원문정보
초록
영어
Accurate determination of rainfall thresholds is essential for effective landslide prediction and early warning. This study compares two methodologies for analysing landslide-triggering rainfall thresholds in South Korea: one using data from the single nearest station selected through recent site verification, and the other based on averaged rainfall data from multiple nearby observation stations. I–D and E–D curves were derived using quantile regression with consistent rainfall event thresholds, and predictive performance was evaluated using ROCbased classification accuracy (AUC). The single observation station approach exhibited considerable variability, whereas the average-based method produced more stable and consistent threshold estimates. The comparison identifies the strengths and limitations of each approach depending on the availability and distribution of observational infrastructure, providing insights into effective data utilization strategies for landslide early warning systems