A dynamic estimation of casualties from an earthquake based on a time-use survey: applying HAZUS-MH software to Ulsan, Korea

Abstract

One essential piece of information used to estimate earthquake casualties is the spatiotemporal distribution of the population. HAZUS-MH is a loss-estimation model nested in a geographic information system that has been developed and freely distributed by the US Federal Emergency Management Agency to estimate loss and damage due to natural hazards. Because inventories of behavioral patterns based on empirical data have not been developed and have relied solely on behavioral information about the US population, they may not provide reliable and meaningful casualty estimations in an international setting. To estimate accurate hourly exposure for an earthquake using empirical data, the study uses a daily time-use survey conducted by Statistics of Korea. The survey contains behavioral data from approximately 21,000 respondents, collected over 24 h. By combining structural damage estimates, calculated by HAZUS-MH, with spatiotemporal behavior patterns estimated using a daily time-behavioral survey in Korea, the study shows the estimated casualties related to a given earthquake scenario in the Ulsan area. The simulation results show that the greatest number of estimated casualties occurred when the earthquake struck between 2 a.m., and 4 a.m. The fewest casualties were expected when the earthquake occurred between 12 p.m., and 2 p.m. The proportion of the indoor population in damaged buildings and the spatial distribution of occupancy type was one of the important factors. During the daytime, casualties increased where the nonresidential occupancy was concentrated. At nighttime, a higher number of casualties were estimated in residential occupancy.