Year: 2025
Type: conference-paper
Abstract: Currently soil erosion is a more serious situation, in this study the convolutional neural network (CNN) and graph neural network (GNN) technology is used. Through the combination of the two algorithms: the remote sensing image features identified by CNN and the geospatial relationship features processed by GNN are combined to analyse the changes in soil and water conditions and predict the degree of soil erosion, so as to provide a decision-making basis for precise prevention and control. In the whole process, CNN provides basic spatial feature data for GNN, while GNN further explores the complex relationship between geographic elements on the basis of CNN, and the two are complementary to each other, so as to jointly achieve more accurate monitoring, simulation, and prevention and control strategy formulation for soil erosion. The data and conclusions fully reflect the advantages of these two algorithms over other methods for soil erosion control.
DOI: 10.1117/12.3059150
Authors: Yuhang Zhu, Yuhan Zhou, Xueyu Huang, Hongjiao Huang, Maoyi Li
Language: English
Subfield: Soil Science
Domain: Life Sciences
SDG: Life on land
Open Access status: hybrid
OpenAlex ID: W4409795241