WORLDWIDE EARTHQUAKE LOSS ESTIMATES, IN REAL-TIME AND SCENARIO MODE

Within 20 minutes, on average, after M6+ earthquakes worldwide, we distribute emails estimating the number of fatalities, the number of injured and the average damage to buildings in all settlements affected by strong shaking. In addition, we telephone rescue agencies, in case of likely serious disasters, and we place the following items on our website. (1) A map showing the general location of the epicenter. (2) A detailed map of the epicentral vicinity, showing the average degree of damage of settlements by a color code on a scale of six. (3) A list of affected settlements containing estimated intensity of shaking, population, numbers of casualties and estimated percentage of buildings in five damage classes. (4) A travel time map for a tsunami, should one have been generated. Sources of errors in communicating an approximately correct loss assessment to rescuers and authorities include: (A) Insistence of local civil defense officials to rely on underestimated loss reports from the periphery of the disaster, as representative of the sum of the losses, (B) inaccurate earthquake depth or location, (C) lack of information on site amplification, (D) incorrect assumptions about the fragility of the built environment, (E) outdated population numbers, (F) lack of information on local attenuation, (G) the simplification of the earthquake source as a point. Nevertheless, our results over three years show that the accuracy of our alerts is satisfactory for rescue agencies, which do not need exact numbers of casualties. They simply need to know whether or not a serious disaster has taken place that requires international response. We also estimate losses for scenario earthquakes. In these cases, likely positions and source parameters are assumed for earthquakes in populated seismogenic regions. From the amount and distribution of losses in these scenarios, strategies for protecting the population can be designed. In March 2005, we published an estimate for losses to be expected in future great Himalayan earthquakes. For the scenario called “Kashmir”, we estimated that 67,000 to 137,000 fatalities should be expected. This turned out to be an accurate forecast of the Kashmir disaster of 8 October 2005, in which about 76,000 people perished. Thirty one minutes after the occurrence of this M7.6 earthquake we called the Swiss rescue team to tell them that thousands of people must expected to be dead in Kashmir. After the Bam M6.6 earthquake of 26 December 2003, we also called the duty person of the Swiss rescue service to tell him that we believed that more than a thousand people could have perished, but “that the losses strongly depended on the hypocentral distance from a city called Bam”, a city we had not heard of until that day. It turned out that the earthquake was much closer and much shallower than the published hypocenter indicated. For that reason we underestimated the number of casualties in Bam, but classified the event as a major disaster, a message sufficient for the Swiss rescue team to offer assistance to Iran immediately.