Feature extraction and analysis of reclaimed vegetation in ecological restoration area of abandoned mines based on hyperspectral remote sensing images

The vegetation growth status largely represents the ecosystem function and environmental quality. Hyperspectral remote sensing data can effectively eliminate the effects of surface spectral reflectance and atmospheric scattering and directly reflect the vegetation parameter information. In this study, the abandoned mining area in the Helan Mountains, China was taken as the study area. Based on hyperspectral remote sensing images of Zhuhai No. 1 hyperspectral satellite, we used the pixel dichotomy model, which was constructed using the normalized difference vegetation index (NDVI), to estimate the vegetation coverage of the study area, and evaluated the vegetation growth status by five vegetation indices (NDVI, ratio vegetation index (RVI), photochemical vegetation index (PVI), red-green ratio index (RGI), and anthocyanin reflectance index 1 (ARI1)). According to the results, the reclaimed vegetation growth status in the study area can be divided into four levels (unhealthy, low healthy, healthy, and very healthy). The overall vegetation growth status in the study area was generally at low healthy level, indicating that the vegetation growth status in the study area was not good due to short-time period restoration and harsh damaged environment such as high and steep rock slopes. Furthermore, the unhealthy areas were mainly located in Dawukougou where abandoned mines were concentrated, indicating that the original mining activities have had a large effect on vegetation ecology. After ecological restoration of abandoned mines, the vegetation coverage in the study area has increased to a certain extent, but the amplitude was not large. The situation of vegetation coverage in the northern part of the study area was worse than that in the southern part, due to abandoned mines mainly concentrating in the northern part of the Helan Mountains. The combination of hyperspectral remote sensing data and vegetation indices can comprehensively extract the characteristics of vegetation, accurately analyze the plant growth status, and provide technical support for vegetation health evaluation.