on Rapid Assessment of Earthquake Death Toll Based on Neural Network

 101 post-1960 fatal earthquakes in China mainland were selected to build a BP neural network model for rapid earthquake fatality estimation. In this model, event time, magnitude, epicenter intensity, population density and the event region were selected as input vectors. While, the fatality was assigned as output vector. An empirical test shows that the BP neural network model can give a more accurate assessment, with a good match to the real death toll. This approach can be applied in rapid earthquake losses estimation, and provide scientific basis for emergency rescue and government decision-making.