Researchers Meeting Abstracts

As the COVID-19 pandemic spreads rapidly across the globe, healthcare professionals are working tirelessly to care for patients despite shortages in personal protective equipment and medical supplies. Current research rarely examines how healthcare workers balance patient care, personal risk, and family responsibilities during a pandemic. To address this gap, we are conducting an qualitative open-ended survey of healthcare workers across four cities in the U.S. and Canada: Los Angeles, CA, New York, NY, Montreal, QC, and Vancouver, BC. We aim to identify specific challenges that healthcare workers are facing to inform emergency preparedness planning for potential wave(s) of COVID-19. This knowledge can ultimately support the improvement of healthcare infrastructure across international borders by collectively enhancing healthcare workers’ capacity to prepare for and respond to future public health emergencies.

During major disasters, such as hurricanes, the construction industry experiences several challenges in responding and recovering from an approaching hurricane. Recent studies suggest that social media plays an important role in crisis communications. This study has revealed communication patterns on Twitter for the construction industry during Hurricane Michael in 2018. Around 275 million tweets were collected using keywords that are associated with a hurricane, construction, and the post-disaster context. Several machine learning techniques, including topic model and dynamic topic model, and data mining approaches have been used to analyze user sentiments and concerns over time and space. The crisis communication patterns, such as planned and nighttime construction, obtained from this study will help construction stakeholders and policy-makers make informed strategic decisions that are community-centered and facilitate effective response and recovery in major disasters. This study potentially can reveal latent challenges that delay the reconstruction of infrastructure and provide opportunities to improve recovery time.

Social vulnerability plays a significant role in differential disaster outcomes and impacts and influences the pace and progression of post-disaster recovery processes. Various disaster assistance programs are designed to support individual survivors, states, local governments, non-governmental entities, and businesses depending on the nature and magnitude of disasters impacts. One such program is the Individuals and Households Program (IHP) provided by the Federal Emergency Management Agency (FEMA) after Major Disaster Declarations. The program supports homeowners and renters for rental housing, repair, replacement, and permanent construction of disaster-damaged houses. It also provides assistance to disaster victims with disaster-related severe needs and necessary expenses. To date, a very limited exploration into connections between social vulnerability and access to IHP disaster assistance funds has occurred. In this study, IHP assistance provided to homeowners and renters between 2010 and 2018 at the county level were analyzed and the relationship between the distribution of funds and the level of social vulnerability—calculated based on Social Vulnerability Index (SoVI)—were explored. The hypothesis is that issues of distributive justice may be influencing the pace and progression of recovery in places where higher social vulnerability is present, and that receipt of disaster funds is linked with underlying socio-economic conditions.

Realistic estimates of damage and loss on a regional scale relies on a rich and complete database of fragility functions that are relevant to the hazards and infrastructure of a community or region. A comprehensive taxonomy and database of region-specific fragility functions appropriate for Oregon were solicited by a research cooperative of Oregon-based lifeline providers. Objectives of this research were to (1) identify available and missing fragility functions suitable for Oregon lifelines, (2) evaluate the quality of the collected fragility functions, and (3) provide recommendations where refinement is needed for future studies. The Python-driven database was structured using a hierarchy of infrastructure systems —electric power, water and wastewater, and transportation systems—hazard intensity measure —earthquake and tsunami— and fragility function attribute—fragility class, probabilistic distribution, damage state, and metadata. A survey was conducted to (a) identify relevant hazards, infrastructure, and multi-hazard scenarios of interest to Oregon stakeholders, (b) identify potential in-house and proprietary fragility functions, and (c) develop metrics to evaluate the quality of the collected fragility functions. This presentation will summarize the results of the study and provide a vision for future work.

The majority of COVID19 responses have been focused on individual medical interventions including social distancing and hand washing. Yet beyond individualized policies, regional and institutional factors such as social ties, health care capacity, and mobility may be core drivers in stopping (or accelerating) infection. Some cities—such as San Francisco, New York, and Boston—moved rapidly to close schools and restaurants while others—such as Dallas, and Orlando—did not. This presentation will discuss a multilevel investigation with individual level data from four neighborhoods in Boston and New York City along with city-level data from 18 cities.

This study is the first to investigate the role of social media in enabling households to reduce disaster losses. The historic 2011 Bangkok flood is utilized as a case study. This event was one of the first major disasters to affect an urban area with a substantial population connected to social media. The role of online information is investigated with a mixed-methods approach, using both quantitative (propensity score matching) and qualitative (in-depth interviews) techniques. The study relies on two data sources – survey responses from 469 Bangkok households and in-depth interviews with twenty-three internet users who are a subset of the survey participants. Propensity score matching indicates that social media use enabled households to reduce mean total losses by 37%, using the nearest neighbor estimator. Average loss reductions amounted to USD 3,708- USD4,886, depending on the matching estimator. Social media offered information that was not available from other sources, such as localized and nearly real-time updates of flood location and depth. With knowledge of current flood conditions, Bangkok households could move belongings to higher ground before floodwaters arrived. These findings suggest that utilizing social media as sensors could better inform populations during disasters, particularly in locations that lack real-time, accurate flood monitoring networks. Expanded access to social media might be useful in developing countries, ungauged basins, and complex urban environments. Overall, the study reveals that online information can enable effective disaster preparedness and reduce flood losses.

Rapid response research is critical to understanding community resilience in the face of coastal hazards, such as tropical storms and flooding events. Yet, on-the-ground data collection efforts following these events are often uncoordinated and occur at different spatial scales and data resolutions. This talk introduces a project that aims to define and document several standardized data formats that could be useful for integrated hazard exposure and human impact research. This project also aims to provide software tools to facilitate the use of these standardized data formats, to increase the value of open access data repositories for researchers studying the human impacts of coastal disasters. This presentation focuses on the challenges in collecting and integrating hazard exposure and human impacts data and paths forward in interfacing such data. The talk begins with an overview of study designs and data formats identified through the literature review as being common in studies of the human impacts of coastal disasters. Then some of the challenges that researchers experience in data integration for such studies and future promising approaches will be discussed.

Hurricane Maria devastated the Island of Puerto Rico leaving significant destruction to the existing housing stock. The limited government response resulted in reliance on informal practices for rebuilding housing. While informal housing is often considered unsafe, it is widespread and has been historically the primary means for building and rebuilding on the Island. Individuals whose homes were destroyed or badly damaged are vulnerable as they experience significant stress due to material and social losses. Their ability to cope with and respond well, or to have personal resilience, becomes critical for the sustainability of these communities. Personal resilience represents the ability to emotionally recover from highly disruptive events and is closely connected to how communities engage in the recovery process. While previous research has identified the need for establishing a “home” in the aftermath of a disaster, there is limited understanding regarding how personal resilience is affected by informal housing. To address this gap, this research used door-to-door surveys (N=305) in two municipalities in Puerto Rico– Loiza and Yabucoa. The statistical analyses assessed whether personal resilience was associated with informal housing, level of damage, level of assistance, social capital, and households socioeconomic vulnerability. Results show that the level of damage, bridging social capital, and age played a critical role in enhancing personal resilience while informal housing did not. The research contributes to the literature on personal resilience by identifying which social and ecological factors play a role into how communities can emotionally bounce back from disasters.

Natural hazard consequences include direct damages that are relatively easy to model, indirect costs that can be quantified in economic terms with more effort, and intangible consequences that are difficult to quantify. This last category has broad impacts for emergency response and long-term resilience planning but is often not directly integrated into modeling efforts due to the scale and the qualitative nature of the data. This presentation introduces an end-user vetted methodology developed for collecting local and regional consequences of site-specific infrastructure damage and for integrating these data with real-time predictive storm models used by Emergency Operation Centers (EOCs) around the country (e.g., ArcGIS Enterprise and WebEOC). This demonstration project, funded by the Department of Homeland Security, focused on Providence (RI)’s coastal critical infrastructure (e.g., wastewater treatment, fire and rescue, health facilities, and seaports) for use in the Rhode Island Emergency Management Agency’s EOC. Recent advances in hydrodynamic modeling, along with readily-available LIDAR elevation datasets have increased the resolution of storm models to the degree that flooding predictions can differentiate flooding probabilities across one land parcel (e.g., ADCIRC-Surge Guidance System). This allows for parcel-scale data to be meaningful in predictive storm modeling tools. By identifying wind/surge/wave/flooding thresholds for critical infrastructure failure, the identified concerns may be directly linked to the storm prediction models in real-time or for planning purposes. The end-user inputs to the model make outputs directly relevant to emergency managers as they allocate resources and anticipate the challenges of an imminent storm.

The COVID-19 pandemic has resulted in statewide stay-at-home orders across the United States (U.S.). This research examines the effects of social distancing measures on physical and virtual social capital networks at the individual-, community-, and organizational-level in the U.S. By analyzing data from an online name generator survey, crowdsourcing platforms, and telephone interviews using choice modeling, social network analysis, and qualitative descriptive analysis, the findings of this study will provide valuable insights into social isolation vulnerabilities, the potential for virtual volunteering to fill gaps in support networks, and the long-term impacts of disruptions to social capital.

The COVID-19 crisis will have lasting impacts on states such as North Carolina (NC), whose economy heavily relies on agriculture. Family farms are central to sourcing local food systems, supporting local economies, and contributing to food security, but they may be particularly vulnerable to the impacts of the COVID-19 crisis as they depend on retail outlets such as farm stands and farmers’ markets which are likely to be closed under the COVID-19 containment measures. This presentation offers preliminary findings regarding family farmers’ challenges, adaptive capacity, and resilience to operate under the COVID-19 crisis.

On the morning of September 19, 1985, an 8.1 magnitude earthquake shocked Mexico City and approximately 10,000 people died, and hundreds of buildings collapsed. This paper examines the most immediate comic book response to that catastrophe in Terremoto:85. It is argued that the Terremoto:85 comic book series depicts the Mexican nuclear family as an allegory of the nation that must remain together at all costs, without questioning the sociopolitical upheaval that inevitably emerges after the occurrence of a natural disaster. This graphic narrative shows that only those characters depicted as exemplary civilians get to have happy endings. Only those characters who embrace normativity, gender norms, and contribute to the procreation and production of the nation are granted happy endings. By doing that, Terremoto:85 obscures the real civil disobedience and action that surged after the occurrence of Mexico City’s earthquake of 1985. This study, therefore, encourages critics of natural disasters to look at graphic narratives as portals that reflect the cultural sociopolitical tensions that erupt after a disaster. Further, it is suggested that by not portraying the civil disobedience nor the social organization that surged after the earthquake, Terremoto:85 erases well-recorded civil responses to the earthquake of 1985. In doing so, this comic book provides key answers to examining representations of current national emergencies in visual culture that erase vulnerable and precarious populations from national narratives.

Infrastructure recovery time estimation is critical to disaster management and planning. Inspired by recent resilience planning initiatives, the authors consider the context where experts need to estimate the time for different infrastructure systems to recover to certain functionality levels after a hazard event. We propose a methodological framework to use expert-elicited data to estimate the expected recovery time curve of a particular infrastructure system. This framework uses gaussian process regression (GPR) to capture the experts’ estimation-uncertainty and satisfy known physical constraints of recovery processes. The framework is designed to find a balance between the data collection cost of expert elicitation and the prediction accuracy of GPR. This study evaluates the framework on realistically simulated expert-elicited data concerning two case study events, the 1995 Great Hanshin-Awaji Earthquake and the 2011 Great East Japan Earthquake.

Alleviating limited access to food that arises during disasters is the focus of many recovery efforts, however, there is no evidence base on which to build effective programs and policies to meet community food needs. For instance, modifications were made to allowable purchases for Supplemental Nutrition Assistance Program (SNAP) beneficiaries after Superstorm Sandy in New Jersey, so families were able to repurchase lost food supplies and Disaster Supplemental Nutrition Assistance Program (D-SNAP) is routinely activated following disasters. These modifications have been made without an understanding of those living on the cusp of food insecurity that may be pushed into insecurity due to the disruption of a disaster to the local food system resulting in a disruption to food access, availability, or accessibility created by the disaster event. In addition, many studies of risk and protective factors for food insecurity use nationally representative samples in the United States, overlooking groups or special populations outside of traditional demographic groups such as those impacted by disaster disruption or those at increased risk for health complications due to disaster exposure. To address these gaps, a sample of individuals at increased risk for COVID-19 were surveyed in NY State during statewide community restrictions or “NYS on PAUSE”. This presentation will describe household behaviors and barriers related to obtaining food during the pandemic in a low income and minority sample of households in NY State, preliminary findings on risk and protective factors associated with food insecurity during the pandemic, and discuss implications of food environment disruption and household food security during pandemics and other hazards.

