Review on research methods of disaster loss accumulation and amplification of disaster chains

In recent years, the frequent catastrophic disasters have caused great losses of human lives and properties in the world. This indicates that disaster losses may be accumulated and amplified through disaster chains,in which one disaster triggers another, and so forth. The losses are much heavier in disaster chains than in a single disaster. Disaster chain is a typical complex form of regional disaster system. Understanding the amplification mechanisms of disaster chains is very important in catastrophe risk governance. This review first focuses on the concepts and understandings of disaster chain and summarizes a comprehensive definition from the geographic perspective through literature research. It is found that disaster chains have two common features, including the casualty relationship between different disasters and the spatial and temporal expansion of disaster losses. Only under a geographical framework, a sound and complete understanding of the disaster chain concept may be possible. A regional disaster system includes the environment, hazards, and exposure units and the complex spatial and temporal interactions of these elements. Second, considering the different research philosophies, five method categories in recently disaster chain research are summarized and analyzed. The five categories include empirical statistics methods, probabilistic models, complex network models, disaster system simulations, and multidisciplinary theories. The appropriateness and disadvantages of each category of methods are discussed with respect to their utility in describing disaster chain loss accumulation and amplification. Empirical statistics methods are a classical one that often use weighted average of a series of indicators. They have great advantage in taking into consideration various geographic factors and the modeling process is simple. But these methods cannot reveal the disaster chain evolution and processes. Probabilistic models can generate an overview of all possible events after a disaster have happened, as well as calculate the conditional probability. But they have the same problem as the statistics models. Disaster chain is a typical complex network. So the complex network theory may be used to describe the evolution of disaster chain networks. Such method should consider the spatial and temporal features of the disaster chain components in order to make the result more precise. Simulation methods are a promising one that can support the understanding of disaster chains dynamics, as well as the mechanism of the accumulation and amplification of disaster chain losses. However, simulation precision should be improved by including the spatial and temporal features of disaster chains in the future. Two types of important disaster chains, the seismic and typhoon disaster chains were used as examples to show the practical application of these methods. Finally, this review shows that the main trend of disaster chain research is to build and improve the dynamical model of disaster chain loss accumulation and amplification processes. The key is to connect all the factors spatially and temporally in a disaster chain system. It is necessary to transform the research approaches from "static-descriptive-explanatory" to "dynamic-process oriented-simulation" in order to understand the complexity of a disaster system.