Key points are not available for this paper at this time.
Both pre-disaster approaches, e.g., mitigation and preparedness, and post-disaster approaches, e.g., response and recovery, play important roles to mitigate the damage from large-scale disasters. From the viewpoint of disaster response, there have been studies on evacuation guiding schemes and applications using evacuees’ mobile devices, e.g., smart phones. On the other hand, disaster preparedness has also been studied mainly on geographical information analysis, e.g., road blockage probability and people flow data. The road blockage probability is the probability that the corresponding road is blocked due to collapse of roadside buildings when an earthquake occurs. The people flow data express the people flow in usual time. In this paper, with the help of evacuation guiding schemes, road blockage probability, and people flow data, we propose a road network risk analysis approach that considers people flow in both ordinary and evacuation situations, which can be used to as a tool to strengthen the urban fabric for fostering better evacuees’ responses in disaster situations. First, the proposed approach derives ordinary road demand, which is the degree of road usage at a certain interval in an ordinary situation, from the people flow data. Then, it calculates evacuation road demand, i.e., the degree of road usage at a certain interval in an evacuation situation, by extending the edge betweenness centrality under the assumption that people located according to the ordinary road demand move to refuges along their evacuation paths. Finally, it detects roads with high risk of encountering blocked road segments by combining the road blockage probability and evacuation road demand. Through numerical experiments under a case study of Arako area of Nagoya city in Japan, we show the proposed approach can detect such high-risk roads. Furthermore, we show the detected roads spatially change according to the people flow in ordinary situations, evacuation behavior, and disaster occurrence time.
Sasabe et al. (Wed,) studied this question.