Abstract: Coal production will inevitably have an impact on the ecological environment. It has become the consensus of all major countries that vegetation restoration should be carried out in coal mining areas. Monitoring the vegetation in the mining area is an important part of the vegetation restoration work in the mining area, and plays an important role in the design, implementation, management and maintenance of the vegetation work in the mining area. The calculation of vegetation coverage based on NDVI is currently the most common method of coal mine ecological monitoring. It was discovered during is the process that calculating vegetation coverage based on NDVI would cause serious errors. Sentinel-2 data was used to calculate the NDVI of the study area using the remote sensing band inversion method tostudy the reasons for the formation of the error zone and provide a suitable method for ecological monitoring of grassland mining areas. Furthermore, the empirical comparison method was used to investigate the NDVI distribution characteristics of the Shengli and Pingshuo mining areas. This phenomenon has also appeared in other research results. The results show that NDVI can accurately reflect the surface vegetation coverage in areas with specific vegetation coverage, but there will be significant error areas in coal-covered areas in the mining area. The error phenomenon will appear in both study areas, with a greater impact in the Shengli mining area. This error phenomenon is caused by the inadequacy of the NDVI’s normalization algorithm, which makes distinguishingbetween coal-covered areas and low-to-medium-covered grasslands with similar characteristics in spectral curves impossible. To avoid the impact of this phenomenon, we propose to mask the relevant areas or replace the vegetation index in the mining area’s vegetation monitoring.