An integrated assessment framework for quantifying and forecasting water-related multi-hazard risk

Abstract

Disaster risks induced by different kinds of hazard may emerge in any place where human activities or properties exist. Most human settlements are exposed to more than one hazard. The multi-hazard risk analysis that assesses the potential loss caused by multiple natural hazards can provide a valuable reference for regional land-use planning, disaster prevention and emergency management. Although an increasing number of risk assessment methods related to multi-hazard have been developed recently, three main challenges remain in the current practices: (1) the disparate characteristics of hazards increase the difficulty of their combination and comparison, (2) the dependence and interactions between different hazards are often neglected, and (3) the results of multi-hazard risk assessment are not quantitative to show the probability of disaster loss. This thesis aims to construct an integrated framework to quantify and forecast the risk of multiple water-related hazards including heavy rainfall, extreme river flow, and storm surge. The framework consists of the three typical components of disaster risk assessment containing hazard, vulnerability, and risk analysis and is applied in the Greater London and the Eden Catchment, UK. For hazard analysis, the joint probability and return period distributions are fitted for the three water-related hazards on the basis of dependence analysis and copula theory. A newly developed 2D hydrodynamic model is enhanced with auto Input-Output control and processing in a multi-GPU platform to drive numerous flood simulations. The frequency-inundation curves due to the combination of the three hazards are generated by connecting the joint return period functions and the results of flood simulations. The distribution of human life and properties in the research area are analysed and classified with different vulnerability curves that quantify the potential damage due to the severity of inundation. The component of risk analysis evaluates the probability of loss for human life or different types of properties according to the results from the hazard and vulnerability analysis. The risk assessment framework considers the interaction and dependence between the multiple hazards by using hydrodynamic modelling and joint probability analysis, respectively. It can produce fully quantitative results such as risk curves quantifying the probability of different damage states, and risk maps illustrating the expected loss in the research region. With the efficient 2D hydrodynamic model and the autoprocessing package, the framework is further applied to give flood and risk forecasting to the Eden Catchment by integrating with a numerical weather prediction model. The framework shows a quantitative approach of multi-hazard risk assessment. It also provides an integrated procedure of flood risk analysis and forecast in consideration of the dependence and interactions between different water sources. The methodology and the findings are of interest to insurance companies, regional planners, economists, disasterprevention authorities, and residents under the threat of flooding. The main source of uncertainties of the framework and the limitations are identified. Future work and further applications in other regions are recommended.