An Assessment Of Global Land Susceptibility To Wind Erosion Based On Deep-active Learning Modelling And Interpretation Techniques

This article discusses the development of a novel methodology using deep learning and active learning models to map global land susceptibility to wind erosion. The results showed that the GRU-AL model was the most accurate, with 38.5% of lands classified as very low susceptibility and 26.1% as very high susceptibility. Interpretation techniques revealed the importance of soil carbon content and precipitation in predicting wind erosion.