A Model for the Spatio-temporal Distribution of Population using Country-Wide Statistical Data and Its Application to the Estimation of Human Exposure to Disasters

摘要

Extent of fatality due to natural disaster depends largely on spatial distribution of population at the moment disaster occurs. In this study, a computational model for estimating day-long spatio-temporal distribution of commuters (workers and students) was developed using a number of country-wide statistical data of Japan such as "population census", "survey on time use and leisure activities", and "economic census". The model estimates behavior of individual commuters by considering their attributes including gender, occupation, place of work (or school), and place of residence. As a case study, the proposed model was applied to the Keihanshin (Kyoto-Osaka-Kobe) metropolitan area, one of the largest metropolitan areas in Japan. Estimated maximum number of commuters unable to return home in an expecting earthquake scenario was between 1.1 and 1.9 million depending on the assumptions on traffic disruption following disaster and the maximum walkable distance of each commuter.