A Bayesian Network Model for Seismic Risk Analysis.

Abstract

Earthquakes are one of the most unpredictable natural disasters. A series of secondary and derived disaster events may occur afterwards and lead to even more consequences. In such situations, a seismic risk analysis that takes into account secondary and derived disaster events is vital in reducing the risks of such disasters. The absence of a holistic seismic risk analysis model-one that takes into account the derived disaster events-may mean that the serious consequences of the disaster chains set off by earthquakes are neglected. This article proposes a comprehensive seismic risk analysis that enables a better understanding of seismic disaster chains and rescue scenarios. The approach is based on a Bayesian network constructed using scenario-based methods. The final network structure is achieved by learning parameters. To determine the critical secondary disasters and the key emergency-response measures, probability adaptation and updating using the Bayesian model were performed. The practical application of the model is illustrated using the Wenchuan earthquake and the Jiuzhaigou earthquake in China. The two examples show that the model can be used to predict the potential effects of secondary disasters and the final seismic losses. The results of the model can help decisionmakers gain a comprehensive understanding of seismic risk and implement practical emergency-rescue measures to reduce risk and losses.