For predictive models to provide valuable insights for pre-disaster planning, they need to capture how physical, economic, and social infrastructure systems within a community interact. However, a substantial amount of data is needed to develop such models. To model physical vulnerabilities, data is needed regarding critical infrastructure and residential buildings. To account for socioeconomic vulnerabilities, data is required regarding household and business demographics, disaster preparedness, and post-disaster behavior. Data constraints can force researchers to neglect certain facets of disaster impact in their models. In consequence, the compound effect of disasters, e.g., disruption in water supply leading to population displacements, is not properly accounted for. The authors argue that integrated assessments of the consequence of disasters can provide better insights for planning and help advocate for actions that will foster disaster resilience. This presentation will discuss the benefits of the integrated assessment disaster impact. A case study of post-earthquake household displacements in San Francisco will be the basis for the discussions. An agent-based model is developed where damage to buildings and infrastructure is simulated using a HAZUS-like approach. Affected households may be displaced due to structural damage, shortage of utilities, or their perceptions about the building habitability. The analysis of the impacts of earthquakes will be conducted with different levels of integration, that is, including or not including water and power availability, household disaster preparedness, and other factors. The results highlight the importance of integrated approaches for evaluating disaster impacts and motivate the systematic collection and sharing of physical, economic, and social data.

Asthma symptoms may worsen following hurricanes due to mold growth and discontinuity in care. Few studies have quantified post-hurricane changes in asthma-related emergency department (ED) visits. The objective of this study was to estimate changes in asthma-related ED visits in North Carolina following Hurricane Irene. Changes in asthma-related ED visits in September-December 2010 and 2011 were examined using data from the Healthcare Cost and Utilization Project databases which capture >97% of all ED visits. A Poisson generalized linear model was used to estimate the association between the Federal Emergency Management Agency disaster declaration and county-level asthma-related ED visits controlling for the month, year, and county. The model accounted for autocorrelation of monthly-county rates within the same year. Following Hurricane Irene, disaster declarations were made for 38 of 100 counties. In September 2010, the rate of asthma-related ED visits was six per 10,000 person-months. In September 2011 following Hurricane Irene, rates of asthma-related ED visits were similar in counties with and without a disaster declaration (seven per 10,000 person-months and five per 10,000 person-months, respectively). When adjusting for covariates, there was little difference in the rate of asthma ED visits before and after the hurricane between counties with and without a disaster declaration (RR[95% CI]=1.02[.97, 1.08]). Although risk factors for asthma exacerbations increase following hurricanes, these results found little evidence of an increase in asthma-related ED visits in North Carolina in the four months following Hurricane Irene. Further research that follows individuals with asthma may identify changes in asthma exacerbations following a hurricane.

The 2030 Global Sustainable Development Agenda of the United Nations highlighted the critical importance of understanding the integrated nature between enhancing infrastructure resilience and facilitating social equity. Social equity is defined as equal opportunities provided to different communities or people by infrastructure. It addresses disparities and unequal distribution of goods, services, and amenities. Infrastructure resilience is defined as the ability of infrastructure to withstand, adapt, and quickly recover from disasters. This study aims to examine the complex interrelationships between infrastructure resilience and social equity using social media data. Social media data is proven to be a reliable source of valuable information for understanding human activities and behaviors in a disaster setting. The spatiotemporal distribution of disaster-related messages helps with a real-time and quick assessment of the impact of disasters on infrastructure and society across different regions. As a first step of this study, this paper presents our work on collecting and analyzing Twitter activities during the 2018 Hurricane Michael in disaster-affected counties of the Florida Panhandle area. The collected Twitter data was organized based on the geolocations of affected counties and was compared against the infrastructure resilience and social equity data of the affected counties. The results of the analysis indicate that in Hurricane Michael (1) Twitter activities can be used as an important indicator of infrastructure resilience conditions, (2) socially vulnerable communities are not as active as general populations on social media in the disaster, and (3) socially vulnerable communities require a longer time for infrastructure-related disaster recovery.

Following a disaster, households are faced with several difficult decisions. Among the most urgent is how to pay for repairs of damage to the home. Major resources for housing recovery include personal savings, homeowners or flood insurance, Federal Emergency Management Agency’s Individual Housing Assistance Program, and the United States Department of Housing and Urban Development’s (HUD)Community Development Block Grant-Disaster Recovery Program, as well as Small Business Administration loans. Differential access to these resources and differences in the amounts and eligibility criteria among these resources combined with varying levels of social vulnerability lead to disparities in housing recovery outcomes. This work aims to address questions about which types of post-disaster resources are obtained and by whom. This presentation will leverage household-level longitudinal survey data from a community impacted by disaster. The longitudinal study, initiated in 2016 following Hurricane Matthew, is focused on the community of Lumberton, North Carolina. The larger project team has executed systematic housing disruption and recovery surveys with the same sample of households four times over the course of four years. This presentation focuses on data collected during the first two years, immediately after Hurricane Matthew and approximately one year later. Quantitative analyses are used to explore patterns of access to various types of recovery resources as well as to identify the relationship between resources and recovery outcomes. Recovery outcome variables include dislocation time, repair time, and recovery progress. Initial damage to the housing unit is included alongside social vulnerability variables such as race, ethnicity, income, tenure, and resource variables such as insurance.

Prior research shows that hospital administrators relied on intuition, experience in previous events, facility knowledge, and availability of information about the approaching storms to inform their decisions to evacuate or shelter-in-place following Hurricanes Sandy and Harvey. Yet, factors that administrators and leaders consider when deciding to close, evacuate, reopen, or reenter healthcare facilities following an earthquake without advance notice remain largely unknown. To address this knowledge gap, this timely research explores factors considered by healthcare administrators making decisions about their facility following the 6.4 and 7.1 magnitude earthquakes that struck Ridgecrest, California on July 4 and 5, 2019, respectively. Semi-structured interviews were conducted with healthcare administrators or leaders involved in decision-making about the closure, evacuation, reopening, and/or reentry of their facilities following the 2019 Searles Valley Earthquakes in Ridgecrest, California. Interviews were professionally transcribed, coded, and thematically analyzed. Understanding the factors considered in the healthcare administrators’ decision-making process around closure, evacuation, reopening, and reentry can help predict and improve healthcare system capacity following other large earthquakes, as well as inform the development of tools such as decision aids and training that can enable healthcare administrators to make informed, risk-based decisions.

During field research (FR) in the natural hazards space, scientists from different disciplines use sophisticated equipment and diverse methods to collect data. This data needs to be managed to facilitate analysis and curation workflows which culminate in publicly shared findings. Throughout the process, a significant challenge is to bridge complementary perspectives for interdisciplinary teams that work collaboratively, and for users discovering FR data. Advanced by CONVERGE, an initiative to advance social science, engineering, and interdisciplinary work in Natural Hazards, the vision for managing and publishing FR data was to enable collaboration and to reflect interdisciplinary work as it is conducted across time from fieldwork to data analysis and publication. This vision was executed in DesignSafe-CI is a National Science Foundation-supported online platform designed to allow hazards and disaster researchers to securely store, analyze, publish, preserve, and share their data along with associated data collection instruments and research protocols. Designing and implementing a model to curate and share FR data required interdisciplinary collaboration. Social scientists, engineers, data curators, user experience designers, and developers brainstormed, drew, discussed, implemented, and tested the FR data publication pipeline in DesignSafe. In this presentation, representatives of the larger team discuss how we blended ideas and workflows into development, show the results as interactive curation interfaces and published datasets, and invite the community to publish and reuse interdisciplinary datasets. This presentation will provide the opportunity to raise and respond to thorny ethical issues related to privacy, data quality, data licensing and access, and academic credit.

The impacts of wildfires on communities have become more pronounced in recent years as the intensity and frequency of wildfires has increased in densely populated areas of the United States. Communities located in the wildland-urban interface (WUI) neighboring high wildfire risk zones are at the highest risk of damage to civil infrastructure and destruction of personal property. This presentation summarizes an investigation on the performance of schools and healthcare facilities in Paradise, California following the 2018 Camp Fire. The authors performed observational and quantitative analysis of the facilities through the use of the National Science Foundation-funded RAPID Equipment Facility and interviews. The loss of functionality of education and healthcare facilities due to both structural and nonstructural damage significantly impacted the initial recovery of Paradise. In agreement with previous research, building materials and defensible space are critical to minimizing damage to buildings during a wildfire hazard. It was found that horizontal framing of structures was most vulnerable to wildfire damage and that nonstructural damage to schools and hospitals had an equal or greater impact on community recovery. Understanding the vulnerabilities of WUI communities to wildfire will help with disaster mitigation and recovery planning and aid in restoring critical services after the disaster.

There is alarming evidence of changing climate conditions, such as increased atmospheric and ocean temperatures, extreme precipitation events, and global-sea level rise. Predictions of the full impacts on engineered systems continue to be a challenge for the engineers, scientists, and policymakers tasked with ensuring the resilience of coastal infrastructure, facilities, and communities. At the United States (U.S.) Coast Guard Academy (CGA), the engineering faculty recognize the need to educate the future of CGA Service and have developed a Coastal Resiliency course that provides a clear view of reality regarding the science of climate change, its impact on coastal infrastructure, and on the necessary socio-economic planning and design of resilient communities. This presentation shares how that course provides preparation for the real-world practice of engineering by exposing students to the importance of risk and vulnerability assessment within the context of changing climate conditions. Through project-based learning and community engagement with local town stakeholders, students are exposed to engineering vulnerabilities associated with extreme weather and rising sea level and to the challenges and concerns with implementing solutions in communities where these vulnerabilities exist. As the U.S. Coast Guard’s primary accession point for engineers and scientists, ensuring future community and industry leaders are informed about the potential challenges that will likely occur during their career is an example of how CGA is developing active hope in an era of environmental extremes.

The impact of natural hazards on the built environment has increased significantly in recent years. Floods, hurricanes, earthquakes, and wildfires are taking a toll on the built environment. This increase in natural hazard damage is occurring as the world has moved from an environment of data paucity to one that is incredibly data-rich. Sensors that monitor building and infrastructure system performance, drone imagery, cell phone location data, social media posts, transit access swipes, and vehicle GPS records provide an incredibly detailed view of how urban areas function under both normal conditions and under the stress produced by natural hazard events. This data explosion can provide a foundation to develop a better understanding of natural hazard impacts and potential ways to mitigate and respond to them. To exploit this new data-rich environment, the natural hazard community needs to develop strategies to engage these rich new sources of information. Collecting data in normal times can provide a baseline to better understand how natural hazards impact people and their use of urban systems. This presentation reviews several existing sources of data on natural hazards and the built environment and develops a strategy to collect, curate, and distribute data on natural hazards data and built environment that can support new research on hazard mitigation and response. The presentation will discuss alternative ways to assemble and maintain this data to support the natural hazards research community.

The engagement in risk mitigation practices is widely recognized as a consequence of a previous benefit-cost analysis of the potential adjustments. In this study a mail survey was conducted with a random sample of 1,500 households in the Red River of the North Basin, North Dakota to estimate the households' perceived economic value (PEV) of the National Aeronautics and Space Administration’s (NASA) Satellite Enhanced Snowmelt Flood Prediction system, measured through their willingness to pay for such product. 211 surveys were completed and returned (14%). The hypothesis of the study was that the relationship between people’s flood risk assessment and their previous mitigation behavior and PEV is mediated by the perceived applicability of the product. Results show the mediation effect is just partial. The model rejects the mediation effect of the perceived applicability of the product in the relation of flood preparedness and flood insurance variables with PEV (direct relationship) but confirms the mediation of the perceived applicability of the product in the relationship between flood risk perception and PEV. In other words, households who have previously engaged in preparedness activities were less likely to be willing to pay for the NASA product while those with flood insurance had higher willingness to pay. Those with a higher flood risk perception valued the applicability of the product higher and thus were more likely to contribute economically in its implementation. Results emphasize the importance of previous mitigation behavior in the willingness to further invest in a public flood prediction system and confirm the complexity of benefit-cost analyses regarding mitigation adjustments, clearly describing a multi-layered non-direct process.

In recent decades climate change, wildfire, and land change have been growing issues in lowland Bolivia and across South America. During the 2019 fire season, millions of acres burned in Bolivia, bringing international attention often without nuance, context, or history, to the issues of wildfire in the Amazon, Chiquitanía, and the surrounding ecological zones. This presentation will share a study of risk and disaster through a multi-sited, mixed-methods research protocol incorporating participant observation, interviews, focus groups, and participant mapping coupled with remote sensing and Geographic Information System analysis of fire history and land change. This study examines the intersections and confluences of climate change, land management policy, resource regime dynamics, and politics in the production of disaster and cascading impacts after the fires in the Bolivian lowlands.

Youth and young adults have been identified as assets in disaster mitigation and management in the United States and abroad. From the perspectives of social justice, civic engagement, and youth empowerment, understanding the role of youth and young adults is an important future direction for disaster-related research. This presentation will reveal findings from a scoping review of the peer-reviewed literature on disaster mitigation, response, and recovery following disasters—hurricanes, earthquakes, tsunamis, tornados, landslides, floods, and severe weather events—with a particular focus on the engagement of youth and young adults. Using the Arksey and O’Malley model of a scoping review, studies are identified that answer the primary research question: What is the role of youth and young adults in community natural disaster mitigation and response? The purpose of this scoping review is to map the existing literature in terms of the volume, nature, and characteristics of the primary research. This scoping review will identify studies published between 1980-2020 that include research or discussion on the role of youth or young adults ages 16-24 in mitigation and response to disasters. Studies meeting the search criteria are charted and summarized to understand the unique role of youth and young adults in responding to natural disasters in their communities. The ultimate goal of this study is to inform the development of relevant, cost-effective, and culturally appropriate interventions that encourage and empower youth involvement in disaster mitigation and response efforts.

