Earthquake loss alerts to save victims

Large earthquakes are unavoidable because globally the plate motions accumulate stress, which leads to ruptures of the crustal rocks hundreds of kilometers long. In developed areas, this brings buildings to collapse, which injures and kills occupants. Potential rescuers are never well informed about the extent of an earthquake disaster because communication along the rupture is interrupted. We have documented that the underestimate of fatality numbers lasts for at least the crucial first few days, often for weeks. For earthquakes that cause thousands of casualties, the extent of underestimation is usually an order of magnitude. To reduce this uncertainty of whether help is required and how much, we have assembled a data set and constructed algorithms to estimate the number of fatalities and injured within  an hour of any earthquake worldwide in the computer tool QLARM. Our estimates of the population and the makeup of the built environment comes from government and internet sources. For large earthquakes, the hypocenter and magnitude is calculated and distributed by the GEOFON group at the Geoforschungszentrum (GFZ) in Potsdam, Germany and the Geological Survey (USGS) in Golden, USA within 6 to 10 minutes. Based on this information, the QLARM operator responds with an estimate of the number of casualties within 30 minutes of the earthquake, on average. These estimates are available to anyone by email alerts without charge. Since 2003, the QLARM operator has issued more than 1,000 casualty alerts at any time of the day pro bono. The USGS delivers a similar service called PAGER, which is based on different data sets and algorithms. The two loss estimates are usually close, which should give governments and news organizations confidence that these alerts are to be taken seriously. The QLARM research group also publishes research results, estimating the likely numbers of future casualties in repeats of historical large earthquakes. In such efforts the QLARM group has discovered that, contrary to the general assumption, the rural population suffers more by an order of magnitude under very large earthquakes than the urban population. It is also clear that the poorer segment of the population in cities and countryside suffer more than the affluent members of society because the former’s houses are weaker and collapse more readily. To be even more useful, a worldwide data set of hospitals and schools is needed in order to provide first responders with locations and likely damage to these critical facilities. Crucially, reliable school location data would enable first responders to focus rescue efforts on schoolchildren who die beneath the rubble of their schools in the hundreds to thousands in large earthquakes. Unfortunately, such data are not available from official sources in most developing countries, and we are not aware of good alternatives. The data on schools in open data platforms such as OpenStreetMap is sporadic. UNICEF runs a global school mapping initiative, but we have been unable to obtain their assistance to date.