L. Comfort, Y. Sungu, M. Huber, J. Piatek, M. Dunn, and D. Johnson

IISIS Project, University Center for Social and Urban Research

University of Pittsburgh, Pittsburgh, PA 15260, USA

Coordination within and between organizations in disaster environments presents a continuing challenge in disaster management. Standard administrative practices have proven ineffective in achieving adequate coordination in the dynamic environment of disaster operations. We propose that disaster response systems evolve through the search, exchange and feedback of information through multi-way communication networks and graphic representation of information. Such networks integrate both technical and organizational components to function as a distinct system, focused on the goal of disaster reduction and response. We further propose that interorganizational coordination in disaster environments may be facilitated by the design and implementation of an interactive, intelligent, spatial information system (IISIS).

A prototype IISIS is currently under development at the University of Pittsburgh. The prototype IISIS integrates three types of information technology: 1) an interactive, distributed system that operates over intra- and inter-organizational networks; 2) geographic information systems that represent information graphically; and 3) intelligent reasoning that facilitates analysis of complex problems through computation. We present a report on the status of this research, and summarize the steps we have taken to build a working prototype IISIS at the the University of Pittsburgh. The University, a community of 32,000, offers a working model of an interdependent urban jurisdiction located in a larger metropolitan region. Field observations show that the process of building the prototype IISIS activates discovery, learning, and self organization both within and among the administrative units of the University. These factors contribute to coordination among the units to create a University response system. The model is applicable to other interdependent communities.

Keywords: coordination, self organization, distributed system, intelligent reasoning, GIS, disaster response

Disaster environments present an extraordinarily difficult context for inter-organizational and inter-jurisdictional coordination. When disaster threatens a community, it requires different responses at different locations from different organizations, which must set aside prior activities and focus attention, time and effort on a common goal. To achieve a coordinated response, these actions must be taken simultaneously. Coordinated response is particularly difficult to achieve with threats such as hazardous materials, in which the general vulnerability is known, but the specific time, location, and magnitude of the threatening event are uncertain.

Coordination under uncertain conditions requires an understanding of shared risk (Comfort, 1997; 1999). When risk is shared, actions taken by any one person may increase the risk, escalating the event into a wider disaster, or reduce the risk, bringing the threat under control and limiting the consequences. Informed action, guided by a shared goal of protection of life and property, becomes a primary strategy for disaster reduction and response. The capacity of a community to take informed action can be strengthened by appropriate uses of information technology (Comfort, 1993; Comfort et al., 1998; National Research Council, 1996). Designing technology to support coordinated action requires both technical and organizational planning. Such planning needs to create an awareness of risk in order to define effective action within and among organizations. It also needs to involve the first level of response, where initial steps taken to reduce risk for the community influence subsequent opportunities for action in a rapidly evolving disaster.

The organizational capacity of a community includes public, private, and nonprofit organizations as well as households. In order to increase this capacity to reduce and respond to disaster efficiently, the community needs to use a range of information technologies to create an information infrastructure that provides decision support to participating members. An effective decision support system would facilitate coordinated action first, within the community prior to a threatening event, and second, between the community and external organizations in an evolving emergency response system. Such a decision support system would assist the missions of local, state, and federal emergency management agencies by supporting the mitigation and management of risk.

The theoretical concept of self organization underlies the design of a decision support system to support coordinated action in community response to disaster. Self organization is the spontaneous reallocation of energy and action to achieve a collective goal in a changing environment (Kauffman, 1993; Comfort, 1994). This capacity to adapt to changes in the environment is observed in both social systems, when organizations adapt their performance to meet unexpected needs, and technical systems, where computers adjust the performance of systems operating in changing environments.

For example, immediately following the 1987 Whittier Narrows Earthquake in the San Gabriel Valley of East Los Angeles, the traffic lights went out. Traffic backed up on city streets, and slowed to a standstill. In frustration, one driver pushed his car to the side of the road and began directing traffic at the next intersection, enabling other drivers of stalled cars to begin moving again. At other intersections across the city, drivers who had observed the first action, began to do the same. Without conscious direction, traffic was moving through the streets, as `citizen traffic cops= spontaneously assumed responsibility for coordinating the flow of traffic, a collective goal (Comfort, 1999). In computer science, the concept of self organization implies a similar process of searching for the most efficient flow of electricity around an unexpected set of obstacles, resuming the performance of an interrupted process with the best available route (National Research Council, 1996). This capacity for flexible adaptation to changing conditions is critical to maintaining effective performance in disaster environments in both social and technical systems.

