Development of population models for risk assessment in New Zealand

Abstract

Population models are key components for natural hazard risk modelling of human impacts, such as evacuation, casualty estimation or household relocation. Existing models designed for this purpose in New Zealand lack the required details for accurate risk modelling. We propose the development of three population models. The first is a model of households containing demographics of interest. The second is a network model describing the movement of individuals from source (usual residence) to other locations (such as work or school). The third is a spatial and temporal population distribution model for locating individuals, such as within office buildings, commuting, or in parks. The development of a pilot household population model for the Wellington region is described. Census data tables were used to create a synthetic population of households, including the attributes of tenure, composition, income and household size. Households were distributed to residential building footprints. The proposed development of a spatial and temporal distribution model is also described, using a variety of data sources such as OpenStreetMap and occupancy rates. The proposed network model has not been explored in detail in this report. The pilot household population model produced reasonable results when compared with the census data tables. The model is likely to be sufficiently accurate for many forms of risk modelling, depending on the variables and analysis required. Future improvements to the model may be made by utilising census microdata rather than aggregated census data tables.