2021 Volume 130 Issue 2 Pages 213-238
In this study, evacuees' rosters and other relevant disaster records are used to geographically clarify the relationship between people's evacuation behavior following the Great East Japan Earthquake, social and demographic characteristics of evacuees, and housing damage caused by tsunami following the earthquake in Yamada Town, Iwate Prefecture. The trend of shelter entry and exit is analyzed on the basis of three scales. The first scale is the municipality unit; trends in the numbers of people entering shelters in Yamada Town are compared to those in other affected municipalities on the coast in Iwate Prefecture and regional differences in these trends are clarified. The second scale is the district or small area unit; regional differences in rates of entry to shelters in Yamada Town and factors affecting these differences are explained geographically. The third scale is the shelter level; regional differences in rates of entry to school shelters in small areas are analyzed in relation to evacuees at two elementary schools. Furthermore, regarding evacuees in a school shelter where a daily list is available, a logistic regression analysis is performed to explain evacuees' decisions on whether or not to stay in the shelter on the basis of variables such as gender, age, and family situation. The regression analysis for the first scale reveals that the decreasing trend in the number of evacuees in shelters in Yamada Town has been slow compared to those in other affected coastal municipalities. The study reveals that progress in the construction of emergency temporary housing is the factor with the greatest impact on an evacuee's decision to leave a shelter. Regarding the second scale, the geographical distribution of the number of shelter residents in Yamada Town is analyzed on the basis of the scale of a district and a small area. The analysis reveals that regional differences in shelter entry rate reflect social network, topographical features, and developmental process of the settlement. Regarding the third scale, the relationship between distance from the shelter and entry rate of affected households is analyzed by small area using the rosters of Yamada Minami Elementary School and Orikasa Elementary School evacuation shelters. A significant correlation is found between average road distance from Yamada Minami Elementary School and affected household entry rate, and it is observed that many residents were from areas located within 1 km from the school. On the other hand, no significant correlation is found between average road distance from Orikasa Elementary school and affected household entry rate. Regarding the Orikasa Elementary School shelter, almost daily entry and exit records could be obtained for the period from April 9, 2011 to August 3, 2011. Using this record, age and family composition of withdrawers during this period could be identified. A logistic regression analysis was performed with gender, age group, marital status, and family type as explanatory variables. With respect to the influence of a resident's age and family structure on his or her exit time, it is found that the withdrawal rate of residents aged 75 years or more was high until the beginning of May. Households that stayed in the shelter with 0 to 18-year-old juveniles tended to find it better to stay in the shelter until their children's schools reopened. After resumption of schools, at the beginning of April; however, they tended to leave in higher numbers. The study concludes that, although schools should support evacuees during the earlier period of an evacuation as large-scale evacuation shelters, various types of shelter should be developed and reorganized to suit the evacuees' needs in case the evacuation is extended for a long period.