Achieving self organization in both organizational and technical systems represents a substantial set of organizational and technical challenges. Four issues are central to the successful development and implementation of such a plan. They are:

Disaster as a mechanism of change.

Perception of a policy problem directly influences the actions that are or are not taken in a social setting (Polsby, 1998). This general proposition is even more accentuated in the perception of risk or disaster (Douglas and Wildavsky, 1982). When disaster is perceived as the product of external forces, an Aact of God,@ no action is likely to be taken, despite known risks and reasoned possibilities for change. But when disaster is perceived as the product of interacting and cumulative decisions between groups of people and their environment, coordinated action is possible if one identifies the critical points at which a response system evolves and provides timely, accurate information to decision-makers at those points. The evolution of the response system depends upon the timely flow of accurate information, and the level of awareness of decision makers at all levels of the system regarding the consequences of their specific actions for the performance of the whole system.

In this context, disaster is perceived as a mechanism of change. Potential risk need not be feared. Rather, recurring disaster can be viewed as a test of the existing policies of a community: political, structural, social, economic. If basic community institutions and functions fail in a hazardous event, the ensuing disaster serves as evidence of the need for change. It creates the opportunity to redesign, revise or rebuild the human environment damaged by the event. Simply Arestoring the community to normalcy@ recreates the likelihood of recurrence of similar disaster at a later time (National Research Council, 1996). Assessing the likely points of failure or vulnerability before an event occurs, and taking appropriate action to reduce the probability of failure represent a more productive strategy. This approach is the key premise underlying the Federal Emergency Management Agency=s `National Mitigation Strategy= (Krimm, 1997).

If disaster is perceived as a mechanism of change, it influences the choices B social, political, scientific, and economic B that are made by multiple sets of policy makers interacting in diverse ways. Actions taken during a disaster become defining elements for the (temporary) resolution of that disaster, but also likely steps toward the creation of the next disaster. A major function of disaster management is the accurate assessment of not only the specific causes of a particular incident B e.g., an unanticipated release of hazardous materials B but also the interdependent functions and structures of the community that are affected by that release. This assessment provides a means of monitoring the impact of hazardous materials upon the community, but its validity depends upon the degree of access to, and quality of information about, the community available to policy makers before the release occurs. It depends as well on timeliness in communicating risk to community managers who have the legal responsibility to protect people and property. The actual choices they make either reduce or exacerbate the evolving crisis, sometimes in both directions simultaneously.

A decision support system that facilitates self organization recognizes the dynamic nature of disaster and focuses on decision points that enhance or reduce the likelihood of its occurrence. Mapping this continuously evolving pattern of interaction is an important, but difficult, function of disaster management. This interaction is interdisciplinary, inter-organizational, and inter-jurisdictional. Only recently has the scientific and technical capacity B including satellite communications, remote sensing, GIS analysis B existed to support this assessment on an ongoing basis. Scientific and technical capacity generate further possibilities for interaction among community participants that exceed the possibility for organizational control. The system either becomes self organizing and establishes a more appropriate order for the changed environment, or it disintegrates into chaos and disaster.

When the complexity of interacting scientific, social, political, and economic conditions exceeds the existing capacity for organizational control, decisions taken by local actors govern the direction of the evolving process (Prigogine & Stengers, 1984; S. Kauffman, 1993; Gell-Mann, 1994; Comfort, 1994; Comfort, 1997). Since disaster means loss of control, investment in improving capacity for organizational response at the local level is likely to generate the greatest benefit to the community. Failure to improve capacity in first response is likely to undercut any other measures undertaken for disaster reduction. An interactive, intelligent, spatial information system designed to support coordination among the participating actors and administrative organizations of the community would greatly reduce the risk of disaster and enhance its capacity for response to extreme events.

Coordination among actors B individuals, organizations, jurisdictions B has long been sought in research in organizational theory, without much success (Caiden and Wildavsky, 1974). The requirements for coordination -- monitoring performance of a complex system and sharing that information to support timely action by many participants seeking a common goal B have been difficult to achieve in dynamic environments through standard administrative practices.

Information technology now provides a means of decision support that enables coordination among multiple organizations and jurisdictions. While coordination of action among participants is frequently recommended as a solution to complex, interdependent problems such as shared risk, planners and policy makers have found it difficult to implement in practice. Administrative theorists have not been able to define coordination in ways that do not imply coercion (Caiden and Wildavsky 1974, 277-279), or to devise means of facilitating coordination without compromising the shared goal (Wilson 1989, 268-274). Theorists critical of coordination have viewed it primarily as a problem of control. They have defined it as a set of organizational procedures that would require multiple participants with different levels of training, understanding and responsibility to follow a common set of rules, often externally imposed, to achieve a complex goal.

