@

초록

This study explores the integration of advanced data analysis and geostatistical modeling to enhance geotechnical resilience in urban environments prone to cascading geo-hazards. Focusing on a 4 × 4 km plain area in Seoul, South Korea, standard penetration test data were meticulously pre-processed to remove outliers and missing values, ensuring a robust input for spatial analysis. Two predictive methods, 3D ordinary kriging and mosaic-based modeling, were applied to interpolate soil strength parameters essential for assessing seismic slope stability and liquefaction risk. The mosaic-based model demonstrated superior reliability, achieving a 43% improvement in prediction accuracy through leave-one-out cross-validation compared to ordinary kriging. The results highlight the importance of conservative modeling approaches for capturing the spatial variability of subsurface conditions in densely populated, high-risk areas. This research contributes to the development of data-driven strategies for mitigating compound and cascading geo-hazards, offering valuable insights for geotechnical earthquake engineering and urban risk management. © ASCE.