AN INTERACTIVE, INTELLIGENT, SPATIAL INFORMATION SYSTEM
(IISIS) FOR DISASTER MANAGEMENT: A COMMUNITY MODEL
(An earlier version of this paper was accepted by the Environmental System Research Institute, ESRI, for presentation to the 18th Annual ESRI International User Conference, July 27-31, 1998 San Diego, California)

L. Comfort, D. Metzler, Y. Sungu, M. Dunn, L. Selavo, K. Brown, J. Myung
University of Pittsburgh University of Pittsburgh

Keywords: Sociotechnical system, disaster management, real time information, intelligent reasoning

Problem Statement

The startling increase in costs of disaster in the U.S. over the last decade has compelled the emergency management community to shift its focus from response and recovery to mitigation of damaging events. These costs have been estimated at $50 billion per year¹, creating a cumulative impact on the nation's capacity to manage continuing risks to communities exposed to natural and technical hazards. Mitigation practices that have relied either primarily on technical solutions, or organizational training and response, have historically failed.² In most instances, these failures resulted not from lack of effort, money, or awareness of risk. They resulted from lack of adequate knowledge about the likely consequences of a disaster in an interdependent, urban environment, and lack of timely, accurate information available to responsible organizations to enable them to take effective action under the stress of an actual event.³ As disaster losses escalate in interdependent environments, the search for improved methods of risk reduction and response intensifies.

This problem involves three broad sets of issues. The first set is practical: how can a community increase its capacity to mitigate and respond to disaster in its environment? The second set is social, cognitive, and organizational: what are the cognitive processes that occur among experienced personnel engaged in interactive disaster response operations; how do these processes change to meet the dynamic demands of an evolving situation; and what kinds of information are needed to support decision making within and among organizations in reference to that event? The third set is technical: how can an information system support the dynamically changing roles of individuals and organizations as they adapt to shifting needs, without overwhelming the personnel in complexity; and how can the system provide distributed intelligent support in response to changing local conditions to assure that actions taken conform to the established guidelines of emergency management?

Existing software programs designed for emergency management have not successfully met the needs of practicing managers for four main reasons. First, existing programs are characterized by a lack of flexibility in representing the diverse perspectives that emergency managers require in a dynamic situation. Second, most programs are designed only for use in emergencies and are hampered in practice by the staff's lack of familiarity with the program in an actual event, and equally, by a lack of currency in the knowledge base of stored information about the community. Third, existing programs are largely designed for use by single organizations and lack the capacity for interorganizational decision support so crucial in a community-wide event. Finally, no existing system includes an intelligent reasoning component that can assist managers in addressing the rapid increase in volume and complexity of information in disaster events.

IISIS Approach

The prototype interactive, intelligent, spatial information system (IISIS) offers an approach to mitigation and response that links information technology with organizational design and learning to create an information system that provides decision support to practicing agencies. It utilizes aspects of advanced computational techniques including intelligent reasoning, graphical information system techniques, and flexible computer human interface design to support communication, coordination, feedback and learning processes identified by direct observation, a field operations and analysis of information requirements for practicing managers in disaster environments. The prototype IISIS facilitates coordination of the multiple aspects of disaster response and provides real-time information as the disaster evolves. Current development will provide intelligent guidance to decision makers confronting complex situations requiring urgent action.

The IISIS prototype is designed to provide interorganizational decision support to multiple organizations seeking to achieve the common goal of reducing risk, and to support appropriate action at multiple levels -- households, private businesses, public agencies, and nonprofit organizations -- when threats occur. ⁴ This approach produces a 'sociotechnical' system of interacting computers, organizations, and individuals working toward a common goal. (Figure 1) The term, sociotechnical, indicates the interdependent relations among the built environment, machines and people who use them in the dynamic environment of a modern urban community. Computers become active components of the response system, enabling a diverse set of human decision-makers to process complex information simultaneously with sufficient accuracy and timeliness to support coordinated action.

The IISIS design supports a dual use system for improved management of disaster at both organizational and community levels. That is, managers may use the system in their respective public, private, and nonprofit organizations for daily, routine operations in monitoring and mitigating seismic risk. When disaster occurs, the IISIS activates a community-wide response network that creates an event-specific knowledge base by drawing relevant information from multiple data bases and many organizations to support interorganizational and interjurisdictional decision making in reference to that event. Dual use ensures that users are familiar with the system and the information is current when a disaster occurs.

The long-term goal of this project is to increase the capacity of communities to mitigate and respond effectively to disasters. This need is especially critical at the local level, where the initial actions taken can affect the subsequent course of decisions and actions dramatically -- positively or negatively -- over the course of disaster operations. The IISIS design specifically incorporates information needs acknowledged by practicing emergency managers and recognizes the importance of organizational design, training, development and learning as central to the effective utilization of information technology in hazard reduction. This project would mark a vital step toward the realization of a global information infrastructure to reduce natural and technological hazards.

