Emergency Management and Underserved Communities: Using Big Data to Improve Emergency Management Preparedness, Response and Resilience

Abstract

In anticipation of high impact weather events such as hurricanes, wildfires, and flash floods, public officials need to make life saving and time sensitive decisions under uncertainty. For example, when a hurricane is forming in the Atlantic, public officials need to decide whether and when to issue an evacuation order. However, there is always a large risk in issuing an order early because of the uncertain nature of weather forecasting. Besides the preparation costs, the public could lose trust in officials and forecast information. Previous studies have identified a number of sociodemographic factors contributing to individuals’ likelihood to evacuate. These research efforts have proven that the probability of evacuation shares a strong positive correlation with both economic and physical mobility, meaning older populations, low-income populations or those with larger families are less likely to evacuate. While these efforts have provided policy makers with valuable insight to provide for these low evacuation populations, there has been very little analysis of the impact of evacuation orders on constituents’ evacuation mobility patterns. To bridge the gap in literature, we investigate the relationship between evacuation policy and observed evacuation patterns during Hurricane Florence (2018). Specifically, we evaluate the evacuation index at the census block group level of communities in Virginia encountering a false positive compared to those in South Carolina experiencing a true positive. By overlaying evacuation order data with cellular mobility data and forecast information from the National Hurricane Center, we aim to capture interactions between policy measures and socioeconomic factors to assess their relationship with evacuation behavior.