Global climate change increases the frequency and intensity of rainfall, driving up the probability of extreme natural hazards. In the past, extreme rainfall has triggered flood and landslide hazards simultaneously, resulting in risk propagating among multiple hazards during personnel evacuation. However, there is currently a lack of systematic assessment methods for such scenarios, limiting the development of effective strategies for risk mitigation. To help inform policy making, this study introduces a novel concept of "cascading risk" to characterize risk propagation across interconnected hazards driven by human mobility. A quantification methodology of cascading risk is established for the first time, which comprises a novel dynamic exposure model to consider the personnel transferring and distribution in road networks. This methodology was applied to Jiangwan Town in China—one of the most exposed areas to the flood-landslide hazards. Results show a 4.78-times amplification of risk when accounting for the cascading effect for in the 2024 event, and 16 times larger in a 100-year return period extreme scenario. Optimal configurations of single-shelter, two-shelter, and three-shelter in Jiangwan Town are calculated, lowering the cascading risk up to 50%. This study provides significant foundations for policymakers to develop cost-effective emergency plans for local risk mitigation.
 

Risk assessment; Cascading effect; Exposure; Landslide; Flood