In much of the resilience planning discourse, the difference between intended and actualized behaviors towards mitigation and adaptation is not recognized nor studied—on both the individual and institutional levels. The nature of disasters is such that there is typically not a pre-event snapshot of a given agent’s behavioral intentions around mitigation and adaptation behaviors that in turn inform their resilience capacity towards a given disaster event, which is comparable to data on interruption and recovery post-event. Furthermore, past research indicates that people do not always engage in disaster preparedness or mitigation – even when they have sufficient resources, preparedness training, or a history of disaster exposure. A significant contributor to disaster preparedness tends to be people’s risk perceptions, however, the Theory of Planned Behavior (TPB) offers additional considerations that affect intended and actualized behaviors. The TPB can be directly applied in the domain of resilience planning – where a category of mitigation and adaptation behaviors are considered. The authors review the very few papers to date that consider the TPB in disaster risk reduction. A theoretical model based on pilot study efforts for application of the TPB to resilience planning is introduced that discusses the importance of attitude, subjective norms, and perceived behavioral control in the prediction of intention for resilience planning, and the relative importance of intention and perceived behavioral control in the prediction of behavior. A series of best practice data collection questions are introduced that relate to businesses' willingness to mitigate.

Wuhan, China was the site of the first confirmed COVID-19 case in November 2019. From January through March 2020, the city and surrounding Hubei province were subjected by the Chinese government to a quarantine and blockade. Mixed-methods online research, conducted in Mandarin Chinese with persons who were in Wuhan during the COVID-19 lockdown, provides insights into the adaptation strategies of a gendered and otherwise segmented urban economy—already seeking to counter the climate crisis—during a pandemic, quarantine, and blockade. Gender, articulated with age, occupation, income and savings levels, and rural to urban migration status, can serve both as a demographic identifier and a marker of vulnerability and resilience in crisis contexts. Applying an intersectional gender-lens, to understand how Wuhan residents have dealt with a mega-city blockade and pandemic, can deepen what is known about adaptation and response to frequent, sudden-onset emergencies and traumas as well as slower onset events, such as climate change.

In the face of escalating climate change, effectively managing the population and assets exposed to climate hazards is critical to limiting damages. Recognizing the benefits of moving out of harm’s way, the United States government has already spent billions of dollars removing over 40,000 households from flood-prone areas. However, without simultaneous efforts to limit new development in hazardous places, the population and infrastructure exposed to floods may increase. Some municipalities may strategically grow by channeling new construction away from floodplains and buying out existing flood-prone houses, while others may ignore floodplains altogether in their decision-making. Yet, with little data available on how development in floodplains has changed over time, knowledge about which municipalities have effectively managed their floodplains–and how–remains limited. In this study, a novel parcel-level dataset is constructed to measure changes in floodplain development over time and identify which municipalities have grown while limiting exposure to flooding. By combining buyout records, real estate databases, and land use data, this study measures the area of floodplain developed and the change in the number of housing units in the floodplain between 2001 and 2016 across the State of North Carolina. Initial results indicate that many municipalities use property buyout programs to remove housing from floodplains while simultaneously permitting extensive new construction in the same floodplains. Overall, the findings indicate a substantial need to harmonize federal and state-level buyout regulations and funding priorities with local policies aimed at minimizing floodplain development. Future work will examine patterns and policies underlying municipality performance.

College students in the United States have been negatively affected by the coronavirus disease 2019 (COVID-19) pandemic. In addition to the economic and health impacts of COVID-19, college students have experienced severe social disruptions due to campus closures. As a result, college students may be at higher risk for dropping out. Increased rates of student drop-out may lead to negative downstream economic and public health consequences. To prevent or mitigate these negative consequences, it is critical to support students during the COVID-19 pandemic. However, there is a lack of empirical knowledge concerning the impact of disasters such as this pandemic on college students and the students’ related needs. The COVID-19 pandemic presents a unique opportunity to address this gap and identify avenues for student support. We are evaluating the impacts of the COVID-19 pandemic on college students using an online survey and semi-structured interviews at two time points. Baseline data collection occurred in April and May, 2020, with time two data collection planned for the fall of 2020. College students, faculty, and academic advisors were recruited through convenience and snowball sampling methods. The study assessed outcomes including stress, anxiety, and depression, as well as risk and protective factors such as social support, coping, and purpose formation. Our presentation will outline the significance of this study and discuss available preliminary findings from the baseline data. Implications for research and practice will be discussed.

An important goal for risk communication studies is to assess the degree to which information transmission influences how residents weigh the credibility of diverse information sources and, in turn, how this assessment affects their protective action decisions. To better understand this process, the present study examined two datasets—the 2009 Boston water contamination emergency and the 2017 Sichuan Earthquake adjustment—which are characterized by different levels of uncertainty. Both studies reveal that respondents attach the highest credibility to authorities (for example government agencies), followed by the public (for example news media) and private intermediates (for example peers), and the least credibility to the ultimate information receiver (for example families and self). In response to the emergency with lower uncertainty, respondents’ adoption of recommended protective actions depends on the credibility of authorities, whereas the adoption of a non-recommended protective actions results from respondents’ beliefs in their own expertise and their household emergency preparedness. On the other hand, the effect of information source credibility on respondents' protective actions in response to an emergency with higher uncertainty are mediated by risk perceptions.

As uncertainty during emergency response operations increases, so does the need for responders to display heightened levels of adaptive performance. Emergency response organizations such as the fire service, however, tend to constrain adaptive behaviors due to their highly formalized bureaucratic structures characterized by extensive rules, policies, and procedures. To bypass these constraints, structural theories suggest that leaders empower employees with more control over their work conditions. To explore this idea, survey data from three separate United States fire departments was collected to test whether mid-level supervisors can empower firefighters by increasing their ability to improvise during complex response incidents and the degree to which this enhances adaptive performance in fire departments. Moreover, in a moderated mediation model, this study tests whether senior leaders must also be effective, empowering leaders to achieve this effect, as many senior leaders in the fire service are criticized for being overly bureaucratic, risk-averse, and resistant to change. The initial results show that supervisors who coach, inform, lead by example, show concern, and encourage participatory decision-making to enhance firefighters’ self-determination. This result also explains why such behaviors among supervisors increase levels of department adaptive performance. Besides, when supervisors and senior leaders are both empowering and provide balanced procedural constraints to subordinates, these indirect effects become stronger, implying that senior leaders also play an important role in the empowerment process. These findings answer crucial questions regarding how response organizations can become more adaptable in conditions of high uncertainty despite various constraints.

This study approaches community-level disaster preparedness planning from a resource-matching perspective to better understand the potential for the utilization of local resources in a disaster scenario. The project is set in the Pacific Northwest, where many disaster preparedness efforts are focused on a potential magnitude 9.0 Cascadia Subduction Zone (CSZ) earthquake event that would likely necessitate that communities be self-reliant for an extended period. Data gathered from a pilot community resilience sample survey illustrates the expectations that community members have for being able to access needed resources in an earthquake scenario. The study seeks to better understand the extent to which community member expectations align with resources that might be available in a disaster scenario, and where gaps in this alignment occur. Understanding the nature of these gaps can help to guide community members, social infrastructure providers, and disaster preparedness planners to better prepare for uncertain events based on the differing needs of communities. Specifically, we look at which resources people expect to access via social ties or local institutions as well as community knowledge of available local resources. This pilot project is currently being expanded to additional Washington State communities in both urban and rural locations.

Over the past decade, post-disaster relocation has been one of the key policy decisions used to protect communities from future devastation. Relocation policy is also considered to be a climate change adaptation strategy. If done correctly, relocating communities will benefit from less exposure to hazards. However, in practice, relocation has long been the least favored policy as it destroys social and economic networks. This is compounded by the reality that developing sites and houses becomes the highest concern, resulting in a lack of community involvement. This research analyzes relocation governance post-disaster, comparing cases from Tohoku, Japan (earthquake and tsunamis); Leyte, the Philippines (storm surge from a typhoon); and Yogyakarta, Indonesia (volcanic eruption) to understand the impact of different governmental approaches to community relocation. It explores how program design and governance structure impact the implementation of community relocation and how that affects community engagement and ultimately relocation outcomes. Key findings include 1) housing programs that existed pre-disaster are the most likely mechanisms to be used for community relocation, 2) governance structure and stability greatly influence processes, 3) allocated relocation incentives affects recipients’ sense of responsibility, and 4) the key way stakeholders approach relocation is a result of their culture. In sum, relocation governance functions better when it is structured to center communities and holistically incorporates stakeholder involvement, rather than governments simply providing housing.

The San Francisco Bay Area is known for its high levels of traffic congestion. Extreme flood events resulting from combinations of storm surges, tides, sea-level rise, and other drivers can lead to closures of major roads and highways in low-lying coastal areas, thus causing further disruption to traffic networks. Such disruption has socioeconomic consequences such as loss of wages and jobs for individuals, economic costs for businesses due to the delay or absence of workers, increases to road accidents, and loss of life. The goal of this study is to understand changes in morning peak-hour traffic patterns in the San Francisco Bay Area caused by coastal flooding. The authors integrate an incremental traffic assignment model with a set of flood maps to simulate regional traffic patterns. This study uses Census data to represent origin-destination pairs for commuters. Disruption is then quantified in terms of employee absences and travel-time delays. This study highlights the far-reaching impacts of coastal flooding, where inundated road segments in the vicinity of the Bay can cause travel-time delays much farther inland. Thus, flood-related traffic disruption is not a coastal problem but a regional one and therefore requires a coordinated regional response. This analysis also suggests that, for sufficiently large regions, flood exposure is not correlated with travel-time delays. Instead, this study shows that the metric on road network density is a good predictor of travel-time delays.

This talk surrounds the theme of “Active Hope: Disaster Researchers' Potential to Contribute to a Path Forward”. Disaster researchers seem to be at a fulcrum in history, where shifting environmental hazards and vulnerabilities wrought in global, political, and economic transformations point to enhanced danger to large populations. Troublingly, just at the time that robust scientific inquiry is needed, attacks on science emerge from all points of the compass. Yet, each day brings new situations where a scientific understanding of risk, hazard, disaster management, and related themes could inform policymaking and enhance human wellbeing. This presentation invites consideration of what research and research ethics might be if construed from the optimistic standpoint of the benefits of ongoing research: what if research were grounded in an appreciation of the capacities of people affected by disasters; by a recognition that research is valuable; by respect for autonomy and choice by those in impacted areas; an emphasis on resilience and efficacy and not vulnerability and insufficiency; and by seeing researchers as accomplished professionals and not risk objects. What, then, might we achieve?

Over the past year, various calls have been made for establishing 'National Transportation Safety Board-style' investigations of disasters. For many disaster researchers, these calls have been frustrating; there is a large community of researchers who already conduct rapid-response research to disasters who rigorously investigate such events and who share their findings broadly. This project offers an epistemic analysis of post-event disaster research. Each side of the previously-mentioned debate makes a series of epistemic and ethical claims about the nature of what ephemeral data must be preserved, about the requisite expertise required to understand disasters, and about what kind of information and analysis is most useful for their mitigation and management going forward. Disentangling these arguments and clarifying the debate offers the opportunity for explicit reflection within our community about what we have achieved through disaster research and where our field can continue to grow. To assist in moving these discussions forward, two sets of empirical data are offered. First, the authors develop and present a taxonomy of different approaches to post-event research from around the globe, contrasting the strengths and weaknesses of each approach. Second, the authors share perspectives from two evaluations of rapid-style research projects with which we have close connections: post-wildfire and COVID-19 ongoing studies. Through these illustrative examples, this talk argues for an evolution in our field's approach to collecting pre-, during-, and post-event data.

Public housing residents and subsidized housing renters face severe challenges in terms of post-disaster housing, but cities with little experience with disasters may have limited knowledge of how to help them. While research on post-disaster recovery policy has grown, there is still a lack of knowledge on pre-disaster planning for post-disaster housing, especially in at-risk communities who have not experienced a recent disaster. Particularly, studies are lacking on policies that determine where public housing residents—a highly vulnerable social group—go after disasters, when and where they can return, and their overall recovery trajectory. This knowledge gap is critical, given the high dependence of public housing residents on government assistance for recovery in the aftermath of disasters. Using Salt Lake County, UT as a case study, this research looks for pre-disaster policies to provide post-disaster housing to public housing residents. The county faces a 57% probability of experiencing a 6.75 magnitude earthquake in the next 50 years. This study uses archival research techniques and semi-structured, open-ended key informant interviews with local emergency management, city planning, and housing officials of the Salt Lake County and its constituent cities to examine their state of post-disaster housing plans; motivations behind the planning; challenges and opportunities for this effort; and coordination between various entities in this regard. Preliminary findings suggest that cities are not prepared to provide housing to public housing residents after disasters, and there is a general lack of predefined policies to help these socially vulnerable populations to recover in the long run.

