Urban structure reinforces attitudes towards tsunami evacuation

Abstract

Evacuation is a critical life-saving action, especially in devastating natural hazards such as near-field tsunamis. However, the development of effective evacuation measures remains challenging to the extent that a successful example has been referred to as a 'miracle'. Here we show that urban structures have the potential to reinforce attitudes towards evacuation and significantly influence the success of tsunami evacuation. Agent-based evacuation simulations revealed that a distinctive root-like urban structure formed in ria coasts reinforces positive evacuation attitudes by effectively gathering evacuation flows and leads to higher evacuation rates compared to typical grid-like urban structures, which can explain the regional differences in the number of casualties in the 2011 Tohoku tsunami. Although a grid-like structure reinforces negative attitudes under low evacuation tendencies, with leading evacuees, its dense feature helps to propagate positive attitudes and drastically improve evacuation tendencies. These findings pave the way for making successful evacuation inevitable through harmonised urban and evacuation plannings.

Figure 1

Simulation domains and results of agent-based tsunami evacuation simulation. (a) Simulation domains presented in a large-scale map. Simulation domain for Ishinomaki is in a plain area located in the northern part of Sendai bay. Simulation domain for Kamaishi is located within Sanriku ria coast. (b) Simulation domain for Isinomaki in a small-scale map. Typical grid-like urban structure was formed in the plain area where a large extent of the tsunami inundation was experienced due to its low-lying flat feature. (c) Simulation domain for Kamaishi in a small-scale map. The distinctive root-like urban structure was naturally formed along with the topography of ria coast. (d, e) Snapshots of the evacuation simulation ($t=60s$) for Ishinomaki and Kamaishi, respectively. The evacuating agents are highlighted with larger spheres indicating their communication extent. More active evacuation behaviours (coloured in red) are found in Kamaishi compared to those in Ishinomaki where many non-active agents (coloured in green to yellow) can be observed. (f) Comparison of the evacuation completion ratios. Even with similar evacuation attitudes, Kamaishi achieved a higher evacuation completion. The box plots consist of the median line, box limits (inter-quartile range), whiskers (1.5 $\times$ inter-quartile range from upper or lower quartiles) and dots (fliers larger or smaller than whiskers). Each box plot includes $n=120$ simulation runs. Figures were made with the Generic Mapping Tools (https://www.generic-mapping-tools.org/), QGIS (https://qgis.org/) and ParaView (https://www.paraview.org/).

Figure 2

Different evacuation behaviours found in different cities. (a) Comparison of the initial attitude distributions of evacuated agents. Agents with relatively lower evacuation attitudes could evacuate in Kamaishi compared to those in Ishinomaki. (b) Comparisons of the timeseries of the evacuating cluster size. The number of agents in clusters of 10 or more agents was counted as large-groups. The time is normalised for comparison considering different evacuation distances in different cities. (c, d) Evolution of clusters of the agents with positive attitudes in Ishinomaki and Kamaishi, respectively. The centres of the clusters are plotted with the number of agents in the clusters. The size of the plots corresponds to the cluster size. While small-group evacuees with high evacuation attitudes are consistently observed in Ishinomaki, large-group evacuees with relatively lower evacuation attitudes are likely to be observed in Kamaishi. Maps were made with GeoPandas v.0.11.1 (https://geopandas.org/).

Figure 3

Attitude reinforcement mechanisms inherent in urban structures. (a) Trajectories of evacuated agents in Ishinomaki. The grid-like structure causes dispersed evacuation flows and makes them difficult to merge. (b) Evolution of clusters of the agents with negative attitudes in Ishinomaki. The centres of the clusters are plotted with the number of agents in the clusters. The size of the plots corresponds to the cluster size. The dense grid-like structure helps the growth of clusters with negative attitudes and results in the formation of large clusters with negative attitudes that negatively influence the evacuation of agents with positive attitudes. (c) Trajectories of evacuated agents in Kamaishi. The distinctive root-like structure effectively gathers evacuation flows and helps them form strong clusters with positive attitudes that cannot be influenced by clusters with negative attitudes. (d) Evolution of clusters of the agents with negative attitudes in Kamaishi. The centres of the clusters are plotted with the number of agents in the clusters. The size of the plots corresponds to the cluster size. The relatively sparse root-like structure limits the growth of clusters with negative attitudes. Maps were made with GeoPandas v.0.11.1 (https://geopandas.org/).

Figure 4

Significant improvement in evacuation tendencies with leading evacuees. (a) Evacuation completion ratios with different numbers of leading evacuees. Attitude reinforcement mechanism helps to effectively propagate positive attitudes from a small number of leading evacuees and improves the evacuation tendencies. (b) Ratios of behaviour change caused by negative attitudes during evacuation under varying numbers of leading evacuees. A low probability of being influenced by negative attitudes in Ishinomaki led to higher evacuation tendencies compared to that in Kamaishi. The box plots consist of the median line, box limits (inter-quartile range), whiskers (1.5 $\times$ inter-quartile range from upper or lower quartiles) and dots (fliers larger or smaller than whiskers). Each simulation case includes $n=120$ simulation runs.