Abstract: As a typical disturbed patch in the mining area, dump is a key object of ecological restoration in the mining area. Studying species diversity of artificially reconstructed dump ecosystem and analyzing vegetation coverage and configuration under different slopes can provide reference for plant species selection and configuration scheme for the vegetation reconstruction project in dumping site. With the north slope of Baorixile coal mine dump as the research area, we used UAV hyperspectral remote sensing technology to obtain images. Combined field measurements, we established a typical vegetation spectrum library. We used the spectral angle method to realize the mapping of vegetation type distribution and diversity analysis. The results showed that: (1) The plant community had succeeded from one with species from single-family to that with 15 families, 28 genera, and 32 species. Perennial herbs replaced annual or biennial herbaceous pioneer species. (2) Asteraceae, Leguminosae, Rosaceae, and Gramineae played a major role in vegetation reconstruction, which is consistent with the main plant families of the surrounding natural grassland, showing a forward succession. (3) Vegetation coverage of the study area was mainly at medium and good levels. Excellent vegetation coverage types had a significant distribution advantage on slope III (5°-15°), with a correction coefficient (k) of terrain area difference of 1.11. (4) The slope range of 14°-15° was suitable for plant growth. When the configuration was Cleistogenes squarrosa (29%) +Chenopodium glaucum (28%) + Serratula polycephala (20%) +Astragalus adsurgens (15%) +Sonchus oleraceus (6%), vegetation coverage was the best, which could achieve a better ecological restoration effect. This study provides scientific basis for the terrain shaping, plant species selection, and configuration of mining dumps.

Key words: dump, hyperspectral image, vegetation succession, species diversity.