Ground Microspectral Threshold Classification Based On Ground Hyperspectral Vegetation Index

Surface micro-blots(plant communities,bare lands,and rat holes)are the main components of the grassland.Their distribution and area change are important indicators of grassland degradation assessment.In this study,the ground hyperspectral data was used as the research object.The data was collected in the typical desertification grassland in Siziwangqi,Inner Mongolia.The vegetation threshold method was used to find the threshold interval and the best separable pixels of the surface micro-blot to find a suitable area.The characteristic threshold classification method classifies high-precision micro-blots on the surface to provide a theoretical basis for the use of UAV hyperspectral remote sensing for quantitative inversion of desertified grasslands.In this paper,the ground hyperspectral data were collected from the ground sample plots and rat hole plots in the experimental area,and based on the ground measured hyperspectral data,based on three vegetation indices(Ratio Vegetation Index,Normalized Vegetation Index,Soil Adjustment Vegetation).The index is used to process ground hyperspectral data,and the vegetation index threshold method and visual interpretation method are used to count and analyze the thresholds of the surface micro-blots(vegetation,bare soil,and rat holes)in three typical desertification grasslands in the experimental area,and to use Spss performs Kappa coefficient verification.The results show that: based on the vegetation index threshold method,the method of identifying and classifying micro-plains in desertification grassland is feasible;the normalized vegetation index threshold method has the best classification effect on plant communities and non-vegetation communities(bare land,rat holes).The optimal separability pixel threshold is 0.4,and the Kappa coefficient has an accuracy of 96.2%;the soil-adjusted vegetation index has the best classification effect for rat holes and non-rat holes(plant communities,bare land),and the best pixel is separable.The sex threshold is 0.185,and the Kappa coefficient has an accuracy of 96.7%.In this paper,the classification and identification of surface micro-plaques in desertification grassland are initially achieved,which is of significance for the inversion of the vegetation index,the bare soil area and the number of ratholes in the typical steppe of Inner Mongolia.