This study describes the importance of the link between women’s roles as caregivers and traditional culture during a disaster. The relationship between women and local traditional culture is defined, before the Great East Japan earthquake in 2011 and information is provided on how local traditional culture affects tsunami evacuation behavior using literary analysis and a theoretical basis of field practice. This project mainly studies community oral testimony collected and edited by a local monk. It is the only document that contains details of how ordinary men and women died and the steps they took during evacuation. Therefore, the monk, the editors of his book, and relatives of the deceased were all interviewed. Results show gender inequality exists in the Tohoku area and its effect on the tsunami evacuation activities during the earthquake. The presentation discusses invisible gender rules in the disaster-stricken area and female circumstances during the earthquake. Methods include an analysis of 620 testimonies of the deceased using the testimony document created by the monk. This information was then coded and collated in a chart. A workshop with other postgraduate students was also organized to ensure the accuracy of the coding and finalize the coding. Factor analysis was then implemented, followed by testing the null hypothesis and an alternative hypothesis. Gender has been calculated as the second most crucial factor in our results. This talk will identify tsunami evacuation behavior through the gender lens. Women can help promote resilience in disaster-affected areas based on the disaster testimony collected by a local monk.

This presentation reflects on what it means to care in disaster research. It explores how current disaster research methods seek to fulfill institutional requirements for ethical research procedures and protocols. Yet, disaster researchers continue to grapple with a range of subjective and relational dilemmas. The concern is not just around how to deal with researchers' emotions, but also how researchers relate to people in disaster-affected communities—‘other’ researchers across post-colonial ‘locals’—and institutions that are entrusted with a duty of care. What are the ethical and personal strategies employed by disaster researchers across such contexts? This work contributes two strategies to this ongoing discussion. First, it encourages disaster researchers to approach research as care—an embodied and relational way of seeing, doing, and being in the world. Around the world, disaster researchers increasingly belong to disaster-affected communities. Researchers’ personal traumas and experiences are intertwined with their professional ways of seeing, doing, and being in a disaster context. It is important for disaster researchers to acknowledge such entanglements. Second, the talk proposes an ethical framework for researching with CARE (collaboration, accountability, responsiveness, and empowerment). Beyond satisfying institutional ethical protocols, the talk examines the concrete everyday practices through which disaster researchers develop networks of care, trust, and reciprocity in traumatic contexts. In conclusion, disaster researchers are encouraged to also learn from people and communities on the frontline of disaster impacts. For example, what can disaster researchers learn from the surge in caring work performed by young climate activists, volunteers, and journalists on the frontlines? How can disaster researchers learn from the strategies of faith, healing, and resilience demonstrated by people who experience frequent devastation and displacement around the world?

Post-disaster information products—such as impact estimates from crowdsourcing, social media, and machine learning models—are developed rapidly to inform decisions that support response and recovery activities. Despite revolutionary growth in the availability of such data, most disaster information products continue to overemphasize metrics of loss or impact that are not transparent, human-centric, or sensitive to ethical issues in response and recovery. The authors developed frameworks to improve the rapid assessment of holistic disaster impacts and recovery needs using a combination of large, openly available datasets and a survey of 815 households affected by the 2015 Nepal earthquake. This project developed 1) a Geospatial Data Integration Framework (G-DIF) to systematically combine available post-disaster datasets into a more accurate and context-specific estimate of immediate damage, and 2) a framework to map Post-Disaster Persistent Needs (PDPN) to predict future needs of households likely to face the most obstacles during recovery. These frameworks allow for needs assessments that better align with stakeholder priorities. This research advocates for three shifts in the way big data is used in post-disaster evidence today. First, the talk illustrates a shift from estimating losses to quantifying needs, moving away from housing-dominated damage assessments towards more holistic measures of multifaceted disaster impacts on communities. Second, the value of shifting towards models that are flexible to diverse inputs so they can be tailored to specific locales (inputs-agnostic and context-specific) are highlighted. Third, the talk demonstrates how models can be developed while supporting ethical priorities outlined by stakeholders, such as equitable recovery.

Governing toxic chemicals utilized to produce extreme energy carries implications for citizens’ right to know and confront contaminants in their groundwater and immediate environment. Many chemicals associated with hydrofracking are sheltered as proprietary intellectual property thus raising questions about who governs their use and who is accountable for their extended effects. State-sanctioned studies and corporate discourses actively work to constrain citizens’ knowledge while purporting the necessity of complex drilling technologies concerning national security. Manipulating data to legitimize the safety of extreme energy technologies discounts the lived experience of those who have suffered poor health outcomes and environmental contamination in favor of sustaining fossil-fueled capital accumulation. Moreover, because eco-governing strategies are predicated on maintaining a global economy lubricated by oil, corporate-state efforts toward deregulation mobilize knowledge infrastructures founded in the primacy of scientific methods and quantitative metrics to legitimize their environmental degradation. Attention toward the immediate material and economic implications of chemicals utilized for extreme energy production simultaneously standardizes the technological determinism of extreme fossil-fuel production and normalizes the systemic toxicity that defines extractive environmentality. Thus, while scholars and activists alike praise the potential of data, metrics, and monitoring to ensure accountability, it is clear that governing rationalities foundational to extreme energy production must also be rendered accountable.

How disaster researchers identify, describe, and address vulnerability to hazards is changing to reflect updated understandings of the capacities and constraints that people face when a hazard threatens. Novel methodologies accompany these updates to theoretical models of vulnerability. This presentation presents the methodology and insights from a project designed to analyze how risk information conveyed over social media may have enhanced capacities and alleviated vulnerabilities immediately before, during, and after Hurricane Sandy. The authors argue that digital communication of informal weather risk information over social media enhances capacity among people who may be at risk from hazardous weather. Increasing digital communication has many advantages (e.g., speed of communications, reaching large populations, etc.), but also disadvantages (e.g., the digital divide which isolates individuals who are not “digitally connected”). How are people communicating about weather risks in creative ways that cultivate capacity, particularly among people who may be more vulnerable to storm impacts? How can social media alleviate vulnerability through communicative capacities overall? To answer these questions, this study developed a novel methodology that uses ethnographic insights from focus groups to guide the analysis of massive amounts of Twitter data collected around Hurricane Sandy’s 2012 landfall in New York City. By using “Deep Data” from focus groups to render “Big Data” tractable and meaningful, results show that social media data can yield rich insights that contribute to novel theories of vulnerability and capacity in the context of hazards.

Local governments’ initial unresponsiveness to the COVID-19 outbreak has dragged many municipalities into a dilemma between containing the virus and maintaining adequate public services. The official information sources also face a challenge in balancing timeliness—competition against rumors—and accuracy due to understaffing. These challenges call for a study of the connections among the emergent norms among authorities and residents, residents’ perceptions, threat perceptions, and protective actions on both sides. Through initial inductive analysis, a series of causal links are identified between timeliness of government emergent norms and residential stakeholder perceptions, between government and residential emergent norms, between misaligned threat-stakeholder perceptions and the unbalanced demand-supply of information and necessities, and between residential emergent norm and order compliance. The theoretical model of this study reflects these causal links and seeks answers of the following questions in a catastrophic context: (1) what are the critical factors of establishing an emergent norm among residents; (2) what are the determinants of residents’ perceived critical resources (e.g. information, necessities, and lifeline systems); (3) what are the relationships among government and resident’s emergent norms and protective actions. To test this model, the secondary data collection uses a quantitative survey distributed to the residents who experienced different levels of stay-at-home orders. The findings can assist local authorities to prioritize services and limited capacities when facing large-scale pandemics and mass quarantines. This study also emphasizes the managerial aspect of effective risk communication in pandemic response.

Recently, communities have encountered more frequent and intense natural disasters causing billions of dollars in property damage with life-threatening risks. Given such heightened risks, disaster scholars are envisioning low-cost alternatives to disaster response strategies so that communities can depend less on investing in physical infrastructures but more on leveraging social factors. For example, recent studies have shown how increased social capital and dense social networks can help to rebuild communities in the aftermath of a disaster. Studies have also shown how emerging social media platforms facilitate disaster communication and early awareness. The main goal of this study is to design and develop a crisis communication and resource sharing platform to facilitate community interaction and the exchange of risk information with trusted information neighbors in the cloud. In order to facilitate user interaction in different crisis scenarios, the application develops a social network graph on top of the geotagged location, based on the analytics and results obtained in our recent social media studies. As soon as a node finds its neighbors, it provides its geographic position to the neighbors and obtains complete information about the list of other nodes and their positions in the network to exchange messages. Each node will be able to perform data gathering tasks (e.g., text, audio, video). This application will be armed with necessary graphical interfaces to show the connected nodes in the network as well as a connected social network. The social graph includes smart functions for efficient crisis communication and resource allocation for future disasters.

Crisis informatics can benefit the overall preparedness and resilience during disasters as well as epidemics by collecting and transforming data into usable forms. For example, it can help increase the number of lives saved, improve quality of life for the affected people, accelerate economic recovery, and create new jobs. This requires examining multi-disciplinary areas and addressing data challenges related to scope, scale, complexity and uncertainty. Recently, Microblog service providers such as Twitter and Facebook are serving as potential sources for handling relevant information during disasters or emergencies. Social media can support ground truthing, sharing information, and crowdsourcing as well as collecting responses from different groups of its users (individuals, agencies, and other stakeholders). Ability of self-generating and sharing data significantly accelerates the speed of information production and spreading, making social media an increasingly popular means for emergency communications as it provides real-time content, sentiment, and trends of public attention and behavior. In this study, emergence of diverse concerns and interactions over social media with the spread of COVID-19 from different stakeholders such as emergency officials, policymakers, local communities, practitioners, and news media were explored. Twitter application programming interface (API) was used to collect large-scale geo-tagged tweets based on COVID-19 related keywords to perform data analytics on service characteristics, retweeting activities, situational awareness, communication patterns, early awareness, public health and travel impacts, and overall response and recovery efforts.

To assess the disaster recovery of communities through the lens of wellbeing, this study explored the use of three different types of physical activity monitoring data. First, this presentation highlights a recent study that used bike/pedestrian counts collected at public trails to assess the impact of 2017 Hurricane Harvey. This type of data is increasingly collected in many urban areas in the United States and has the potential to understand disaster impacts and recovery in near real-time. Second, the authors analyzed social fitness app data from the communities impacted by Harvey. The data, provided by Strava Metro, anonymizes, aggregates, and provides activity data (e.g., cycling and running) from Strava app users. This analysis revealed different physical activity impacts of Harvey across communities and associated recovery trends. Third, the authors explored the possibility of using individually shared physical activity data. An online platform was developed to allow users of fitness monitoring apps (FitBit, Google Fit, and Strava) to grant us permission to access their data through the OAuth 2.0 protocol. The authors conducted semi-structured key informant interviews with those impacted by Hurricane Harvey and the Tubbs Wildfires to understand their use of apps pre-event and post-event. App users were generally willing to share aggregated and anonymized data with researchers and public officials in order to support disaster response and recovery. This study provides insights into the value of physical activity monitoring data in supporting disaster management and highlights opportunities and challenges that future studies may address.

In October 2016, Hurricane Matthew wreaked havoc on Lumberton, North Carolina, an impoverished, highly socially vulnerable community of roughly 20,000. Less than two years later, in June 2018, the city government announced that $12.9 million in Hazard Mitigation Grant Program (HMGP) assistance would allow over 100 homeowners to relocate from or elevate their disaster-affected homes, seemingly signaling the dawn of a brighter day. Rather, it proved simply to be the calm before another, even more powerful, storm—Hurricane Florence, which struck Lumberton in September 2018. In less than two years, scores of homeowners in large swaths of Lumberton twice faced the decision to rebuild or relocate. In April 2019, the authors traveled to this majority-minority community to document the stories of 15 homeowners who received HMGP assistance. Through harrowing accounts of evacuating rapidly flooding homes—in many cases, twice—and a yearning for assurance of no future disasters, the experiences of those homeowners with mitigation assistance in their journeys to recovery are shared in this presentation. Specifically, homeowners’ perceptions of the city government’s coordination and disbursement of HMGP assistance are explored in the context of recipients’ decisions to rebuild or relocate. We unveil varying degrees of satisfaction or frustration toward the city government, the latter more commonly vocalized by recipients who elected to remain. Homeowners’ stories of loss and longing, hope and hopelessness, and deep-seated despair and distrust are interwoven into a complex tapestry that reveals resilience in spite of largely not yet financed—and therefore, unfinished—journeys to recovery.

