Earthquake population loss estimation using spatial modelling and survey data: The Bam earthquake, 2003, Iran

Abstract

This paper contributes to ongoing global efforts to estimate probable earthquake population losses. Several models have been developed to estimate the population losses such as the number of killed, injured and hospitalized, injured and non-hospitalized, trapped, and non-injured people. However, there was lack of research on the estimation of the number of trapped people in Iranian context which is the main reason in conducting of this study. Data were collected based on a two-stage cluster sampling method through an interview survey from 396 people who experienced the Bam earthquake in Iran, 2003 regarding building damage and population losses in each household. The cross-tabulation analysis was performed to calculate the probability of each class of population losses, and point density analysis in GIS was applied to identify the spatial distribution of population losses throughout the study area. Moreover, a Wilcoxon Signed Rank Test, Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Fuzzy Inference System analyses were carried out to measure the validity of the statistical and spatial results in comparison with the Actual Earthquake (AE) data. These analyses clarified that survey data significantly improved estimation results of population losses against KHM model. Furthermore, the estimation of “the number of trapped people” according to data gathered in Iran is promising, as it can play a significant role in disaster response operations. This factor can make a key contribution to the efficiency of allocation of limited resources, particularly Urban Search and Rescue (USAR) operational teams.