MODELING DEMOGRAPHIC SENSITIVITY TO TSUNAMI HAZARDS IN THE PACIFIC NORTHWEST USING GIS-ENABLED FACTOR ANALYSIS

Tsunamis generated by Cascadia subduction zone (CSZ) earthquakes pose significant threats to coastal communities in the U.S. Pacific Northwest. A regional inventory of community exposure to Cascadia-related tsunamis on the Oregon coast indicates that tens of thousands of people live, work, and play in tsunami-prone areas. Impacts of future tsunamis to individuals and communities in the region will likely vary due to pre-event demographic differences. Given the catastrophic potential and quick arrival times of tsunamis generated by local CSZ earthquakes, emergency managers must understand who is vulnerable to tsunamis so that they can prepare realistic and effective evacuation and response procedures.

Assessing vulnerability through an inventory of demographic attributes will help managers identify isolated issues (for example, an elderly population needing assistance to evacuate quickly) but it fails to address how multiple demographic characteristics of an individual or neighborhood interact and likely amplify each other. To describe the multivariate nature of individuals living in areas prone to CSZ-related tsunami inundation, we adjust the Social Vulnerability Index (SoVI) model based on factor analysis to operate at the census-block level of geography and focus on community-level comparisons along the Oregon coast. The SoVI model is based on the use of principal component analysis, one of the most common multivariate factorial approaches, to reduce a large number of census variables into a smaller set of multivariate components. Variable members of each model component exhibit similar variation across the study area and each component explains a certain amount of the total variance of the entire dataset. This model is then merged with geographic-information-system (GIS) tools to develop block-level maps of demographic sensitivity to tsunamis.

A principal components analysis of populated census blocks in the Oregon tsunami-hazard zone results in 11 broad components that explain 64.6 percent of the variance. The model components that represent the highest percentages of the database variance relate to demographic attributes of wealth and education, age and tenancy, and type of employment and housing type. The number of residents from census blocks in tsunami-prone areas considered to have higher social vulnerability varies considerably among 26 Oregon cities along the Oregon coast and most are concentrated in four cities and two unincorporated areas. Model results suggest there is no apparent relationship between the number of residents considered to have high social vulnerability and the percentage they represent of the total number of residents in the tsunami-hazard zone. Variations in the number of residents from census blocks considered to have higher social vulnerability in each city do not correlate with the number of residents or city assets in tsunami-prone areas. Modeling methods presented here provide emergency managers with the means to depart from one-size-fits-all mitigation strategies that inadequately address differences in social context and, instead, to develop strategies tailored to local conditions and needs.