Identification of risk amplification effects in rainstorm disaster chains using coupled disaster system theory and Bayesian networks

: The risk amplification effects inherent to disaster chains are crucial for disaster prevention but have rarely been empirically substantiated because of the lack of suitable evaluation methods. Thus, this study proposes a coupled framework integrating disaster system theory and Bayesian networks ( CF-DST-BN ). By leveraging multivariate indicators spanning subhazards, bearing bodies, and damage, the framework captures three amplification effects from 359 rainstorm disaster chain events in the core area of the agricultural‒pastoral transition zone in northern China. (i) The cascading hazard transmission increases the expected secondary hazard-event count by 31.76%, primarily along 4 tertiary, 2 quaternary, and 1 quinary diffusion paths. (ii) The amplification effect of exposure manifests in two ways within three- to five-level chains: a superimposed effect resulting from multiple distinct secondary hazards acting upon the same type of exposed objects and a scope expansion effect occurring because multiple distinct secondary hazards affect different types of exposed objects, thereby expanding the overall scope of affected targets. (iii) Disaster damage is amplified via two mechanisms in higher-order chains: first, multiple secondary hazards produce compounded damage; second, a single secondary hazard expands the damage scope. Unexpectedly, the risk diffusion paths of rainstorm-induced disaster chains become irreversible as the number of layers increases, which is a process that is shaped primarily by the disaster-prone environment. This hybrid approach offers dual utility, as it captures both risk amplification effects and decodes the causal relationships that underlie risk propagation, thus providing practical guidance for policymakers and researchers involved in disaster prevention and mitigation.