Abstract: The Lingbei mining area is the main production area of ion-type rare earth mines in southern my country. The serious land desertification caused by rare earth mining has become a key constraint to the sustainable development of the ion rare earth industry. This paper uses random forest to extract land desertification classification information, and completes the 29th Lingbei mining area land desertification classification thematic data set from 1986 to 2019. In order to improve the reliability of quantitative inversion of desertification, the quality and accuracy of this dataset are verified by sample points selected on Google images. Taking 2018 data as an example, the overall accuracy of land desertification classification is 93%, and the kappa coefficient is 90.67%. This data set reflects the degree of desertification in the rare earth mining area of Lingbei in a specific period and the temporal and spatial changes of land desertification in the mining area before and after rare earth mining, and provides basic data and technical support for the future control and transformation of desertification in the study area.
Keywords: rare earth mining area; desertification; random forest; temporal and spatial changes