Social Vulnerability, Locational Vulnerability, and Uncertainty in Wildfire Risk Index Construction

Over the past two decades, methods to measure and account for social vulnerability, or the ability of a community to prepare for, cope with, and recover from a natural disaster, have rapidly proliferated. Recognizing the multifaceted nature of vulnerability, researchers and practitioners now seek to combine social with locational (physical) vulnerability into singular indexes designed to identify areas of high risk. Central to such efforts are common demographic data sets (e.g., those provided by the U.S. Census) and spatially referenced environmental (locational) data sets specific to a hazard, such as historic flood plains or remotely sensed indicators of live fuel moisture for wildfires. There are inherent uncertainties, however, associated with both social and locational data. Examples are myriad and could include issues of scale, precision, and sampling. Although more thoroughly addressed in the physical geography domain, these uncertainties are largely ignored in many of the common constructions of social vulnerability indexes that dominate the literature. In this article, we make two contributions. First, from a theoretical perspective, we draw on the traditional error propagation framing and methods from physical geography and apply these to a human geography setting. Second, we use this method to propagate data uncertainties through a wildfire risk index and show how this error propagation destabilizes conclusions and common map interpretations, even in a simplified application. Our results suggest that ignoring the uncertainty could lead to misallocated resources for prevention, mitigation, and recovery efforts. We highlight the difficulties of index construction and make recommendations for different possible approaches.