This paper highlights the ethical challenges of gathering perishable data from communities impacted by disaster in the Caribbean region and provides a justice framework to navigate these ethical challenges. The post-disaster environment presents a multitude of ethical and logistical challenges for researchers interested in gathering timely and unpreserved data. However, there are social barriers and unspoken societal norms that can be greater obstacles for researchers considered outsiders to the community. Although a researcher may gain physical access to the “space” where a disaster occurred, gaining social access and the trust of community members require vastly alternative methods. In the Caribbean, this social barrier is compounded by a colonial and neo-colonial legacy of exploitation and the mistrust of predatory research conducted by outsiders. The authors highlight the challenges encountered from post-disaster fieldwork and community studies conducted between 1989 and 2017 along with the methods utilized to overcome these barriers. The disaster events covered include fieldwork conducted in St. Thomas and Montserrat after Hurricane Hugo (1989), Grenada after Hurricane Ivan (2004), the United States Virgin Islands following Hurricanes Irma and Maria (2017-2018), Dominica following Hurricane Maria, and the Bahamas following Hurricane Dorian (2019). Also, informing this study are community studies conducted in Bermuda (2008-2009), Jamaica (1995), and the Bahamas (1996-2012). Additionally, the authors conducted interviews with emergency managers and residents in four Caribbean countries (the Bahamas, St. Kitts & Nevis, Trinidad & Tobago, and Dominica) to gauge their opinion on external researchers and the possibility of their offices serving as coordinating centers. Ethical challenges covered include research ethics committees or institutional review board approval; local institutional oversight and capacity building; engaging in the acute phase; local and equitable collaborators; reciprocal travel by survivors; cultural competency; and sufficient compensation for research participants.

disaster resilience of the nation’s critical infrastructure. According to the National Academy of Sciences, research on understanding and analyzing the disaster resilience of our infrastructure systems is a “national imperative.” To address this need, this paper proposes an automatic keyphrase extraction methodology to extract relevant phrases on disaster resilience from documents in the infrastructure domain. In developing the proposed methodology, a document collection including research papers and public reports were prepared. Noun phrases are first extracted from every sentence in the collection and form the candidates for keyphrases following a filtering procedure. Each candidate phrase is then represented as a global semantic vector and a local semantic vector. To select relevant phrases on disaster resilience, a semantic similarity measure is proposed to incorporate the semantics of candidate phrases in both the general and infrastructure domain. Ten physical resilience concepts from a pre-developed community resilience hierarchy are selected as the target concepts to evaluate the performance of the proposed methodology. When evaluated on the document collection, the proposed methodology achieved 66% of precision on the top 20 extracted keyphrases on average.

Although recent emergency management studies in public administration have focused on collaborative governance networks, little is known about challenges faced by public managers in steering the network of public, private, non-profit, and community actors in international contexts. Towards filling this gap in the literature, this paper focuses on the challenges faced by public managers that are involved in managing post-disaster recovery. It is based on a study of public managers at the National Reconstruction Authority (NRA), the primary governmental body established for post-disaster recovery in the aftermath of the 2015 Nepal earthquake, which claimed more than 9,000 lives and affected 5.6 million people. Based on a comprehensive review of governmental and non-governmental documents and national newspapers, and interviews, this talk argues that public managers face four major difficulties in steering the network actors in post-disaster contexts. These difficulties relate to the political-economic context of the state; formal institutions (rules and regulations related to organizations); organizational capacity (e.g., human resources); and, the interests, priorities, and goals of other public and non-public agencies involved in post-disaster recovery. Based on these findings, this talk offers recommendations on how to address the challenges faced by public managers who manage collaborative governance networks in post-disaster contexts.

Disaster recovery is widely regarded as the least understood phase of the disaster cycle. In particular, the literature around lifeline infrastructure restoration frequently mentions the lack of empirical quantitative data. Despite these limitations, there is a growing body of research modeling lifeline infrastructure restoration using empirical quantitative data. This study reviews this literature and identifies the data collection and usage patterns present across modeling approaches to inform future efforts using empirical quantitative data. The modeling approaches are classified into simulation, optimization, and statistical modeling. The number of publications in this domain has increased over time with the most rapid growth of statistical modeling. Electricity infrastructure restoration is most frequently modeled, followed by the restoration of multiple infrastructures, water infrastructure, and transportation infrastructure. Interdependency between multiple infrastructures is increasingly considered in recent literature. Post-earthquake restoration is most frequently modeled, followed by the restoration after multiple/unspecified hazards, hurricanes, and other wind hazards. The most commonly utilized data type is post-event outage/restoration data, while infrastructure-specific data are less frequently used due to the lack of accessibility. The data are gathered from a variety of sources including collaborations with utility companies, national databases, and post-event damage and restoration reports. This study provides discussion and recommendations around data usage practices within the lifeline restoration modeling field. Following the recommendations would facilitate the development of a community of practice around restoration modeling and provide greater opportunities for future data sharing.

The Natural Hazards Center, in partnership with Save the Children United States (U.S.), is conducting an evaluation of Save the Children’s Building State and Voluntary Organizations Active in Disasters (VOADs) Capacities to Protect Children in Disasters. Two broader goals of this project are to increase VOAD and emergency management knowledge and awareness of children’s needs and to influence the prioritization of children’s needs within these organizations. The larger project includes a range of qualitative and quantitative approaches, but this paper focuses on findings from a social network analysis survey conducted with Arkansas and Nebraska state-level VOADs. A social network analysis was conducted to (1) identify the current types of engagement between and among organizations to better understand the nature of organizational relationships throughout the disaster cycle; and (2) to analyze these existing networks to understand how individuals and organizations are already connected. This allows for identifying opportunities for collaboration among VOAD member and affiliated organizations so as to better prepare for the needs of children before, during, and after disasters and emergencies. This work highlights key findings from these state-level surveys while discussing critical methodological and ethical considerations associated with this approach. At the conclusion of the larger evaluation, the research team will provide Save the Children U.S. with a final report and aid in the creation and review of basic state-level indicators that measure a state’s progress in addressing key recommendations by the National Commission on Children and Disasters as well as a Joint Children’s Needs Response Assessment and Referral Pathways.

In recent years, social disruption and property damage caused by wildfires has grown dramatically, resulting in a number of large-scale evacuations and destruction of housing stock. Despite their growing prevalence, hazards research has not yet investigated the effects of wildfire damage on population change, as it has for other hazards such as hurricanes and flooding. Given this research gap and predictions of increased wildfire intensity under climate change, this talk investigates whether wildfire damage affected migration patterns within the United States since the early 2000s. Employing a difference-in-differences research design that compares fire-damaged counties to neighboring non-damaged counties over time, the authors leverage a natural experiment framing to identify causal effects. Changes in migration flows in response to wildfire damage were examined, comparing several metrics of both migration and damage to do so. For measures of migration, the Internal Revenue Service’s County-to-County dataset (IRS) and the Federal Reserve Bank of New York/Equifax’s Consumer Credit Panel (CCP) are used. For fire damage metrics, the Spatial Hazards Events and Losses Database for the United States (SHELDUS) and the newly mined United States National Incident Command System Incident Status Summary Forms (ICS) are used. The authors expect increased migration out-flows and reduced migration in-flows to counties with high levels of wildfire damage. The authors further anticipate that our results will differ across these four metrics due to differences in their spatial and temporal scales, and hypothesize that the CCP and ICS datasets will show the strongest relationships between fire damage and migration.

With the development of the Internet, social media has become an important channel for posting disaster-related information. While analyzing attitudes hidden in these texts, known as sentiment analysis, is crucial for the government or relief agencies to improve disaster response efficiency, it has not received sufficient attention. This research aims to fill this gap by investigating attitudes towards disaster response on social media using Twitter. The contributions of this research are threefold. First, this research analyzes both the public attitudes towards different types of disasters and the general sentiments of the identical catastrophe over time. For example, Python was implemented to grasp Twitter data and assess the feelings of people quantitatively by these opinionated texts. Second, three machine learning (ML) models were proposed for sentiment prediction including Logistic Regression, Support Vector Machine, and Random Forests. Using machine-learning techniques, a comparison of these models was qualitatively conducted via the prediction accuracy and the optimization method of machine-learning models is discussed from the perspective of model parameters and input data structures. Finally, the demand for different essential relief supplies during disasters is explored based on public opinions on Twitter by focusing on a series of disasters. This not only validates the feasibility of the proposed ML models, but also helps provide decision supports for emergency managers and public officials.

Natural disasters, like hurricanes, are often seen as infrequent events that occur with mostly low intensity. However, due to the exacerbated impact of climate change and increased population growth in coastal areas, more frequent and intense hurricanes in the future can be anticipated. Therefore, special attention should be given towards understanding and predicting hurricane damages. In this study, using survey data collected from households who were affected by Hurricane Sandy in 2012, five types of property damages were investigated. In addition, the effectiveness of hurricane preparedness and housing characteristics were incorporated in explaining variations in property damages. Better understanding of hurricane activity and the determinants of damages can provide individuals, policymakers, and insurance agencies the proper way to prepare for future hurricanes. This study emphasizes the role of disaster preparedness efforts in minimizing exposure and promoting the effectiveness of hurricane mitigation.

County-level post-disaster power outage data provide high temporal resolution (two to three hours) snapshots of the number of household accounts without power. These data have been used to derive electrical power restoration rates and map the spatial extent of damage. Yet, such publicly available datasets give little insight into more granular levels of detail for answering questions such as who is the most impacted by extended power outages or where the impact is felt the most. Big data acquired through satellite imagery can provide high-quality, scalable, and timely information on power outage. In this study, the extent of disrupted electric utility infrastructure in the Florida Panhandle following Hurricane Michael was estimated using both county-level power outage data and NASA’s Visible Infrared Imaging Radiometer Suite (VIIRS) Day/Night Band (DNB). Additionally, the 2013-2017 American Community Survey estimates at the block group level were used to downscale the percent loss of electrical service to a neighborhood level. Preliminary results indicate that the longest power outage duration was observed in the rural Calhoun and Jackson counties, followed by Bay and Gulf counties that experienced a direct hit by the hurricane. The results also reveal an uneven distribution of service loss and disparities in power restoration within each county. The use of near-real-time satellite imagery can address important ethical dilemmas related to the timely delivery of emergency services to the most impacted households and communities, and shorten the feedback loop between post-disaster damage assessments, policy formulation of restoration priorities, and community recovery plans.

The pedagogical space for preparing the public for disasters is extensive and includes not only school-based initiatives and public information campaigns, but also family and community learning, adult education, and popular culture also known as public pedagogies. Moreover, with technological developments such as social media, citizen journalism, and blogging, there are increasingly sophisticated ways for citizens to access disaster information. To facilitate decision-making during disasters, a novel people-centered approach is needed to incorporate residents' perspectives and expectations, while authorities become more competent communicators willing to engage in long-term dialogues with communities. This is particularly critical in disaster-prone areas such as Miami, Florida where the population has increased ten percent in the last eight years as those new residents require basic knowledge to prepare for hurricanes. Despite a few recent initiatives, little is known about leveraging different strategies to benefit community-centered learning in disaster contexts. In this study, a survey of south Florida residents was conducted to reveal local community-level preferences for receiving disaster preparedness training along with different mechanisms of educational techniques. An ordered probit regression model is used to determine the most efficient method of disaster training. While one-third of the respondents had previous training of some sort, the rest reported no prior experience and reported the necessity of having some form of training. Online learning in video formats was mostly preferred over face-to-face learning. Results also indicate preferences over medium to longer spans of training rather than shorter ones.

The increasing convergence of inland and coastal communities through coastal erosion, sea-level rise, and rural to urban migration is a phenomenon that Louisiana is experiencing on an unparalleled scale although the issue is not unique to the state. Following recent storms and aided by federal disaster funds, many residents moved away from the coast to reduce their risk—only to be flooded repeatedly from rain events over inland watersheds. This study seeks to understand the multi-scalar connection between government policies and local adaptation practices by asking (1) How do government policies, like the National Flood Insurance Program, influence local building practices and community wellbeing in southeast Louisiana? (2) How can a greater understanding of environmental risk and wellbeing increase adaptive capacity in coupled inland-coastal regions? and (3) How can wellbeing and adaptation scholarship be better incorporated into development decision-making at all levels to bolster community resilience? To answer these questions the study used policy and document analyses, focus groups, and interviews with residents, elected officials, and design professionals. The project also incorporates community-based planning and design processes that allow residents to prioritize elements necessary for wellbeing to improve resilience even as climate risks increase. Initial findings show stakeholders view government policy as increasing future flood risk for inland communities by incentivizing development practices that can be detrimental to existing neighborhoods. This suggests the need for aligning government policies with development practices that can bolster long-term community resiliency in floodplains.

Disasters occur frequently in Japan but some residents cannot or do not evacuate to shelters. The purpose of this study is to link the lifestyle custom and hobby activities to the community disaster prevention plans. By Conducting an interview survey with many residents the relationship of customs and hobbies with acknowledgment of community disaster prevention plans were identified. The study area is the Funada district of the Kihou town in Mie prefecture. Interview surveys show that residents who have a custom to talk with many others tend to recognize the community disaster prevention plan. Findings also indicate that residents can understand the community disaster prevention plan better when it is combined with an everyday custom and hobby. This joint event was planned by residents and held as a monthly walking and cooking class from June to December 2019. A leader and an assistant leader, from the study area, would decide on a walking route and the recipes of dishes. They decided on a monthly cooking menu, and the cooking class ingredients to purchase and the ingredients they could harvest locally. In addition, a workshop and survey about community disaster prevention plans were conducted at several such events. The analysis of the survey data shows that communication about community disaster prevention plans increased among these groups over time. This study demonstrates the potential for raising awareness about community disaster prevention plans by making every day and non-everyday actions seamless in community education efforts.

