Including Dynamics in a Network Based Stochastic Multihazard Model

We outline a conceptual approach to forecast multihazard risk from a cascade of natural hazards events. Network models have been proposed for cascades of natural hazard events, for example storm, flooded river, breached stop banks, damaged infrastructure. These have generally not taken time into account, with the cascade of events effectively assumed to occur instantaneously. We extend the methodology to account for multiple temporal processes, often occurring on quite different time scales, and hence incoporating variable delays. Further, since state of the art physical models generally involve heavy computation, we advocate the use of computationally simple probability distributions to describe the dynamics and interaction of the hazard events in our proposed network model. All model components have estimable parameters, which permits application to specific situations. This enables a larger number of simulations of the model, ensuring greater accuracy of probabilistic model forecasts. The modelling approach takes into account the dynamic and evolving nature of the temporal processes. Thus, it may be possible to identify key elements of the system that are most vulnerable, develop strategies for mitigating risks, and examine restoration strategies.