Citation

Nichols JM, Beavers JE. Nat. Hazards Rev. 2008; 9(4): 179-189.

Copyright

(Copyright © 2008, American Society of Civil Engineers)

DOI

unavailable

PMID

unavailable

Abstract

A method to estimate the likely fatalities in earthquakes in the twenty-first century is a statistical analysis of the data for fatalities in earthquakes from the last two millennia. The National Oceanic and Atmospheric Administration (NOAA) collects and collates earthquake fatality data for the world. The NOAA database has records dating to 186 Before Common Era. The fatality information is incomplete, like all such historically collected data sets. A standard statistical analysis of the known fatality data provides earthquake fatality models for the twentieth century and the last two millennia. The analysis uses a generalized Poissonian distribution to provide a mathematical mapping for each fatality data set. The generalized Poissonian distribution provides a method that can allow for over- and underdispersion of the data not accommodated by a standard Poissonian model. The results show a higher mean for the two millennia data set, which is expected due to the undercounting of the smaller fatal events prior to the twentieth century, so, for example, someone dying in an earthquake in Papua, New Guinea, in the twelfth century will not be in the NOAA database. The generalized Poissonian model is consistent for the two millennia and the 20th century. The generalized Poissonian model for the twentieth century appears to represent an acceptable statistical method to estimate deaths in the current century. This analysis of fatality distributions in earthquakes points to a likely peak death toll for this century in the range of 600,000 to one million people in a single urban center, with this event most likely occurring in an intraplate region rather than an interplate region of the world.