Incorporating human behaviour in large-scale dynamic risk assessment and reduction modelling

Disasters due to natural hazards pose a large risk to societies all over the world. Although initiated by the occurrence of a hazard event, the actual risk to societies is largely determined by their exposure and vulnerability. With exposure being the representation of who and what is affected, and vulnerability the susceptibility of these values to the characteristics of the hazard. When trying to understand current and future disaster risk and designing disaster risk reduction strategies, it is important to untangle each of these three components of risk: hazard, exposure and vulnerability. An important step in achieving this is understanding the impact of human behaviour and local contextual factors on each of these components. To support future risk assessments, we have developed a spatially explicit and dynamic multi-hazard decision support system, UNHARMED. We will present a model application of UNHARMED to Melbourne, in which we have incorporated an activity-based land use model to simulate dynamic exposure profiles on a 100 m. grid. This type of approach enables large scale behavioural modelling as it works with population density levels per grid cell, an aggregation of social and economic differences into a limited number of societal groups, and behavioural profiles for each of those groups, making it possible to apply the approach to country and continental scales with data normally available at those scales. Behavioural characteristics of the various groups, together with local contextual factors, drive the residential choices, resulting in plausible spatial and temporal distributions of these societal groups, creating spatially explicit exposure information for each. Group specific vulnerability information is then used to assess the risk profiles, as well as the targeting of risk reduction options to meet the needs of the specific groups.