The current standard practice of measuring disaster preparedness in the United States is to measure disaster preparedness as an individual household attribute based on the amount of stocked supply, hazard mitigation actions, and planning in place. Individual households are expected to survive on their own for at least 72 hours or more without power, food, or transportation following a disaster. This study argues that social capital is a critical resource in disaster response and therefore should be included in measurements of disaster preparedness. This research outlines the reasons for measuring social capital in disaster preparedness and tests the results of doing this. Accounting for social capital explicitly in disaster preparedness requires expanding the concept and measures of disaster preparedness to include social capital measurements; called community disaster preparedness (CDP) in this presentation, as opposed to individual household disaster preparedness (IDP), which the Federal Emergency Management Agency currently measures. This research hypothesizes that CDP is greater than IDP, especially for tightly knit communities identifiable by race and ethnicity, even after controlling for income. Also, social capital is assessed and its association with both CDP and IDP are examined. This study uses a combination of survey strategies, including a mail survey of a random sample of households in Seattle stratified by zip codes, a convenience sample through a network of ethnic health boards in Seattle/King County, and Google paywall intercept surveys in Washington State.

To get to hope, people must go through loss and grapple with change. Working in this field has allowed disaster researchers to accompany communities as they traverse painful post-disaster landscapes. Researchers find hope in the responses that grieving loss, resisting, and then accepting change invokes. These collective experiences have coalesced into a strengthened sense of commitment to the places, cultures, and ecosystems that inspire, nurture, and sustain our work. Hope is found in 1. The ideas that shaped the field and the first generation of disaster researchers that left the current generation with an interdisciplinary canon, tools to teach others, and change practice. 2. Understanding inconsistencies and contradictions in current neoliberal policies, models of development, and the imbalances of power that create disasters. 3. Being armed with the knowledge that current researchers can address these challenges and affect change. 4. Having clear ideas of what altered approaches, goals, and values look like. 5. A network of scholars and practitioners that are allies in the effort to advocate for change and lift up the voiceless. 6. The ability to operationalize ideas as tools to change practice. 7. Watching students internalize the call to action and have a clear idea of what needs to be done.

The construction industry experiences extreme weather events every year. Decision-making in construction sites under extreme weather events will be hampered by unknowns and uncertainties and will pose difficulties for construction managers, stakeholders, contractors, subcontractors, suppliers, workers, and policymakers. Those difficulties such as worker safety in such events have not received adequate attention, and the existing literature does not provide enough guidance on the sources, types, and efficacy of warning information available to construction workers. The goal of the study that will be presented in this talk was to create a multi-disciplinary decision support framework based on the critical factors contributing towards the disaster resilience capacity, measured as robustness and rapidity, of active major construction sites. This research aimed to (a) create and analyze metrics and performance measures related to site robustness and rapidity; (b) provide policy insights to construction stakeholders to enhance resilience capacity in future construction activities; and (c) recommend better ways of risk assessment and information dissemination to workers on-site and identify consequences of these improvements in construction scheduling. The study findings are based on focus group interviews of researchers, experts, and practitioners actively involved in construction activities. This study is one of the first few studies that consider the notion of construction site resilience linking three phases of disasters: preparedness, response, and recovery.

Humans have a remarkable capacity for resilience and growth after stressful and traumatic experiences, including pandemics. Understanding psychological and environmental factors that promote resilience or psychological distress during or after a pandemic will facilitate prevention efforts, intervention strategies, and research programs. Therefore, we examined various factors theorized to promote and inhibit resilience and symptomatic trajectories during the COVID-19 pandemic in a sample of international and non-international students at our medium-sized university in the southern United States. Factors theorized to promote resilience were social support and meaning in life. Meaning in life is the perception that one’s life is significant and is positively related to psychological well-being. While stressful or traumatic life events disrupt one’s worldview, individuals make meaning out of stressful events to experience growth and reductions in psychological distress. Social support, too, has been consistently implicated as a positive predictor of resilience. Because social support and meaning in life can fluctuate rapidly based on naturalistic environmental changes, a tremendous gap in the literature is an understanding of how social support and meaning-making might impact resilience or distress during the pandemic. Beyond promoting psychological resilience, it is also vital to identify psychological factors that promote effective disaster preparedness and prevention behaviors during the pandemic. Using the Extended Parallel Process Model framework, we also examined the impact on weekly changes in self-efficacy and threat perception on disaster preparedness. Employing a prospective longitudinal design, students completed eight weekly assessments and were compensated $5 for each completed assessment. Latent Growth Mixture Modeling and Multilevel Modeling were used to (1) differentiate trajectories of psychological resilience and distress, (2) differentiate trajectories of disaster preparedness, and (3) examine predictors of these trajectories over the 8-week study period. Data collection is ongoing; therefore, the presentation will focus on the theoretical basis of the present study.

The National Science Foundation-funded Social Science Extreme Events Research (SSEER) network was formed, in part, to respond to the need for more specific information about the status and expertise of the social science hazards and disaster research workforce. Core to the mission of SSEER is to identify and map social scientists involved in hazards and disaster research in order to highlight their expertise and connect social science researchers to one another, to interdisciplinary teams, and to communities at risk to and affected by disasters. Ultimately, the goal of SSEER is to amplify the contributions of social scientists, to advance the field through expanding the available social science evidence base, and to enhance collective wellbeing. This article describes an ongoing SSEER effort to generate the first-ever census of social scientists who study hazards and disasters. Drawing on data from 1,013 SSEER members located across five United Nations regions, this article reports on the demographic characteristics of SSEER researchers; provides a novel inventory of methods used by social science hazards and disaster researchers; and explores how methodological approaches vary by specific researcher attributes including discipline, professional status, researcher type based on level of involvement in the field, hazard/disaster type studied, and disaster phase studied. This analysis showed that SSEER respondents range in age from 21 to 78 years old; 52.02% identify as female; 59.62% identify racially as White; 62.59% report having earned a doctoral degree; and 57.85% work in an academic institution. Most SSEER respondents described themselves as core researchers (45.11%), meaning they view themselves as highly involved in and deeply committed to hazards and disaster research. About an equal number of respondents identified as emerging researchers (23.99%) and periodic researchers (20.04%), with a smaller percentage indicating that they are situational researchers (6.22%). In terms of research methods, case study (60.41%) was the most often selected approach among SSEER researchers, followed by survey research (59.53%) and in-depth interviews (58.74%). Just over 80% of SSEER respondents said that they use four or more primary methods in their hazards and disaster research. The SSEER initiative responds to calls to more systematically characterize the social science hazards and disaster research workforce. The results have implications for training, mentoring, and workforce development initiatives geared toward ensuring that a diverse next generation of social science researchers is prepared to study the root causes and social consequences of disasters.

The goal of this paper is twofold: to clarify the tenets of convergence research and to motivate such research in the hazards and disaster field. Here, convergence research is defined as an approach to knowledge production and action that involves diverse teams working together in novel ways—transcending disciplinary and organizational boundaries—to address vexing social, economic, environmental, and technical challenges in an effort to reduce disaster losses and promote collective wellbeing. The increasing frequency and intensity of disasters coupled with the growth of the field suggests an urgent need for a more coherent approach to help guide what we study, who we study, how we conduct studies, and who is involved in the research process itself. This article is written through the lens of the activities of the National Science Foundation-supported CONVERGE facility—which was established in 2018 as the first social science-led component of the Natural Hazards Engineering Research Infrastructure (NHERI). Convergence principles and the Science of Team Science undergird the work of CONVERGE, which brings together networks of researchers from geotechnical engineering, the social sciences, structural engineering, nearshore science, operations and systems engineering, environmental engineering, and interdisciplinary science and engineering. CONVERGE supports and advances research that is conceptually integrative, and this article describes a convergence framework that includes the following elements: (1) identifying researchers; (2) educating and training researchers; (3) setting a convergence research agenda that is problem-focused and solutions-based; (4) connecting researchers and coordinating functionally and demographically diverse research teams; and (5) funding convergence research, data collection, data sharing, and solutions implementation.

Many universities have closed their campus residences since the COVID-19 pandemic started. The consequences may be staggering for international/out-of-state students, a vulnerable but often neglected population. Their evictions may also impact the public’s health due to an increase in air travel. We document and compare the experiences and coping strategies of evicted international/out-of-state university students in the US and Canada. The relevance of this research to disaster studies is that it sheds light into the universities’ organizational responses to disasters, and how these converge to impact its internal and external populations.

Technological advances within the last decade have generated new avenues for conducting research, including the analysis of “big data” from social media and new online communications platforms. These developments in communications also benefit disaster volunteerism and emergent behavior by facilitating volunteering, organizational decision-making and coordination across great distances. Beginning in 2017 following Hurricane Harvey, the authors began studying many of these emergent groups that conduct volunteer rescues during disasters using social media and smartphone apps. Such groups are composed of organized volunteers who self-deploy to areas impacted by disasters, most typically flooding events, to rescue stranded or distressed individuals. This project was expanded in 2019 by using traditional ethnographic research techniques and new virtual ethnographic techniques through organizational group communications platforms. Within the communications platforms ethnographic observation of organizational decision-making and coordination were conducted, particularly relating to volunteer management in conducting rescue activities. This type of data collection is beneficial in that it allows researchers to observe and collect data without having to immediately converge upon the site of a disaster, a practice that has raised ethical questions about the impact an influx of researchers may have on local populations. However, conducting ethnography in these spaces also generates new ethical questions and challenges including how to maintain informed consent and respect privacy concerns. This presentation focuses on this novel data collection practice for disaster scholars and the ethical considerations around crowdsourcing and virtual ethnography, which have yet to be explored.

The frequency and magnitude of hazards and associated risk in modern society can be exacerbated by globalization and environmental change at local, regional, and global levels. Risk perception and related behavior constitute a fundamental theme in risk analysis. Despite the inherent dynamic nature of risk events, the temporal dimension of risk perception and behavior has been understudied in the current risk science literature. Longitudinal research design is largely lacking in this field as previous studies mostly used cross-sectional data. Infectious disease outbreaks provide a key setting for analyzing the changing perception of and response to natural or human-induced hazards. In order to examine the dynamic risk perception and behavior in response to the Coronavirus Disease 2019 (COVID-19), we conducted a series of three online surveys of residents in four major US cities (Seattle, Los Angeles, Chicago, and New York City) during March – June, 2020. This presentation will share preliminary findings on temporal changes in perceived COVID-19 risk and the dynamic relationship between risk perception and behavior response.

Twitter is a popular platform for understanding and analyzing human socio-behavioral dynamics during natural hazards and crisis events because of the high volume of content and relative ease of access. Twitter's Terms of Service, however, restrict the quantity and types of data which can be collected from the platform. Accordingly, there are multiple ways to retrieve data from Twitter with no consensus among researchers as to standard data collection procedures. In this work, the authors compare two tweet datasets gathered around Hurricane Harvey via different criteria and show that many types of common social media analyses are significantly dependent on how Twitter data is collected. This study finds significant differences in the user base, user behavior, social network, and content of tweets between the two databases and illustrates how these differences impact resulting analyses. The authors provide guidelines for tweet sourcing for different methods of social media analysis and advocate for increased transparency and record-keeping surrounding the sources of social media data so as to reduce bias and increase transparency.

Climate Change is a phenomenon that intersects social, economic, political, and environmental challenges. Countries and Nations must face such challenges that are inherently multifaceted. In this regard, achieving gender equality has a unique role in the global multilateralist agenda. Currently, scholars know that vulnerability has a noteworthy gap when it is gendered: women suffer the most when there are natural and human-made disasters. However, less is known about how gender equality influences countries' overall vulnerability and fosters adaptation and resilience. The relevance of studying gender equality and climate change resides in the determination of its role concerning vulnerability and in fostering resilience among societies around the world. The research question that guides this study is: "How gender equality influence countries' adaptive capacity to climate change consequences?" The hypothesis tested suggests that the more gender equality in a State, the more adaptive capacity to the negative consequences of climate change. This study runs a multivariate regression analysis with five variables: adaptive capacity, gender equality, GDP, population, and conflict. Adaptive capacity is operationalized with the ND-Gain Index, while gender equality is reflected in the UNDP's Gender Inequality Index. Both variables correlated for the specific periods of 2000, 2010, and 2017. Results show that the relationship between the dependent and independent variables is statistically significant, and there is a strong association. On a scale 0 to 1, a 0.1 unit increase in gender inequality predicts a decrease in adaptive capacity by 5.18 units for the year 2000, 5.12 for 2010, and 5.29 for 2017. This study contributes to current knowledge by demonstrating that more equal societies are better adapted and, therefore, less vulnerable to adverse impacts of climate change.

Much data is collected before, during, and after disasters by numerous agencies, organizations, and private citizens. Additionally, data that are collected by researchers are typically heterogeneous and encompass the physical, chemical, biological, social, economic, and health, among dimensions and are collected using various methods, criteria, and data quality control and objectives. This presentation will focus on how data are used to arrive at scientific conclusions and will elucidate how these conclusions compare to reality and the perception of reality by affected communities. Data from Hurricane Harvey related to environmental, social, economic, and community health hazards and risks are used to illustrate the ethical boundaries that may be at play. The opportunities to develop and leverage tools that can close this big data divide will be discussed.

