Tsunami risk framework based on agent-based evacuation modelling: the case of Kuroshio, Japan

Tsunami risk is dynamic as all elements (i.e., hazard, exposure, and vulnerability) have long- and short-term variability. The short-term variability is mainly due to the variation of human exposure during evacuation. Reliable evacuation modelling requires the adoption of agent-based frameworks that may be complicated to set up; therefore, several parameters must be defined appropriately. Once the agent-based model is deemed reliable, it is possible to integrate the agent-based evacuation modelling with stochastic tsunami hazard scenarios to compute the tsunami risk in terms of the number of people affected by the tsunami during the evacuation. We present the case of the Saga district of Kuroshio Town in Kochi Prefecture, Japan, where the anticipated tsunami from future tsunami from the Nankai-Tonankai subduction zone is very high, exceeding 20 m. For agent-based modelling, MATSim software is used. Stochastic earthquake source models for the Nankai-Tonankai tsunami are generated to perform stochastic tsunami inundation simulations. Subsequently, agent-based tsunami evacuation modelling is performed by considering four types of transportation modes. The stochastic tsunami simulations and agent-based evacuation modelling results are integrated to estimate the risk. Such an integrated framework is used to recommend tsunami fatality risk mitigation strategies in tsunami-prone regions. Results show that up to ~1,000 out of 2200 residents may get affected if car is the sole evacuation mode. Therefore, the coastal community is recommended to evacuate on foot. The developed integrated method can be extended to consider possible road blockages due to debris due to intense shaking.
The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)