Dynamic Monitoring of Desertification in Ningdong Based on Landsat Images and Machine Learning

The ecological stability of mining areas in Northwest China has been threatened by desertification for a long time. Remote sensing information combined with machine learning algorithms can effectively monitor and evaluate desertification. However, due to the fact that geological environment area is easily affected factors such as resource exploitation, it challenging accurately grasp development process area. In order better play role remote technology monitoring areas, based on Landsat images, we used variety feature combinations Ningdong coal base. performance each model was evaluated various indexes. Then, optimal selected extract long-time base, spatial-temporal characteristics were discussed many aspects. Finally, driving change quantitatively studied. results showed random forest best combination had recognition than other models. Its accuracy 87.2%, kappa 0.825, Macro-F1 0.851, AUC 0.961. 2003–2017, land increased first then slowly improved. 2021, situation deteriorated. force analysis human economic activities have become dominant factor controlling desert rainfall plays an auxiliary role. study comprehensively analyzed It provide scientific basis combating construction green mines.