Disasters tend to magnify pre-existing inequalities, with women often more likely to face more severe consequences due to gendered inequalities. American households with children under 18 continue to have a gendered division of labor in which women do more domestic work, particularly tasks related to children’s health and education. This is the case even though a majority of women with children under 18 are working. The COVID-19 pandemic is a disaster that has resulted in school closures as well as many people working from home. Due to social distancing requirements, many parents do not currently have childcare. This study seeks to begin investigating the gendered consequences of the COVID-19 pandemic in the US. In particular, it will investigate how the pandemic is affecting the gendered division of household labor among families with children under 18. The study will use an online survey and in-depth interviews with men and women who have children under 18 in order to understand their experiences. This presentation will discuss the study design and preliminary results.

Urban growth patterns, geophysical character, and proximity to the Gulf of Mexico have rendered Houston, Texas one of the most flood-prone cities in the United States. In this study plans that drive growth decisions in Houston were analyzed as a critical first step to identifying policies that increase vulnerability to flooding by failing to protect critical resources or direct development away from flood-prone areas. Given the demographic and socio-economic diversity in Houston, a second concern of the study was whether plans are addressing vulnerability equitably. This study focused on three high-income super-neighborhoods in the west and three low-income super-neighborhoods in the north-east of Houston to examine (1) the extent to which the plans in the six neighborhoods consistently reducing vulnerability to flooding and (2) the differences in the ways that plans address vulnerability in low-income versus high-income neighborhoods. Using the Plan Integration for Resilience Scorecard method, content analysis was conducted on 35 plans from 2002 to 2019, including 17 neighborhood-level plans, six city-level plans, and 12 regional plans. Consistent with the literature, the Hazard Mitigation Plan, and post- Hurricane Harvey plans have more policies for reducing vulnerability compared to other plans. Preliminary analysis also suggests a lack of consistency in policy intent to address flood vulnerability between neighborhood scale and city-regional scale. This finding is consistent across all six neighborhoods. Moreover, while a large number of policies aim to reduce vulnerability across the city, few policies focus on the needs of the low-income neighborhoods, eliciting the need for more equitable policies.

The growing cost of climate-driven coastal impacts requires an improved understanding of how coastal populations engage with adaptation decisions. While studies explore factors driving coastal adaptation, generally, few evaluate how residents consider relationships between in situ protective adaptation versus retreat from at-risk areas. This presentation addresses that gap by posing and responding to the following questions: What is the relationship between residents’ exposure, perceptions of climate trends, and concerns about the future? How do these factors influence attitudes and openness to different adaptation strategies? Are these strategies considered to be progressive – where protection is indexed to minor threats and retreat occurs when protection measures fail–or are these dichotomous choices? In this study structural equation modeling is applied to evaluate these decision pathways using a 2017 household survey in North Carolina’s Albemarle-Pamlico Peninsula (n=147). The results reveal that residents commonly view protection and retreat as mutually exclusive, rather than progressive methods for reducing risk, and that their preferences are correlated with different understandings of climate threats.

Hurricanes can cause significant damage to housing in high-risk areas (for example SouthEast United States and Northern Australia). Such damage can lead to significant monetary loss and negative mental health outcomes. There is a range of engineering solutions (for example hurricane shutters and roof upgrades) that can reduce this damage and, therefore, mitigate the negative knock-on effects. However, the uptake of hurricane-related mitigation measures in hurricane-prone regions has been low. To promote hurricane mitigation behavior in high-risk areas it is important to understand what facilitates or impedes an individual to perform such behaviors. This talk will discuss the findings from a research project investigating the psychological factors that explain why people do, or do not, prepare for hurricanes. This project involved surveying residents of North Queensland, a hurricane-prone area of Australia, using an online questionnaire. Based on well-researched theories of human behavior, the questionnaire assessed a range of psychological factors and respondents’ intentions to perform a range of hurricane mitigation behaviors. The results of the study suggest that thinking and talking about hurricanes, known as hazard intrusiveness, leads to mitigation behavior but risk perception does not. It was also found that perceiving a mitigation measure as having secondary benefits (for example increasing property value) was more important than perceiving primary benefits (for example reducing damage). The results show that it is important to adapt current psychological theories to explain adoption of relatively high-cost hurricane mitigation behaviors. The findings also suggest that promoting hurricane mitigation behavior requires changes to both messaging and policy.

This pilot study will examine the economic and social consequences of COVID-19 as they intersect with questions of racial stigma and racial violence that disproportionately affects AAPI, an ethnically and socioeconomically diverse demographic that nonetheless frequently shares overlapping experiences of racism and xenophobia across multiple ethnic categories. We will conduct a general survey and interview small business owners, gig economy workers, low-wage laborers in “essential industries,” individuals working in intimate labor, and healthcare and other paid care workers. We will also interview members of community-based organizations that are mobilizing to support AAPI businesses and workers disproportionately affected by the pandemic.

The disaster risk management community is steadily expanding its usage of artificial intelligence and machine learning technologies for the creation and analysis of disaster risk and impact information. Limited attention has been paid to questions of fairness, ethics, and biases of these systems as compared to other domains such as facial recognition, search algorithms, and criminal justice, where significant problems have been identified by scholars, activists, and journalists. To address this gap, this project convened a working group on the topic of "Responsible AI for Disaster Risk Management." The group consisted of over 30 practitioners and researchers working on the use of artificial intelligence tools in this field. The working group met over a period of four months to examine case studies of how these tools are being used and discuss examples in adjacent domains where AI was shown to have negative social impacts. The authors also conducted one-on-one interviews with members of the group and other experts. The authors then collaboratively drafted the findings, which include potential harms that might arise from the use of AI in disasters, strategies for mitigating these harms, and guidelines for the effective and responsible use of AI in disaster risk management. This talk presents, for feedback and discussion, the results of this work as well as lessons learned from facilitating community discussion around the ethical uses of technology for understanding disaster risk and impacts.

Rural areas in Japan have suffered from their population recovery, especially in the impacted region from the Great East Japan earthquake and tsunami in 2011. Previous disaster studies mainly focus on post-disaster unstable situations and fewer studies focus on long-term social changes between pre-disaster stable situations and post-disaster situations. This makes it difficult to evaluate the long-term impact by earthquakes and the following reconstruction projects and, therefore, to plan reconstruction projects. This study quantitatively identifies what social space changes may happen in regions after disasters, including damages and reconstruction projects, using the case study of the 1995 Kobe earthquake. This study defines social space as statistical resident characteristics calculated by cluster-analysis and principal component analysis using 50 indicators from the national census in Japan. To achieve this purpose, this study statistically and chronologically compares the social space changes between impacted regions and the standard trends from the whole of Japan. The results show that 1) damages and reconstruction projects caused short-term impact in demographic transition, but long-term social spaces changes were closer to transitions predicted from the pre-disaster situation and nationwide trends rather than disaster-related factors and 2) the disaster caused less impact to demographic trends in urban areas and more impact in suburban areas in pre-disaster problem areas, such as aging and population decline. These results explain that with reference to pre-disaster situations, we may predict post-disaster long-term situations before an event. This study proposes that it is necessary to consider post-disaster situations based on pre-disaster situations for planning long-term reconstruction.

Widespread disasters can overload official agencies’ capacity to provide assistance and often community members emerge to assist with disaster response. This was the case with 2017’s Hurricane Harvey that devastated Greater Houston. This study seeks to more fully understand how rescuees and volunteer rescuers use mobile and social media during rescues. The guiding research question for this talk is: How can we use photo-elicitation interviews to gather rich forms of data from private social media sources that illustrate the rescue process? The data collection method used for Phase I of this National Science Foundation-funded study was a photo-elicitation interview (PEI) method), an approach that allowed the authors to directly elicit private social media data while interviewing rescuees and rescuers. The research team collected a total of 1,212 images (from Facebook, WeChat, WhatsApp, Snapchat, Nextdoor, and GoogleDocs from 44 interviews). A series of qualitative analyses was also conducted to better understand our findings from Phase I. Those images were then used in a series of machine learning studies during Phase II of our study to evaluate automated detection of the features of images that characterize flooding rescues. To address our method-focused research question, this talk shares examples of the type of data gathered using this method. Specifically, the talk focuses on a subset of data illustrating how members of a community not only physically rescued flooding victims, but also how they worked behind the scenes to provide vital information. The talk will conclude with participants practicing this method.

Mobile homes are the largest source of unsubsidized affordable housing in the United States. Yet, there are few studies that examine the spatial distribution of mobile home parks in metropolitan regions or the environmental inequalities that residents may face. This paper presents the findings of one of the first, and most comprehensive, studies of mobile home parks in a U.S. metropolitan area. We inventory and spatially analyze mobile home parks in the greater Houston Metropolitan Statistical Area, a 9-county region with a population of 6.9 million people. We locate 2,126 mobile home parks that include over 30,000 units of housing. We then compare geographical areas with higher numbers of mobile home parks to those with fewer, in terms of their socio-demographic characteristics; access to amenities; and exposure to environmental hazards. We find that mobile home parks are concentrated in geographic areas with greater concentrations of households in poverty and with other characteristics associated with vulnerability, less infrastructure, and greater exposure to flood hazards, compared to those areas with fewer mobile home parks. Yet these places do not have the same level of services as other municipalities. We conclude by discussing the planning and policy implications of our findings.

The limited existing studies on rural resilience largely point to rural communities being vulnerable due to single sector dependence, and less access to education, employment, and healthcare. However, the degree to which these factors, or others, reduce rural resilience, and how they differ from urban resilience, are largely unknown. After an EF4 tornado impacted a 32-mile stretch of Northeastern Kansas on May 28, 2019, a team of local researchers dispersed to document the perishable physical damage data. The impacted area was primarily rural, with unpaved roads reaching out to some residents. Only three days after the tornado, many residents were displaced; therefore, accessibility and trespassing issues became an important variable for the research team. Longitudinal studies are imperative to understanding which fundamental factors drive rural resilience. Thus, the research team re-engaged six months after the tornado to document repair progress, and to survey households to understand their recovery experiences. However, during both outings, large dogs, KEEP OUT signs, and NO TRESPASSING signs were frequently encountered leaving the research team questioning if efforts were misplaced. Though the study has gained positive local attention, the researchers were left pressed with ethical and methodological considerations on when to deploy, who to collect data from, and how to gather that data. This presentation will discuss these issues, along with follow-up discussions on what happened next, including communicating lessons learned back to the public and the process of de-sensitizing human subjects data.

There is a long history of the social sciences documenting and measuring which factors and to what extent those factors contribute to social vulnerability. These studies have captured individual social vulnerability factors at single instances in time. To understand the dynamic nature of social vulnerability, this presentation leverages a longitudinal study of household-level data collected in a community impacted by disaster. The longitudinal study takes place in Lumberton, North Carolina, and was initiated in 2016 after Hurricane Matthew. The larger project team has executed systematic housing disruption and recovery surveys with the same households four times over four years. This presentation focuses on data collected during the first two years, immediately after Hurricane Matthew and approximately one year later. When a community experiences a major disaster, its population stability is put at risk. Quite often a portion of the population permanently leaves. This presentation aims to answer questions, such as (a) Which social vulnerability factors influence a household’s decision to stay? (b) When a community is impacted by disaster, how does its social vulnerability profile change? and (c) With hazard exposure being pre-disaster, initial damage being immediate, and dislocation, repair, and recovery being longer-term outcomes, how do social vulnerability factors contribute to our dependent variables occurring at different stages of the disaster? Correlation testing and regression analysis are executed to answer these questions, through examining social vulnerability indicators and their intersection, including race, ethnicity, income, tenancy, household structure, insurance, and housing type, and through examining a series of dependent variables.

Pre-disaster damage predictions and post-disaster damage assessments are challenging since they result from complicated interactions between multiple drivers, including exposure to various hazards, structural resilience, and demographic characteristics. This study evaluated the contribution of eight drivers of structural damage due to Hurricane María in Puerto Rico, leveraging machine learning algorithms to determine the role that human factors played. These algorithms are well suited to parsing out the role of social indicators in disaster studies because they are readily interpretable, accommodate large datasets from a variety of different sources and formats, and provide quantifiable measures of variable importance. Random Forest and Stochastic Gradient Boosting Trees algorithms analyzed a rich set of publicly available data including wind, flooding, landslides, and vulnerability measures. Algorithms were trained with these data to predict the structural damage caused by Hurricane María in Puerto Rico and the importance of each variable was calculated from the top-performing models. Results indicated that vulnerability measures were the leading predictors of damage, followed by wind, flood, and landslide measures. Each predictive variable exhibited unique, often nonlinear, relationships with damage. These results demonstrate that social vulnerabilities play critical roles in damage pattern analysis. Furthermore, they indicate that targeted, pre-disaster mitigation efforts should be enacted to reinforce household resilience in socio-economically vulnerable areas. Recovery programs should focus on the highly impacted vulnerable populations to avoid the persistence or exacerbation of pre-existing inequalities in the wake of a disaster.