The transition from routine to disaster operations presents a stubborn problem in disaster response. The inability of personnel to grasp the magnitude of changes occurring in disaster environments may be due to missing data, but it may also be due to a cognitive lag behind the pace of rapid change. An information system that simply reported all of the events occurring in a disaster may not provide management personnel with a coherent, comprehensible interpretation of the whole situation. Assessing the rates of change in the frequency of demands for assistance, and the degree of complexity in interactions among the components of the evolving response system may be summarized, analyzed, and presented by the system to the set of professional disaster managers simultaneously. Such measures would provide crucial assessments of the `state of the community' that would enable professional managers to adjust their actions in the crisis effectively and efficiently.

Figure 1: Sociotechnical System, IISIS Prototype

Design Issues

IISIS poses four fundamental system design issues that set this project apart from current practice in building information systems for organizational use. First, this research addresses information needs for decision-makers operating in the dynamic, urgent environment of disaster operations following a major disaster. Most information systems are designed for well-understood tasks and relatively static situations, demonstrated in practice in fields as diverse as software engineering and expert system development. The full consequences of a major disaster in a complex urban environment, however, cannot be predicted reliably in advance. A multi-agent decision support system for such an environment must necessarily be dynamic, incorporating a continuous stream of new information into the system and adapting performance of different subsystems accordingly (O'Hare and Jennings, 1996:67-75).

Second, in this dynamic operating environment, participating personnel often take on new roles, or create novel relationships which did not exist prior to the disaster. An adaptive information system needs to be more flexible than a typical information system in order to offer personnel a range of views regarding the evolving situation as needs and available resources change (Hoschka, 1996).

Third, disaster environments place a large cognitive and emotional burden upon participating personnel. The proposed prototype will provide not only a knowledge base of relevant information and a dynamic description of unfolding events, but also an intelligent guidance system to support individual reasoning and cooperative decision making processes in that event. Rather than the narrowly focused architecture of traditional expert systems and decision support systems, this system will provide a robust and broad base of support for human reasoning. It will offer: 1) local advice for strategic situations; 2) a general implementation of the standard operating procedures for disaster response (based on the qualitative inputs provided by human users); 3) a broad based assessment of the overall situation regarding aspects such as rates of change of activity and levels of risk, and 4) a capability to query the system regarding the reasons for which conditions do not hold, or events did not take place, and the consequences of hypothetical actions (Metzler and Martincic, 1998).

Fourth, given the dynamic, urgent environment of disaster operations, the information system needs to be both robust and sustainable. The dual-use design of this system will serve to increase both qualities, as managers would use it in their daily operations for single-organizational tasks and update the knowledge base on a continuing basis. Dual use ensures that users are familiar with the system and the information is current when a disaster occurs. This design overcomes the twin weaknesses of single-use emergency information systems: lack of familiarity with the system among personnel when the disaster actually occurs, and obsolete data in the knowledge base. If organizations use the system to support their daily activities, their personnel will not only maintain the currency of the knowledge base, but make the transition to community-wide use more easily.

System Design

The current IISIS prototype consists of three basic components: 1) interactive communication among distributed knowledge bases; 2) dynamic representation of graphical information as the event unfolds, using GIS; and 3) intelligent reasoning by the computer to integrate incoming information from the environment with stored information about the community, its infrastructure and population to assess the likely consequences of the event for the community. Figure 2 illustrates how observations of a threatening event are transmitted via a Web page interface to an information system that provides a continuous assessment of the 'state of the community'. The intelligent reasoning component will calculate the probabilities of risk from specific events for designated community systems, and, when completed, will calculate the cumulative probability of risk for the whole community. As information from the environment enters the information system, it is passed to the intelligent system, which first compares the change in new information about the community with stored information from the knowledge base to assess the degree of discrepancy. The intelligent system draws upon multiple sources of knowledge concerning stored data about the geographic and physical characteristics of the environment, social and economic characteristics of the population, legal and organizational requirements for action. Based upon discrepancies between incoming and stored information, the intelligent system will calculate probabilities of risk, and return this information to the user, via the Web interface. As events change, new information triggers a continuing reassessment of the state of the community, providing practicing managers with current assessments of the probability of risk at specific locations, as well as an overall assessment of risk for the whole community.

The IISIS prototype supports decision-making at multiple levels of jurisdiction and action. The human-computer interface is a set of Web pages that receive input from the user, transmit it to the information system, and return new information, processed through the system, to the user. The screens are designed to keep the manager's focus for his/her particular level of operations, but to allow movement to other jurisdictional levels of operation and other phases of disaster management as the event evolves. For example, if the user designates response as the phase of disaster operations, all subsequent screens will capture information regarding response operations. The designation of jurisdictional level has the same function. The user may, however, shift the stage or level of operations by changing the selection to a different phase or jurisdiction. The system is designed to be flexible and allows adaptation to meet specific needs.

The set of functions for the IISIS prototype is organized in three basic subsets of functionality. The first identifies the initial conditions of the community as the basic knowledge base for coordinated action. The second subset reports the event and assesses its impact upon the community dynamically. The third subset provides a continuous record of disaster operations for the specific disaster event.

Figure 2: System Design, IISIS Prototype