Modeling protective action decision-making in earthquakes by using explainable machine learning and video data.

Earthquakes pose substantial threats to communities worldwide. Understanding how people respond to the fast-changing environment during earthquakes is crucial for reducing risks and saving lives. This study aims to study people's protective action decision-making in earthquakes by leveraging explainable machine learning and video data. Specifically, this study first collected real-world CCTV footage and video postings from social media platforms, and then identified and annotated changes in the environment and people's behavioral responses during the M7.1 2018 Anchorage earthquake. By using the fully annotated video data, we applied XGBoost, a widely-used machine learning method, to model and forecast people's protective actions (e.g., drop and cover, hold on, and evacuate) during the earthquake. Then, explainable machine learning techniques were used to reveal the complex, nonlinear relationships between different factors and people's choices of protective actions. Modeling results confirm that social and environmental cues played critical roles in affecting the probability of different protective actions. Certain factors, such as the earthquake shaking intensity and number of people shown in the environment, displayed evident nonlinear relationships with the probability of choosing to evacuate. These findings can help emergency managers and policymakers design more effective protective action recommendations during earthquakes.

A highway vehicle routing dataset during the 2019 Kincade Fire evacuation.

As the threat of wildfire increases, it is imperative to enhance the understanding of household evacuation behavior and movements. Mobile GPS data provide a unique opportunity for studying evacuation routing behavior with high ecological validity, but there are little publicly available data. We generated a highway vehicle routing dataset derived from GPS trajectories generated by mobile devices (e.g., smartphones) in Sonoma County, California during the 2019 Kincade Fire that started on October 23, 2019. This dataset contains 21,160 highway vehicle routing records within Sonoma County from October 16, 2019 to November 13, 2019. The quality of the dataset is validated by checking trajectories and average travel speeds. The potential use of this dataset lies in analyzing and modeling evacuee route choice behavior, estimating traffic conditions during the evacuation, and validating wildfire evacuation simulation models.

Using GIS to Understand the Influence of Hurricane Harvey on Spatial Access to Primary Care.

Hurricanes can have a significant impact on the functioning and capacity of healthcare systems. However, little work has been done to understand the extent to which hurricanes influence local residents' spatial access to healthcare. Our study evaluates the change in spatial access to primary care physicians (PCPs) between 2016 and 2018 (i.e., before and after Hurricane Harvey) in Harris County, Texas. We used an enhanced 2-step floating catchment area (E2SFCA) method to measure spatial access to PCPs at the census tract level. The results show that, despite an increased supply of PCPs across the county, most census tracts, especially those in the northern and eastern fringe areas, experienced decreased access during this period as measured by the spatial access ratio (SPAR). We explain this decline in SPAR by the shift in the spatial distribution of PCPs to the central areas of Harris County from the fringe areas after Harvey. We also examined the socio-demographic impact in the SPAR change and found little variation in change among different socio-demographic groups. Therefore, public health professionals and disaster managers may use our spatial access measure to highlight the geographic disparities in healthcare systems. In addition, we recommend considering other social and institutional dimensions of access, such as users' needs, preferences, resource capacity, mobility options, and quality of healthcare services, in building a resilient and inclusive post-hurricane healthcare system.

Warning Triggers in Environmental Hazards: Who Should Be Warned to Do What and When?

Determining the most effective public warnings to issue during a hazardous environmental event is a complex problem. Three primary questions need to be answered: Who should take protective action? What is the best action? and When should this action be initiated? Warning triggers provide a proactive means for emergency managers to simultaneously answer these questions by recommending that a target group take a specified protective action if a preset environmental trigger condition occurs (e.g., warn a community to evacuate if a wildfire crosses a proximal ridgeline). Triggers are used to warn the public across a wide variety of environmental hazards, and an improved understanding of their nature and role promises to: (1) advance protective action theory by unifying the natural, built, and social themes in hazards research into one framework, (2) reveal important information about emergency managers' risk perception, situational awareness, and threat assessment regarding threat behavior and public response, and (3) advance spatiotemporal models for representing the geography and timing of disaster warning and response (i.e., a coupled natural-built-social system). We provide an overview and research agenda designed to advance our understanding and modeling of warning triggers.

Spatial and temporal variation in evacuee risk perception throughout the evacuation and return-entry process.

Developing effective evacuation and return-entry plans requires understanding the spatial and temporal dimensions of risk perception experienced by evacuees throughout a disaster event. Using data gathered from the 2008 Cedar Rapids, Iowa Flood, this article explores how risk perception and location influence evacuee behavior during the evacuation and return-entry process. Three themes are discussed: (1) the spatial and temporal characteristics of risk perception throughout the evacuation and return-entry process, (2) the relationship between risk perception and household compliance with return-entry orders, and (3) the role social influences have on the timing of the return by households. The results indicate that geographic location and spatial variation of risk influenced household risk perception and compliance with return-entry plans. In addition, sociodemographic characteristics influenced the timing and characteristics of the return groups. The findings of this study advance knowledge of evacuee behavior throughout a disaster and can inform strategies used by emergency managers throughout the evacuation and return-entry process.

Visualizing diurnal population change in urban areas for emergency management.

There is an increasing need for a quick, simple method to represent diurnal population change in metropolitan areas for effective emergency management and risk analysis. Many geographic studies rely on decennial U.S. Census data that assume that urban populations are static in space and time. This has obvious limitations in the context of dynamic geographic problems. The U.S. Department of Transportation publishes population data at the transportation analysis zone level in fifteen-minute increments. This level of spatial and temporal detail allows for improved dynamic population modeling. This article presents a methodology for visualizing and analyzing diurnal population change for metropolitan areas based on this readily available data. Areal interpolation within a geographic information system is used to create twenty-four (one per hour) population surfaces for the larger metropolitan area of Salt Lake County, Utah. The resulting surfaces represent diurnal population change for an average workday and are easily combined to produce an animation that illustrates population dynamics throughout the day. A case study of using the method to visualize population distributions in an emergency management context is provided using two scenarios: a chemical release and a dirty bomb in Salt Lake County. This methodology can be used to address a wide variety of problems in emergency management.