earticle

영어

This study aimed to identify key factors influencing landslides and analyze their relationships in mountainous areas altered by forestry activities, such as timber harvesting and forest road construction. The study areas included Yeongju and Yecheon (2020) and Jecheon and Chungju (2023), where landslides had previously occurred. A total of 661 landslide occurrence sites and 661 randomly selected non-occurrence sites from the same watersheds were analyzed using logistic regression. Ten variables were used, including nine environmental factors Topographic Wetness Index (TWI), slope, curvature, slope length, soil depth, tree diameter class, bedrock type, aspect, and forest type and one variable representing forestry activity type. The dataset was divided into training (70%) and testing (30%) sets, and a logistic regression model was developed using backward elimination conducted across 100 iterations. The model with the highest AUC (AUC = 0.792) was selected as optimal. Based on z-values, the most influential factors were slope curvature, forestry activity (forest roads), and tree diameter class. Odds ratio analysis revealed that forestry activity (forest roads), tree diameter class, north-facing slope aspect, slope curvature, and slope angle had the strongest influence on landslide occurrence. Overall, forestry activity areas, particularly forest roads and timber harvesting sites, were more strongly associated with landslide occurrence than other environmental variables. These results indicate the need for more targeted research on forestry-impacted slopes and emphasize the importance of incorporating indicators that reflect the specific nature of forestry activities, rather than simply their presence.