Simulating Multihazard Interactions Using Higher-Order Network Analysis

Abstract

Our built environment is often subjected to multiple (natural) hazards’ concurrent and/or sequential impacts. Modeling multihazard interactions (i.e., possible interrelationships between hazard events with their associated frequencies and severities and resulting impacts on a specific location or region of interest) and simulating multihazard scenarios (or event sets; i.e., realizations of possible multihazard interactions) to design and assess civil infrastructure is crucial. This paper presents a probabilistic framework for simulating multihazard scenarios, aiding in civil infrastructure design and risk assessment. Specifically, the proposed approach leverages temporal hypergraphs for characterizing the hazard interactions. Temporal hypergraphs are a higher dimensional generalization of a graph that can capture multi-way connections (i.e., not just pairwise) that can change over time. The temporal attributes of the hypergraphs are defined using a series of occurrence models and simulation-based approaches to generate the arrival times and features (e.g., event characteristics and local intensities) of all considered hazards over a defined space-time interval. A sliding window procedure is also proposed to support decision-makers and other end users in identifying critical multihazard scenario windows that can significantly impact the resilience of the built environment. The framework’s applicability is demonstrated using an illustrative example of a hypothetical region susceptible to earthquake-induced ground shaking, liquefaction, tsunami, and riverine flooding. The proposed framework can help decision-makers design and test efficient disaster risk mitigation and management policies. Furthermore, the proposed multihazard scenario modeling framework can be combined with vulnerability/impact assessment methodologies to quantify the multihazard risk of exposed infrastructure and communities.