The severity of earthquake events -- statistical analysis and classification

Earthquake events are natural disasters that can pose a threat to people's safety as well as their homes and possessions. In this paper, the severity level of earthquake disasters is addressed using the US National Oceanic and Atmospheric Administration (NOAA) database. A total of 5841 earthquake incidents are recorded that happened between 2150 BCE and 2015 CE. Few studies have done a comprehensive statistical analysis of the consequences of earthquakes. To address this gap, and after determining the probability distribution function of the number of fatalities, we evaluate the distribution of earthquakes with extreme fatalities to determine the severity levels according to the fatality-based disaster scale introduced by Wirasinghe, Caldera, Durage, and Ruwanpura [(2013). Preliminary analysis and classification of natural disasters . Proceedings of the ninth annual conference of the International Institute for Infrastructure, Renewal and Reconstruction (IIIRR), Queensland University of Technology, Brisbane, Australia, July 2013, Section B1.2, p. 11]. To this end, three different methods of determining the extreme events are considered: peak over threshold, R th order, and event-based and location-based block maxima. Moreover, a comprehensive collinearity analysis is performed to investigate any correlation and linear dependency between the earthquake parameters (magnitude, intensity, and focal depth) and the consequences in terms of earthquake fatalities. The severity classification based on block maxima has more detailed severity classes; hence, it is superior to the other two methods. For block maxima, the probability of a lower level disaster (Emergency to Catastrophe Type 1) being the extreme disaster is higher for the location (country)-based data set compared to the event-based worldwide data set, while the probability of a higher level disaster (Catastrophe Type 2 and above) being the extreme disaster is lower. These probabilities are to be expected because a single country, even over the full time period, is less likely to have a massive disaster compared to the world when a large number of extreme events, in this case 100, are considered.