Integrated tsunami risk framework considering agent-based evacuation modelling: The case of Saga, Kochi Prefecture, Japan

Abstract

This study develops an integrated tsunami risk framework combining stochastic tsunami hazard and agent-based evacuation modelling. The framework is applied to the case of Saga, Kochi, Japan, with a population of ∼ 2,200 anticipated to experience significant tsunami events (∼ M9.0). First, stochastic earthquake source models are generated for two values of magnitudes (M8.8 and M9.0) and used to carry out stochastic tsunami inundation simulations. Agent-based tsunami evacuation modelling using MATSim is then performed by considering four mode scenarios: single mode (pedestrian and car) and two multimodal scenarios (mix of car and pedestrian). The stochastic tsunami simulations and agent-based evacuation modelling results are integrated to estimate the risk. The effect of existing tsunami evacuation points and tsunami evacuation towers on risk reduction is also assessed. Such an integration framework is finally used to recommend tsunami mitigation strategies in the tsunami-prone regions. Results show that a significant tsunami hazard (i.e., tsunami depth up to 15 m) is expected in the Saga district, with an arrival time between 5 and 30 min. In addition, the evacuation points on higher grounds effectively save local residents' lives, as shown by the small number of affected people, specifically for the pedestrian and multimodal models (i.e. 10-100 people, excluding the car model). However, up to ∼ 1,000 people may get affected if car is the sole mode of evacuation. Therefore, the coastal community is recommended to evacuate on foot. Finally, identifying and securing sufficient higher grounds and vertical evacuation points are essential for risk reduction.