accuracy of both methods exceeded 90%, confirming the effectiveness of measured spectral data in identifying these grass species on a more fine scale. This article provides new concepts in identifying plant species based on measured spectra, while laying the foundation for future large-scale grass species identification and degradation monitoring using hyperspectral remote sensing imagery.