Data collection and fieldwork in the aftermath of a disaster is an essential part of hazards and disaster research. However, there are numerous ethical, physical, and psychological complexities with such research activities. This was recently highlighted in a Nature article by JC Gaillard and Lori Peek: “Disaster-Zone Research Needs a Code of Conduct.” While this may require more work and training activities for research staff, such protocols can improve the safety and security of researchers in the field, improve data quality and project completion rates, promote better research ethics in challenging contexts, and ultimately improve outcomes for both the research and the communities affected. Through a literature review focused on Southeast Asia, interviews with stakeholders and community members, case-study development, reviews of existing pre-research training and consultations with researchers and humanitarian organizations in the region, the authors are developing a protocol and training that addresses questions surrounding data ethics, the involvement of local collaborators, researcher trauma, and ethical interaction with affected communities. This research offers one articulation of a code of conduct and a set of processes that will enable ethical and safe fieldwork in post-disaster contexts. Given the research-base of Singapore, this research is focused on Southeast Asia and the particular dynamics that animate post-disaster research in the region. This talk demonstrates both the need for and the processes that enable (1) a non-extractive ‘community first’ approach to post-disaster settings; (2) sensitive collection and handling of data; and (3) involving, acknowledging, and benefitting local collaborators.

It is widely accepted that levels of emergency preparedness and resilience differ based on individual and community-level factors. What is less clear, however, is how various types of communities (i.e. urban, rural, marine dependent, in-land) differ in these functions and how individual, community, and institutional characteristics interact to create overall levels of risk and preparedness to natural hazard events. This research uses a mixed-methods approach to understand the interaction of individual, community, and institutional components and applies it to the case of communities in British Columbia, Canada. Using census data, collected by Statistics Canada, our results indicate that key differences are found in the capacity to respond to small and large scale events. While smaller communities may have more experience and are better prepared to handle short-term disruptions, as compared to large population centers, they are found to be less prepared to handle significant disruptions, particularly those with high financial costs. Data gathered from workshops and interviews additionally finds community differences, particularly in the capacity to prepare for emergencies and infrastructure capabilities (e.g. the number of local critical facilities). At the institutional level, sense of responsibility and coordination were found to differ based on community type. While many of these results support previous findings, they draw attention to the need for diverse perspectives that include many sectoral and disciplined backgrounds to make sense of what drives community and individual preparedness and resilience. It is the combined factors that allow for a full understanding of regional emergency preparedness and resilience to hazards.

The rapid spread of COVID-19 requires decision-makers to act quickly to reduce the risk of disease spread. The adverse impacts of government-mandated social distancing measures on individuals are generally unknown. Vulnerable populations, such as low-income individuals, face unique challenges managing their safety, economic means, and autonomy in response to the policies implemented on local levels to curb the spread of COVID-19. This presentation will discuss aims to understand the invisible variables that contribute to vulnerable individual’s personal security during the hardships imposed by the COVID-19 pandemic and institutional response.

As record-breaking natural hazard events continue to occur the impetus for finding scientifically-supported strategies for response and recovery practices is growing. This often means that teams of researchers from physical and social science backgrounds are entering the field in the immediate aftermath of a disaster to collect perishable data. This influx of researchers creates an increased potential for harm, as researchers are intruding on the lives of overwhelmed and vulnerable people and communities. This perishable data, however, is critical to collect if research is going to play a role in reducing the impacts of similar future events (i.e., benefit society). Recently, there have been calls for a code of conduct for disaster researchers. This project builds on these calls to illustrate how ethical considerations in the research design phase can address some of the shortcomings of post-disaster field reconnaissance. To do this, we describe the research methods of a disaster recovery project, Building Resilience and Innovation through Diverse Group Engagement (BRIDGE), conducted in Robeson County, North Carolina, and highlight how prioritizing ethics resulted in a design that minimized harm and maximized benefits. The methods discussed include prioritizing community recovery needs, merging social and physical science research projects, giving power to the participants during interviews, working with local citizens in the research process, and connecting participants to resources. The results do not suggest specific methods, but rather, illustrate a process for making ethically-informed decisions related to disaster research design.

People with intellectual and developmental disabilities (PWIDD) face unique risks during emergencies such as the current COVID-19 pandemic. Similarly, direct support professionals (DSPs) who support individuals with disabilities may also face substantial risks. To understand how the pandemic has impacted quality of life (QoL) among PWIDD and DSPs in the US, we conducted a rapid mixed methods investigation. We used a retrospective pretest approach to survey PWIDD and DSPs about their QoL before and during the pandemic, and a rapid qualitative inquiry approach to conducting dyadic interviews with PWD and their DSPs about changes in their lives as a result of the pandemic, how they have worked together to maximize QoL during the pandemic, and resources they have used or needed during this time. Preliminary research findings will be shared, along with recommendations for resources, policies, and best practices that will improve health and quality of life for PWIDD and DSPs in similar infectious disease emergencies.

Framing disasters as something natural has the potential to absolve those creating risk of all responsibility or create a convenient scapegoat, nature. This can lead to policy and practice that perpetuate a status quo determined to use technocratic measures to reduce risk and build resilience. This talk shares findings from a research project using an exploratory correlational experiment tool through Project Implicit with a sample of more than 400 individuals to demonstrate the impact of understanding disasters as natural rather than socially-constructed. The findings provide support for the assertion—made in previous systematic review of disaster literature by the authors—that failure to recognize social/political/economic root causes of disasters serves to protect powerful interests and works counter to systemic change.

Flooding is the costliest natural disaster in the United States. Although federal governments play an essential role in the hazard mitigation process, the dominant functional responsibility for reducing hazardous risk lies at state and local levels. Following the Great 1993 Midwest Flood, the Federal Emergency Management Agency (FEMA) introduced the property buyout program as a mitigation strategy to reduce the flood risk in the United States. While the buyout program has been in place for several decades, its effectiveness from the community perspective has received little attention in the research. This study aimed to fill this gap by examining the internal and external factors to the adoption of FEMA’s property buyout program in the State of Virginia. To this day, 299 buyout projects have been implemented across the Virginia flood-prone communities. The diverse characteristics of these adopters and the non-adopter (communities with flood hazards, but without buyout projects) present an ideal setting to study the proposed research questions. Data includes a survey of 296 local floodplain managers and state officials in the Commonwealth, American Community Survey socio-demographic and socio-economic data, the FEMA buyout database, and interviews of key informants. Qualitative and quantitative methods are applied to identify the factors. Preliminary analysis showed that internal factors that influence the buyout policy adoption of local governments include flood hazard exposure, property damage, institutional capacity, and social-demographic and economic characters. The external factors include networking, policy diffusion, and upper-level policy environment. This talk will also provide recommendations for future hazard mitigation policy adoption.

The authors seek to contribute to the understanding of post-disaster community interdependencies by examining the recovery linkages between businesses and households. Specifically, this research looks to identify the household dimensions that affect recovery quality in businesses in Lumberton, North Carolina after 2016 Hurricane Matthew. Through an interdisciplinary field study 15 months after the hurricane, businesses were asked about loss of customers and various labor disruptions they experienced. Logistic regression used to examine the impact of these variables on the likelihood of a business reporting being fully recovered, controlling for damage, accessibility issues, business characteristics, owner or manager demographics, and financial assistance. This research found that customer loss in particular had a higher effect magnitude than initial damage in terms of hindering recovery. Labor disruption caused by transportation issues and childcare or school closure issues had a smaller relative effect, but also significantly lowered a business’s odds of full recovery. Not all labor variables were significant—including employee personal household damage—stressing the importance of survey data to understand the specific dimensions of households affecting business recovery.

Natural hazards are an inherently intergovernmental problem, requiring coordination of local, state, and federal governments. Scholars have examined public preferences for different levels of government intervention for various hazards, but usually in isolation. This project takes a more all-hazards approach to the United States federalist system and examines preferences for government—both state vs. federal—risk management for wildfires, earthquakes, and climate change. These three hazards vary on key characteristics such as duration, predictability of onset, loss of life, among others. Using survey data from Oklahoma and seemingly unrelated regression techniques, which account for correlation between the preferences for all hazards, this study finds that respondents prefer state intervention for wildfires and earthquakes but federal intervention for climate change. Additionally, the study finds that ideology is associated with government level preferences for earthquakes and fire, but not climate change. These results suggest both hazard characteristics and variations in individual level public opinion matter for determining appropriate levels of governance. Policymakers should more carefully consider if their level of government is appropriate for addressing certain hazards risks and how constituent characteristics will shape the popularity of these decisions.

Flooding remains the most dangerous natural hazard and that which poses the greatest threat to property and the safety of human communities. The resilience of the built and natural environments is strongly influenced by the development and growth management guidance provided by a community’s ‘network of plans’, which often includes land use, hazard mitigation, and transportation plans, among others. The ways these multiple and independent plans interact can significantly impact community vulnerability. This study investigates the influence of a series of factors on community plan integration for resilience at the district scale in six United States (U.S.) coastal cities using hierarchical linear modeling. We create and compare a set of hierarchical linear models to answer the following research questions: (1) What is the level of plan integration? (2) What factors influence plan integration for resilience in six U.S. coastal cities? Findings show that communities with a larger “temporary” population and a lower socioeconomic status are less likely to incorporate hazard mitigation in local plans, controlling for community planning capacity. The relationship between the planning capacity and plan integration performance is stronger in some cities (e.g. those with planning mandates, much hazard experience, or low physical vulnerability) than in others (e.g. those without planning mandates, little hazard experience, or high physical vulnerability). Investigating the host of factors influencing plan integration may suggest a way forward, both for research and for communities struggling with issues of ‘siloing’, conflicting plan guidance, and disaster preparedness.

During the COVID-19 pandemic, children may spend more of their time at home, with very limited social interactions with their peers and no access to regular social events and gathering. It is particularly challenging for families with young children from birth to age five (i.e., infancy to pre-kindergarten age). Unlike children of elementary school age and beyond, young children usually do not receive structured remote instruction from their teachers due to their limited capacity for remote learning. A related challenge is that parents of young children may not receive similar guidance or support from childcare services as do parents of school-age children, as early childhood teachers usually do not extend classroom learning to home during the outbreak (i.e., remotely assigning homework or learning projects). By using an online-questionnaire method to collect two data points from parents on self-reported quality of parent-child relationships, this study aims to understand the COVID-19 impact on parents’ parenting practices, children’s behaviors, and parents' emotional well being. Because the COVID-19 pandemic may have different levels of impact on the parent-child relationship, based on family socioeconomic status (SES) and ethnic background, this study also investigates how family background contributes to the impact of the COVID-19 on families.

Identifying critical components for an Infrastructure System (IS), such as a water distribution system, paves the way to make more cost-efficient decisions to enhance IS resilience. However, the computational burden of model-based or simulation-based system performance evaluation, especially for large size networks, limited most existing studies to single critical component identification. In this study, a novel methodological framework to identify the Critical Combination of Components for Infrastructure System (CCCIS) is proposed to identify multiple components whose simultaneous failure leads to significant network performance degradation. This problem is formulated as a learning problem and leverages network embedding techniques. Based on domain knowledge, three categories of features are selected: component design features, network topology features, and demand loading features. They are defined and extracted from the network samples, normal network, or residual networks after a disruption event. Observing the similarities between infrastructure networks in CCCIS identification and documents in Natural Language Processing, we adapt the Bag-of-Words model used widely in NLP to generate network-level features based on link-level features to transfer high dimensional link-level features to low dimensional network-level features. This guarantees the computational efficiency and the generalizability of the proposed method to ISs with large size. Based on the generated network-level features, three learning techniques (regression, classification, ranking) are used to identify the CCCIS. Experiments for transportation roadway networks demonstrated that the proposed framework provided not only efficient and accurate performance evaluation, but also is generalizable for networks with various sizes, topologies, and demand patterns.

The behavior of people in the first stage of an emergency evacuation can dramatically impact the time required to reach a safe place. Scholars have widely investigated this behavior with a macroscopic approach fitting random distributions to represent the pre-evacuation time. However, microscopic investigations on how building occupants respond to social and environmental factors are still rare in the literature. This study aims to leverage machine learning to investigate factors affecting evacuation decision-making during the pre-evacuation stage of disasters using a case study of fire evacuation. Interpretable machine learning techniques were applied to capture nonlinear relationships between the input variables and the outcome and to reveal the interactions among the inputs. The behavior of 569 building occupants split between five unannounced evacuation drills in a cinema theater were investigated. Using a well-established machine-learning algorithm — random forest — people's emergency behavior pre-evacuation was modeled and predicted. The results indicate that both social and environmental factors affect the probability of responding to evacuation orders. Several independent variables, such as the time elapsed after the alarm has started and the decision-makers’ group size, present strong nonlinear relationships with the probability of switching to the response stage. Findings show interactions between the row number where the decision-maker sits and the number of responding occupants visible to her. An interesting finding is that a decision-maker is more sensitive to the proportion of responding occupants than the number of them.