Coordination in disaster operations is largely achieved by different means. While emergency response agencies often share similar training and operate within a similar framework for response, e.g. international fire training or the Incident Command System, coordination depends substantially upon processes of information search, exchange and feedback that produce both intra- and inter-organizational learning. The detail, accuracy, and timeliness of the information exchange among them shapes the process, as the organizations strive, individually and collectively, to achieve a common goal. Through learning processes, coordination becomes mutual adaptation, or self organization. The goal of a community emergency planning process is to enable the community to become self organizing in the reduction of risk and response to disaster.

The design of a self organizing system is based upon the concept of an N-K system developed by Stuart Kauffman (1993) to assess change in dynamic environments. In his assessment, Kauffman identifies the number of organizations (N) that are interacting to achieve a collective goal (P), and the number of interactions (K) among them. This design has been extended in studies of response systems following earthquake disasters to include the number and type of transactions performed by organizations directed toward this common goal (T), the duration of their involvement in the response system (D), and the source of funding (S) for their activities (Comfort, 1999). These measures serve as indicators of an evolving response system that changes over time. They are summarized as follows:

N = number of organizations participating in the system

K = number of interactions among organizations participating in the system

T = type of transactions performed by participating organizations to achieve system goal

These measures guide the detailed task of mapping the decision processes for an interorganizational response system in practice. The next section summarizes current research that is developing a prototype decision-support system to facilitate self organization in an actual community exposed to risk.

Interactive, Intelligent, Spatial Information System (IISIS)

A prototype interactive, intelligent, spatial information system (IISIS) is under development at the University of Pittsburgh to support inter-organizational coordination in hazardous materials management (Comfort et al., 1998). While the general design of IISIS supports a self organizing response system, the actual system needs to be fitted to the conditions, context, and actors for the specific community in which it is implemented. This requirement has both advantages and disadvantages. The advantage is that an IISIS creates a community-wide knowledge base that can be used to provide inter-organizational decision support for many types of policy problems. The disadvantage is that maintaining the knowledge base in order to provide current, valid information is a continuing task that requires commitment, cooperation and coordinated action among the members of the community who use it for decision support. The two conditions are reciprocal. A community-wide knowledge base, when it is current and valid, demonstrates its value in addressing complex policy problems and likely increases commitment by members of the community to maintain it. Should maintenance falter and the data become invalid, its value to decision processes decreases, and consequently, its ability to support coordinated action also decreases. The system requires continual monitoring, learning and adaptation for organizational and technical systems.

An IISIS links two types of information processing systems: 1) the technical system of computers; and 2) the human cognitive system of decision makers and their respective organizations (Weick, 1990). Both systems are amplified by networks of communication among multiple computers for the technical system and many decision makers for the organizational system. The load, rate, and complexity of information that is transmitted within and among organizations in dynamic disaster operations is massive. Without technical support, the information processing demands overwhelm the cognitive capacity of individual managers and organizations to absorb, process, and use the flood of incoming information as a basis for timely, informed action. Using distributed knowledge bases and a network of computers operating in parallel, the technical computing system is designed to support the human organizational system in its conduct of distributed, parallel operations in response to a given hazard. A distributed system requires an effective means of electronic transmission that is reliable and readily available. The Internet performs this function, but given its increasingly heavy load, it becomes less reliable for emergency communication. A major question to be investigated in any community development process is to determine the most reliable and robust mechanism for the electronic transmission of data under emergency conditions.

IISIS is designed to operate as a distributed system on at least two levels: professional and public. In its professional form, the prototype IISIS will use a custom-designed browser that is accessible only to authorized professional managers for use in community-related decision processes. This professional IISIS allows managers to order, store, recall, and exchange information relevant to hazardous materials or other risks in three different ways: 1) within organizations; 2) across organizations within jurisdictions; and 3) within a network of organizations that crosses jurisdictions. This capacity enables practicing managers working in different positions of responsibility within the community's emergency management system to build quickly a shared knowledge base in reference to a specific threat. Shared information allows them to coordinate their actions more efficiently, thereby reducing the threat to the community and restoring threatened operations more quickly and effectively. In its public form, the prototype IISIS would allow community residents access to timely, valid information regarding the risk via any standard Internet browser, enabling them to mobilize their own actions more appropriately. Since status information about an incident will be summarized from the professional browser and served to the public browser, IISIS will create a bridge for information exchange between professional managers and the public in conditions that require community-wide action.

