Emergency Management and Underserved Communities: Using Big Data to Improve Emergency Management Preparedness, Response and Resilience; Disaster Response: A Digital Humanitarian-Based Approach to Reaching Vulnerable Groups
Snavely, Conner, School of Engineering and Applied Science, University of Virginia
Shafiee-Jood, Majid, Department of Engineering Systems and Environment, University of Virginia
Baritaud, Catherine, Department of Engineering and Society, University of Virginia
An increase in the earth’s temperature has led to a propensity for natural disasters, augmenting both frequency and intensity. Research has shown that natural disasters have a tendency to disproportionately affect vulnerable populations. As such, it is important to understand hurricane evacuation patterns of various demographics and understand the efficacy of mandatory evacuation orders. Cell phone mobility data and online social media data provide a promising means to understand human behavior at high spatial-temporal resolution. The technical project analyzes mobility data at the census block group level, coupled with evacuation order data and hurricane forecast information from Hurricane Florence in 2018 to gain insight into population behaviors prior to disaster. Moreover, the STS research uses Actor Network Theory (ANT) to illustrate the socio-technical interconnectedness of Virginia’s disaster response network. ANT exposes how vulnerable groups are oftentimes overlooked, while revealing the potential of using social media data to best reach everyone. Together, the projects provide many benefits to emergency management. The technical work institutes a unique approach to understanding mobility patterns of the public, while the STS research proposes how to locate and provide for those most affected.
The Virginia Department of Emergency Management (VDEM) currently has little understanding of evacuation patterns of the general public and how residents respond to evacuation orders. Historical efforts to quantify public evacuation patterns have employed an assortment of datasets ranging from traffic counters to community survey responses. However, they provide a low-resolution approximation of human behavior. This research uses anonymized cell-phone mobility data to engage in statistical analysis of evacuation patterns. Analyzing this data, accompanied by a first-of-its-kind evacuation order database provides an analysis of the differential response to evacuation orders in VA, NC, and SC.
By analyzing hurricane forecast information, it was revealed that evacuation orders are typically enacted five days prior to projected landfall. However, mobility analysis revealed that most residents tend leave their residence roughly three days prior to landfall. It was established that average evacuation index before and after an order is issued is statistically significant, suggesting that evacuation orders, to an extent, influence decision making. Several linear models were developed using census data from the American Community Survey to explore how demographic factors play a role in evacuation rates. The percentage of renter occupied units, age, education level, and income are all significant predictors of evacuation index for a given census block group. By integrating census data, mobility data, and evacuation orders, one can partially attribute higher evacuation response rates to demographic factors. This research provides an understanding of at-risk communities during an evacuation emergency.
It is the goal of the STS research to understand how social media data can be used to reach the most vulnerable groups at the local level. John Law and Michael Callon’s ANT brings to light a rift between global and local networks during emergency response in Virginia. Fortunately, recent advances in technology have provided opportunities for those around the world to volunteer digitally to assist in response and recovery efforts. The research proposes how these groups, digital humanitarians, can provide a means to bridge the gap between said networks.
It was found that emergency managers do not tailor policies and response procedures to specific social and economic groups in Hampton Roads, Virginia. Professional responders have begun to understand that benefits of using online data to collect intelligence during an emergency, but have faced difficulties in efficiently using this data. Digital humanitarians, engaging in citizen science, can be a potential solution to this rift and the lack of specific consideration for vulnerable populations. Working alongside VDEM, they have the ability to help identify and locate at-risk groups to best provide for residents most affected.
Ultimately, as hurricanes become more frequent and intense, it is more important than ever to understand how the public behaves and how to address vulnerable populations. Certain demographic groups tend to evacuate more than others, so it is necessary to ascertain how those less likely to evacuate are accounted for by the disaster response network.
BS (Bachelor of Science)
Actor Network Theory, Hurricane evacuation, Vulnerable groups, Social media, Digital humanitarians, Mobility data
School of Engineering and Applied Science
Bachelor of Science in Systems Engineering
Technical Advisor: Majid Shafiee-Jood
STS Advisor: Catherine Baritaud
Technical Team Members: Zachery Key, Andrea Parrish
All rights reserved (no additional license for public reuse)