Author(s): Kai Guo; Xiangbing Kong; Yuzhong Wen; Huashi Guo; Yinan Wang; Chunjing Zhao; Jintao Zhao

Linked Author(s):

Keywords: Deep learning; Human-induced soil erosion; Remote sensing interpretation; Soil and water conservation; Supervision

Abstract: Uncontrolled human-induced soil erosion can cause land resource degradation, damage to ecosystem service functions, and is also one of the main causes of floods, landslides, and debris flows. Using high-resolution remote sensing images, a ResNeXt AttU-Net deep learning model was designed to conduct automatic interpretation experiments on human-induced soil erosion patches in the typical watershed of the Fenhe River. Combined with human-computer interaction interpretation and preliminary judgment of compliance with prevention and control responsibilities, the human-induced soil erosion areas in the Fenhe River Basin from 2020 to 2023 were interpreted and extracted. Based on topography, administrative regions, socioeconomic factors, and other factors, the distribution and dynamic changes of soil erosion were analyzed. The results showed that the interpretation accuracy of human-induced soil erosion based on the ResNeXt AttU-Net deep learning model exceeded 80%. The more developed the regional economy, the more serious the human-induced soil erosion, among which the economic development zone had the most concentrated and severe human-induced soil erosion. The serious human-induced soil erosion in the Fenhe River basin mainly occurred in hilly areas, and the project types were mainly open-pit coal mines, processing and manufacturing projects, industrial park projects, and open-pit non-metallic mines. The number and area of human-induced soil erosion patches in the Fenhe River basin showed a downward trend, and remote sensing monitoring projects for human-induced soil erosion played an important role. The article provides a solution for accurately grasping the situation of regional anthropogenic soil erosion, providing a reference basis for the scientific management, systematic governance, and policy formulation of anthropogenic soil erosion in the Fenhe River Basin.

DOI: https://doi.org/10.3850/iahr-hic2483430201-363

Year: 2024