Estimating the number of casualties in earthquakes from early field reports and improving the estimate with time

We investigate the time dependence of the number of deaths reported through Internet after earthquakes and/or earthquake-generated tsunami. An approximate relation N(t) = N ₀ 1 − exp(−αt)] is used to describe such temporal variation, in which N(t) is the number of deaths reported at time t, N ₀ is the final number of deaths, and α is the coefficient reflecting the rescue process. We considered 12 earthquake cases since 2001 using the information from the web, which shows that the N-t relation approximates the data, and the logarithm of α is reversely proportional to the magnitude of earthquake, albeit with significant uncertainties. Quick and rough estimate of the final death toll can be made using this simple and approximate relation, with the empirical α-M relation as a reference. For the 12 cases under consideration, quick and rough estimate of fatalities can be obtained 2 days after the earthquake, fitting the real situation in the order of magnitudes. Although being very rough, this estimate can assist the emergency decision-making and can be revised as time lapses. When more and more data becomes available, curve fitting can provide both N ₀ and α at the same time. The method is tested against the data of the recent Yushu earthquake on April 14, 2010.