Citation

Faucher JE, Dávila S, Hernández-Cruz X. Int. J. Disaster Risk Reduct. 2020; 49: e101606.

Copyright

(Copyright © 2020, Elsevier Publishing)

DOI

10.1016/j.ijdrr.2020.101606

PMID

unavailable

Abstract

Phenomena like tsunamis cannot be predicted, and simulation models are continuously being used in an attempt to understand how individuals will react in an emergency evacuation event. More specifically, pedestrian evacuation simulation can be used to better understand how pedestrians will escape an at-risk area. This work presents a pedestrian evacuation model (PEM) for the municipality of Rincón, Puerto Rico. This PEM provides insight on the vulnerability of Rincón to tsunamis using an iterative, optimization-based algorithm that fuses anisotropic least cost distance (ALCD) and agent-based (AB) approaches. This work advances the ALCD literature by providing a computationally-feasible PEM implementation capable of: (1) strategically initializing the location of pedestrians based on infrastructure and land use considerations, (2) managing individual as well as group-based evacuation, (3) assigning evacuation speeds and responses using probability and machine learning approaches, and (4) penalizing evacuation times using fatigue and reaction delays. The results of this PEM will provide emergency managers with a more realistic depiction of the time to reach safety using a systematic framework.

RESULTS, based on the worst-case scenario, indicate that 32.44% of the population in the tsunami evacuation zone (TEZ) would reach safety under 5 min, 41.96% between 5 and 15 min, and 25.60% would require more than 15 min. Unless effective public policy and mitigation strategies take place, the aftermath of a tsunami that inundates the Rincón coastline in less than 5 min could amount to up to 3312 casualties or 68% of the population in the TEZ.

Language: en

Keywords

Agent-based; Anisotropic least cost distance; Pedestrian evacuation models; Simulation; Tsunamis