Using Bayesian networks in analyzing powerful earthquake disaster chains

Abstract

Substantial economic losses, building damage, and loss of life have been caused by secondary disasters that result from strong earthquakes. Earthquake disaster chains occur when secondary disasters take place in sequence. In this paper, we summarize 23 common earthquake disaster chains, whose structures include the serial, parallel, and parallel–serial (dendroid disaster chain) types. Evaluating the probability of powerful earthquake disaster chains is urgently needed for effective disaster prediction and emergency management. To this end, we introduce Bayesian networks (BNs) to assess powerful earthquake disaster chains. The structural graph of a powerful earthquake disaster chain is presented, and the proposed BN modeling method is provided and discussed. BN model of the earthquake–landslides–barrier lakes–floods disaster chain is established. The use of BN shows that such a model enables the effective analysis of earthquake disaster chains. Probability inference reveals that population density, loose debris volume, flooded areas, and landslide dam stability are the most critical links that lead to loss of life in earthquake disaster chains.