Thomas J. Cova - US grants

The objective of this project is to design a geographic database of the Southern California wildfire evacuations of 2003. The primary research question is, "Where and when did wildfire evacuations occur and what were their general characteristics?" Example characteristics of interest include the time the event (wildfire) was detected, the timing, medium, and content of the evacuation order, the area that was evacuated and the decision-making process that led to the given protective action. The project will be divided into two phases: 1) information gathering, and 2) geographic database design. The first phase will focus on collecting information from in the form of articles, photos, video, and eye-witness accounts. Other secondary data to be collected include air photos, satellite imagery, land ownership, and roads. Structured interviews of emergency managers (informants) will be conducted to gather more detailed information on each of the evacuations and the decision-making process that led to a particular protective action (e.g. mandatory versus voluntary evacuation). The focus of the second phase will be assembling the information into a comprehensive geographic database.

This project will: 1) increase our understanding of the complex spatial dynamics of wildfire evacuations and associated decision making for a major event, 2) provide a sampling frame for future behavioral studies on warning and evacuation, and 3) lead to better evacuation modeling and prediction for other communities at risk to wildfire.

While the infinite complexity and dynamics of geographic worlds have long been recognized in the geographic information science (GIScience) literature, current geographic information systems (GIS) technology has not yet incorporated data models, query functions, or analytic tools that can adequately handle geographic complexity and dynamics. This research project aims to integrate these features into GIS data models, query, and analysis. Such integration will lay a foundation for the next generations of GIS technology to further empower GIS support for scientific understanding and discovery of geographic worlds. To achieve this integration, the investigators will examine geographic complexity and dynamics. The basic premise is that geographic conceptualizations need to go beyond separate field- and object-based views of geographic worlds. The investigators will focus on geographic complexity that arises from the interwoven properties of fields and objects embedded in phenomena and relationships at different spatial and temporal scales. They will consider geographic dynamics that reflect on propagation and evolution in space and time as analyzed by Lagrangian (focusing on the stationary action of flows) or Hamiltonian (focusing on the motion of a particle of mass) dynamics. Field- or object-based conceptualization alone cannot capture complexity and dynamics critical to an accurate representation of geography. The investigators will incorporate two additional views of geographic worlds: fields of objects (o-fields) and objects of fields (f-objects) to incorporate geographic complexity and dynamics. They therefore expect to extend the dual geographic conceptualization to a spectrum of objects, f-objects, o-fields, and fields, with scale and resolution are as functions that allow a shift in perspective along the spectrum. With the spectrum of geographic conceptualizations, they will develop a data model that incorporates geographic complexity and dynamics with uncertainty, formulates queries and analytical functions, and builds a prototype system for proof of concepts.

The collaborative project brings together researchers from the University of Oklahoma, University of California-Santa Barbara, and University of Utah to expand on their work on geospatial data modeling. In separate ventures, they have examined the use of combined fields and objects on geographic representation and demonstrated that such combination effectively extends geographic representation to incorporate much richer, more complex geographic semantics. Central to the research project is the idea of modeling GIS data based on geographic complexity and dynamics, as an alternative to the conventional data models that are built upon how data are captured. The research project promises a broader, more comprehensive inspection of issues related to the integration of fields and objects and the development of a holistic theory of the representation of geographic complexity and dynamics. This new approach to GIS data modeling extends static representation to a complex and dynamic view of the world and thus enhances GIS technology to be better suited for scientific research.

Sin Nombre hantavirus (SN) is a recently discovered virus carried by deermice that causes disease with high mortality in humans. Several recent studies have proposed that human disturbance of habitat significantly affects the number of deermice infected with SN. Given unprecedented rates of disturbance and limited understanding of the mechanisms governing variation in SN infection, a thorough study of how disturbance affects SN dynamics is warranted. The central focus of the proposed study is to determine how human disturbance affects SN prevalence in deermice and other reservoirs. To address this issue, a multifaceted research program is proposed that includes empirical and theoretical work. The field data will be used to determine the underlying mechanisms responsible for differences in prevalence. These ground-based data will be used to generate predictive mathematical models of prevalence using aerial and satellite images.

The broader impacts of this study include education, interdisciplinary research and national security. Several undergraduate and graduate students will be trained as part of this research. The project unites scientists from diverse fields (geography, mathematics, ecology, virology). Lastly, the research will yield critical information on the biology of Sin Nombre virus, which is listed as a biological agent of concern for national security.

