Suffusion is the phenomenon responsible for internal erosion, and is the process by which finer soil particles are moved through the constrictions between the larger soil particles by seepage forces. Generally, gap-graded soil is known to be susceptible to suffusion. Meanwhile, suffusion of well-graded silty sand and the resulting soil behavior are not well understood. Moreover, the previous researches on laboratory suffusion tests focused on the study of the critical hydraulic gradient, which triggers the internal instability of the soils within a short period of time. Therefore, in this study, long-term suffusion tests were conducted on well-graded silty sand under a hydraulic gradient lower than the critical value. As a result, abrupt increases in permeability and amount of soil discharged were observed due to the progressive migration of the soil particles, resulting in suffusion even at a relatively low hydraulic gradient.

The increasing frequency of extreme weather events such as typhoons and heavy rains, driven by climate change, has intensified debris flow risks during Korea’s monsoon season, causing considerable human and economic losses. In South Korea, where mountainous terrain covers 64% of the country, localized downpours exacerbate the risk of debris flows, endangering communities and critical infrastructure. To enhance resilience and ensure sustainable risk management, the Korea Expressway Corporation developed a quantitative debris flow risk assessment system based on sensitivity and vulnerability indicators. An early warning system utilizing rainfall thresholds was subsequently introduced. However, discrepancies between rainfall data from local AWS stations and actual site conditions compromised its predictive accuracy. This study addresses those limitations by integrating the Parameter-elevation Regressions on Independent Slopes Model (PRISM) into the early warning system to enhance prediction accuracy at debris flow occurrence and non-occurrence points. Comparative analysis revealed that the PRISM-enhanced system significantly improved predictive performance. Furthermore, cumulative rainfall data from five highway sites validated the system’s reliability in short-term prediction while offering a sustainable, data-driven framework for long-term debris flow risk management. This approach strengthens adaptive infrastructure strategies, promoting more resilient transportation networks and improving public safety while minimizing environmental impacts.