Impact of Road Network Development on the Spatiotemporal Soil Erosion in the Three‐River Source Region

ABSTRACT

As a vital water conservation region in China, the Three River Source Region (TRSR) plays a critical role in safeguarding regional ecological security and ensuring the sustainable utilization of water resources. In this study, the Revised Universal Soil Loss Equation (RUSLE) was employed to quantify soil erosion in the study area across three periods (2014–2016, 2017–2019, and 2020–2022). The Self‐Organizing Map (SOM) was applied to classify spatial erosion patterns within road network areas, and Partial Least Squares Structural Equation Modeling (SEM) was used to quantify the direct and indirect effects among topography, ecology, soil, road networks, grazing, and erosion, thereby elucidating the spatiotemporal evolution and driving mechanisms of soil erosion under road network expansion. The results show that (1) the RUSLE‐estimated soil loss rates for the three periods were 4114, 4278, and 4012 t/(km ²  a), respectively, exhibiting an overall initial increasing trend followed by a decline. SOM identified five erosion clusters within road network areas, forming a gradient from valley to mountain to plateau. Cluster I primarily represents areas of slight to mild erosion, which are distributed across the western region and tend to expand southward into valley zones. Clusters II, III, and V correspond to moderate erosion areas, which are located mainly in the eastern and south‐central regions, with Cluster III showing a temporal pattern of northward expansion along transportation corridors. Cluster IV represents severe erosion zones concentrated in the south‐central region. Overall, all clusters maintained relatively high spatial connectivity across the study periods. (2) SEM analysis revealed that road network expansion and grazing activities jointly amplified erosion in the early stage, whereas engineering measures and institutional interventions, such as slope protection, drainage facilities, fencing, and rotational grazing, significantly mitigated their positive effects in the middle‐to‐late stages. (3) On the basis of the response characteristics of the five cluster types, we propose differential, zone‐specific management measures that integrate roads, grazing, forestry and water for coordinated prevention and control, thereby providing scientific support for precision zonal governance. The SOM–SEM coupled framework developed in this study elucidates the mechanisms by which road network expansion and the intensification of human activities drive soil erosion through mediating pathways such as ecology, soil and topography and offers decision‐making guidance for integrated human–road–land–water management.