Emergency managers recommend protective actions in the face of many threats to minimize loss of life and property and to maximize use of limited resources. In the context of wildfire, two common recommendations are to evacuate or shelter those at risk. Given these two options, questions arise as to which protective-action is best in a given scenario and when it should be issued. This project will examine: 1) the factors that are important in determining which protective action is best in a given wildfire, 2) the strategies that decision makers use to combine the factors, and 3) the effect of uncertainty on the decision making process. The research is based on a three-step experimental approach that relies on interviews, static information boards, and an interactive wildfire simulator to elicit knowledge from both expert and novice decision makers in wildfire management. Causal models of the decision making process will be developed and tested that include the relevant factors and their importance, the method by which they are combined, and the effect of uncertainty.

The results of this research will advance protective-action decision theory and provide a basis for improving the quality of decision-making in emergencies.

The goal of this research is to improve our understanding of the factors and thresholds that lead emergency managers to recommend protective actions to the public in the face of an environmental threat. Three key questions need to be addressed in this context: 1) who should take action, 2) what is the best action, and 3) when should this action occur? As straightforward as these questions may seem, the stakes can be extremely high and they are frequently addressed under time pressure and uncertainty. The focus of this project is "trigger points" (or triggers), a novel decision aid used by emergency managers to combine an event (e.g. time, place, condition) with a recommended protective action (e.g. evacuate, shelter-in-place, refuge shelter) for a threatened sub-population, such that the action is recommended if the event occurs. Triggers therefore provide a valuable framework for addressing the three questions noted above, and their analysis represents a potential leap in improving our understanding of emergency warnings. The objectives of this research are to: 1) extend current theory on protective actions to include triggers; 2) identify the factors that determine how a trigger is set in space and time, the relative importance of the factors, and the decision rule(s) used to combine the factors; and 3) develop and test cognitive and physical models of how triggers are set and detected. The research is based on a three-step experimental design comprised of: 1) interviews and observation to elicit knowledge about the types, nature, and efficacy of triggers, 2) experiments with a web-based wildfire scenario simulator to ascertain relevant factors in setting triggers and their respective importance to decision makers, and 3) a comparative study between triggers set by experts and ones derived through physical modeling.

The results of this project will advance our knowledge regarding a critical yet under-researched decision aid in protective-action decision making. The project will strengthen ties between the disaster research and management communities, and the results will be disseminated to practitioners through presentations, workshops and trade journals with the goal of improving protective-action training and ultimately public safety. This project will also enhance graduate and undergraduate educational infrastructure at the University of Utah while strengthening interdisciplinary collaboration at the Center for Natural and Technological Hazards. In addition, undergraduate educational modules will be developed using the wildfire simulator for classes in Emergency Management, Naturalistic Decision Making, and Fire Human-Environment Interactions. Finally, all software and data from this project will be made available to the disaster research community via a project website.

Hospitals are a critical infrastructure during and after disasters to care for vulnerable populations. The problem is that hospitals are also at risk to the same disasters that the communities they are part of are. Just as hurricanes prompt the evacuation of populations, hospitals also find themselves needing to make decisions about whether to evacuate their patient population prior to landfall or shelter-in-place during the storm. There is little evidence about how hospital administrators make this crucial decision in protecting the health and lives of their patients. This project will interview administrators and emergency managers from a sample of hospitals affected by Hurricane Harvey in Southeast Texas. The research will explore what information was considered in the decision, how was the information obtained, and how are the dangers associated with moving patients balanced against the dangers associated with staying in a hospital that may be flooded or sustain other infrastructure damage. The information gathered in this project will help emergency managers better understand the role of hospitals as a critical infrastructure component during a disaster.

While there is a significant literature on hospital evacuation, the majority of articles reflect case studies of the evacuation process and do not discuss the decision-making process of whether to evacuate or to shelter-in-place. The research objective of this project is to collect the data needed to understand hospital administrators' decision making processes when deciding to evacuate the hospital or shelter-in-place in advance of or during a hurricane event. Understanding this process is important for planning how communities will respond to disasters and their reliance on hospitals as a critical infrastructure component. The research is guided by the Protective Action Decision Model. The research will explore how the perceptions of the severity of the hazard, the level of preparedness to evacuate or shelter in place, characteristics of the patient load, the perceived net benefit of evacuating over sheltering in place, the presence/absence of evacuation orders, and environmental cues from the storm. The research team will conduct key informant interviews with hospital administrators from a purposive sample of hospitals in Harvey-effected counties. Online news articles suggest that 25 hospitals in the region evacuated at some point of the storm. The sampling methodology will be stratified by hospitals in the coastal region that received hurricane warnings based on predicted storm surge and those inland that were primarily at risk from flooding, and also based on evacuation status (evacuated pre-impact; evacuated post-impact; and did not evacuate). The research team will select 25% of hospitals (42) ensuring that there are hospitals from each of the 6 defined strata. Questions will examine which individuals were included in the decision process; what information was sought; where was the information obtained from; and how did decision makers weigh the risk of evacuation vs. shelter-in-place? Understanding the factors contributing to a hospitals ability to stay open during a disaster is a critical component for the resilience of the community.