Dynamic Decision Making in Complex Environments: Before, During, and After Disaster
Louise K. Comfort, Director
Center for Disaster Management
University of Pittsburgh, Pittsburgh, PA 15260
Policy Problem: Interdependent Decision Processes in Disaster Environments
Decision processes before, during, and after disaster shape the consequences and costs of extreme events on communities exposed to risk. Decisions taken before a hazardous event occurs create the basis for actions taken in response to an actual event which, in turn, enable or constrain actions taken by the community to recover from that event. Understanding the interdependencies among these sets of decisions is central to the development of community resilience in coping with disaster. Documenting and measuring these interdependencies systematically is not trivial. This study proposes to document the mitigation and preparedness decisions taken in multiple states, counties, and cities before Hurricane Sandy made landfall in the early hours of October 30, 2012, assess the extent to which those activities informed response actions in the damaged communities, and examine the degree to which decisions taken in response, in turn, shaped the priorities, constraints, and costs of recovery strategies. Hurricane Sandy provides a major test of changes in policy and investments in resources and time made by the Federal Emergency Management Agency (FEMA) to improve disaster resilience since its much criticized performance following Hurricane Katrina in 2005. This study addresses the research question: what factors, if any, increased or decreased the resilience of states and communities damaged by Hurricane Sandy to cope with this extreme event?
A key approach to measuring resilience in communities exposed to recurring disaster risk is to integrate knowledge of the spatial characteristics of risk, vulnerability, cost, and ability to pay in the design of policies and practice to reduce disaster risk. This approach builds on the concept of “hazard of place” (Cutter, 2006), but extends it to include not only assessment of the geolocation of hazards, probability of occurrence, and vulnerability of the population at risk, but also the cost of mitigation and preparedness measures, estimated reduction in losses achieved from measures taken, and timeliness and cost of recovery in communities afflicted by disaster events. This analysis creates a baseline of performance before disaster against which decisions made during response operations can measure adaptations to the impact of the disaster on the community. Decisions made in response operations then affect the requirements of recovery and likely the order and timing of recovery strategies.
While the merit of mitigation and preparedness in facilitating response actions and speeding recovery after an extreme event has long been accepted as a primary requirement of building disaster resilience (McLoughlin, 1985; Witt, 1995; Kaufman, 2012), there has been little systematic effort to measure the degree of interdependence among decisions made in these three phases of disaster that have different degrees of urgency, constraints on action, and consequences for the population. Key decisions in each of these phases of disaster management are often made by different policy makers operating at different levels of authority and responsibility, but the consequences of each set of decisions affect the possible strategies available for the next set of decisions. Capturing this decision process and tracking the dependencies from preparedness to response to recovery reveals the degree of resilience that the community is able to achieve in an actual event. This complex set of interacting tasks can most effectively be assessed in the immediate aftermath of a disaster.
This approach is based on estimates of the value of the built infrastructure in the region at risk before an extreme event occurs, as well as the capacity of the population, organizations, and jurisdictions in the region to pay for services to reduce that risk. A long-standing policy issue in hazard reduction has focused on the disconnect between exposure to hazards and the cost of response and recovery measures when known hazards strike regions where populations and critical infrastructure are located in vulnerable zones, but fail to take appropriate mitigation measures to reduce risk (Platt, 1999; Mileti, 1999). In this research, we compare the decisions taken by local, state, and federal agencies to increase mitigation and preparedness for a region at risk, the estimated losses from Hurricane Sandy incurred by communities, the capacity to pay for risk reduction measures on a regional basis, and the estimated costs in time and resources for recovery in the damaged communities. The chain of decisions taken measured against documented consequences and costs allows a beginning estimate the value of resilience. The still mounting estimates of costs from the consequences of Hurricane Sandy to coastal and inland communities in eight states vividly illustrate this policy problem of measuring the value of resilience on local, state, and national scales.
Since the damaging consequences from Hurricane Katrina in 2005, advances in satellite monitoring of meteorological conditions, visual representation of changing states of risk, technical increases in capacity to exchange, store, update, and communicate information regarding risk on a regional basis have created the capacity for organizations and communities to learn and understand collectively the risks to which they are exposed. These sources of data provide an overview of the state of the region before, during, and immediately after the storm that can be verified with data from operational reports, interviews with decision makers, and content analysis of local newspapers.
Research Design: Integration of Sociotechnical Measures with Representation of Risk
This study focuses on the coastal communities of New York and New Jersey. The goal of the research is to explore how integration of different measures of severity of exposure, probability of occurrence, number and types of population at risk, number and types of critical infrastructure at risk, estimated cost of mitigation and preparedness measures for the region, and potential sources of financial support informed collective decision making processes for the region.
To characterize the advancing risk to the coastal communities, we use data from publicly available sources such as the US Army Corps of Engineers satellite imagery of the region and the US National Weather Service records of the advancing storm. At the state level, we seek situation reports from the New York and the New Jersey Emergency Management Agencies for one week before the storm, one week during and immediately after disaster operations, and two weeks in the early period of recovery. At the local level, we are conducting a content analysis of local newspapers for one week before the storm and three weeks after landfall, as well as situation reports from selected county operations agencies, and professional reports. These sets of data will allow us to identify networks of organizations participating in three periods of decision making: 1) one week before the storm; 2) one week of operations during and immediately after landfall; and 3) two weeks during the early recovery period. We will validate the decision process by reviewing professional reports and conducting a series of semi-structured interviews with key decision makers at local (municipal and county), state, and federal levels of jurisdiction.
This field study will collect data regarding three primary points of decision in assessing resilience: 1) the extent to which the provision of timely, valid information regarding risk presented before the approaching storm informed collective decision making at the community, state, regional, and national levels regarding preparedness; 2) the extent to which key decisions taken to prepare for the storm informed or constrained options in response operations as Hurricane Sandy struck the coastal communities; and 3) the extent to which decisions made in response operations facilitated or constrained options for rapid recovery. Characterizing these three decision points in the events surrounding Hurricane Sandy will provide a measure of resilience achieved by this coastal region.
In this study, we use the geospatial analysis of the region at risk to identify and estimate the economic value of existing infrastructure in the region, and network analysis to identify the networks of principal actors participating in decision processes in each of the three periods of emergency operations. We use the content analysis of news reports to identify and document strategies for communication and feedback that generate interactive learning processes among key actors engaged in the disaster management process. We use semi-structured interviews and professional reports to validate the organizational networks and the feedback processes within and among them. From this analysis, we will develop Bayesian network models to assess the interdependencies within each of the major networks engaged in disaster operations and among the set of interacting systems for the region. We plan to use computational modeling to construct a system dynamics model for the meta-system of disaster operations to evaluate potential patterns of changing resilience for the region, based on different parameters for interaction among the sub-systems and conditions of operation within the region.
The initial data collection for this project is funded by the National Association of Workforce Boards. Supplemental support is provided by the Graduate School of Public and International Affairs and the Center for Disaster Management.
Graduate Student Researchers:
Jee Eun Song