Developing a new fatality model for Iran's earthquakes using fuzzy regression analysis

Abstract

This paper presents the results of a study carried out to develop an empirical fatality model based on data of fatal and non-fatal earthquakes occurred in Iran since 1962. Two reliable databases (RAISE-DAT and REXPO-DAT) which incorporate the number of deaths and the affected population in the past earthquakes of Iran were compiled to make correlation between fatality rate and shaking related parameters. The uncertainties of these parameters (i.e., the number of fatalities in the past earthquakes and the ground motion values provided by the ShakeMap) were considered in the analysis, using the fuzzy regression and Monte Carlo simulation tools. In addition, the appropriateness of three distinct parameters including PGA, PGV and MMI were examined to be employed in the fatality model. The results show that PGA represents the best estimation of fatality (especially for strong earthquakes), while the other parameters (PGV and MMI) underestimate the fatality rate. As a case study, the model was applied in the Ahar-Varzaghan (6.5 Mw) 2012, Iran earthquake. The result shows a good agreement between the actual number of deaths and the estimated fatality. The proposed model can provide an initial assessment of death toll after destructive earthquakes. This is important for organizing rescue and relief activities during golden hours' after an earthquake. Accordingly, the model is used for rapid loss assessment system of Iran's earthquakes, called RAISE. The model can also be used for simulation potential impacts of earthquakes; therefore, it can be employed for seismic risk mitigation planning in Iran.