Estimating human losses in earthquake models: A discussion

Abstract

Several studies and methodologies have been developed in recent years to model the number of victims and injuries caused by natural disasters such as earthquakes. Unfortunately, models and simulations developed up to now show substantial variability in the numbers of victims when compared with real values, because they do not consider a multi-parameter analysis including variables such as seismic intensity, degrees of building damage, percent of occupancy at the time of the event, individual behaviour (age, gender, mobility within the house during the shaking, etc.) or emergency response (effectiveness in response). When dealing with this topic we should separate the situation of estimation of human losses for emergency preparedness from the estimation right after a given event. In this paper the second of these issues will be analysed. People’s reactions prior to and during the shaking together with the building behaviour cause great differences in the number of deaths and injuries for a given earthquake. The European Macroseismic Scale (EMS-98) provides five grades for damage classification from “Negligible to slight” damage (D1) to “Collapse” (D5). While D5 class includes total or near total collapse of the buildings, we propose a class D5+ to represent “totally collapsed” structures separately from “almost collapsed” so as to establish a direct relation between damage grade and death rate. Data from a few events in Portugal and Italy illustrate the difficulties in estimating human losses.