Technical advances allow the integration of visual data with intelligent reasoning and interactive communication, using a blackboard model that provides virtual real-time information to support decision making for personnel operating in the dynamic environment of disaster operations. Logical inference by the computer will incorporate probability estimates to capture the anticipatory logic characteristic of disaster managers who seek to bring order to a dynamic, uncertain set of conditions. The prototype IISIS will function as a distributed, parallel processing information system that is expected to increase the efficiency of decision support for practicing managers. A working demonstration of the IISIS design may be viewed on the Internet at: http://quake.ucsur.pitt.edu:5555/

This report summarizes the developmental process for a prototype information infrastructure to support inter-organizational decision making and coordination for reduction and response to risk from hazardous materials at the University of Pittsburgh for the last seven months, October 15, 1998- April 15, 1999. The first phase, nearing completion, outlines the systematic development of a prototype IISIS. This developmental process has sought to produce three outcomes:

The prototype IISIS that emerges from this development will demonstrate the capacity of the University community to engage in a self organizing process of reduction and response to hazardous materials and other risks on campus. This demonstration will reflect the work done by the participants in the process, and represent their commitment to the campus-wide goal of risk reduction and response. The research issues involved in this project are both technical and organizational. The findings from this project will serve the critical function of documenting the design, functions, and implementation of a sociotechnical prototype for risk reduction and response. This project represents the first effort to translate a theoretical model of a self organizing, sociotechnical system into practice for an actual community exposed to hazardous materials or other risks.

A major function of this developmental process has been to identify both the resources and the vulnerabilities of the University community that could escalate or reduce risk from hazardous materials. Some of this work has been done through the initiatives in emergency planning already underway (University of Pittsburgh Emergency Plan, 1997). Some of the work has yet to be defined. No integrated knowledge base at the level of detail for individual buildings, departments, and rooms that is scalable to a campus-wide profile for effective disaster response existed prior to the project. The task of building a current, valid knowledge base to support inter-organizational decision making for reduction and response to disaster for the University, or any community, is substantial. This developmental process has begun the task, but there is still much work to be done to support informed, inter-organizational decision making among administrative units responsible for protection of life and property of this community of 32,000 people. The tasks involved in the developmental process are both organizational and technical. While these are substantively different sets of tasks, they have reciprocal effects on performance of the campus response system. Each set of tasks will be discussed separately to make explicit the steps necessary to achieve effective performance within their functions, but information will be exchanged between the organizational and technical teams as they explore the constraints and possibilities that each brings to the design of a functioning sociotechnical system. The result will be a more informed, broader group of policy makers who understand the limits and possibilities of both technical and organizational components of the University system, and who will be able to mobilize the strength of one component to offset the vulnerability in another. This is the benefit of a self organizing system in practice.

Some of the initial analysis of actors, tasks and responsibilities for action in emergency response was

included in the University=s Emergency Response Plan, prepared in 1997. The project=s planning process has built on this work, moving to a more detailed and specific analysis of the information flow within and among University administrative units. This analysis has identified points at which the timely exchange of valid information among participating actors enables the system to reallocate its attention and resources in order to absorb changes in its operating environment and continue to function with reasonable effectiveness. It also identified points at which the operations of the system may be overwhelmed and require external assistance or a declaration of a state of emergency. In effect, the developmental process for the prototype has explored the flexibility and limits of the campus organizational system to adapt to altered circumstances. Achieving the most efficient and effective flow of information through the University response system has both organizational and technical requirements. Planning for these two sets of functions must proceed with close coordination among the organizational and technical personnel responsible for the University system.

While the overall goal of the IISIS Project is the design, implementation, and evaluation of a decision support system to support coordinated action to reduce risk, this report summarizes only the developmental phase as it has been carried out in the field environment of the University of Pittsburgh. The prototype focuses specifically on hazardous materials risk, a severe threat to the campus, but the general design for decision support will be applicable to other types of hazards as well. The research issues in the development of a functioning decision support system are complex, and warrant careful planning, deliberation, review and revision by the relevant parties. The objectives of the developmental process are:

Data collection for this developmental process sought to identify the basic characteristics of a response system evolving over time for a community exposed to risk from hazardous materials, as well as the threshold points of change in the evolution of a response system. Data collected at the administrative unit level is organized for the knowledge base according to known standards, e.g. the National Spatial Data Infrastructure (NSDI) standards. Adhering to professional standards in the development of the knowledge base will facilitate common interpretation and shared meanings under the urgent pressure of an actual disaster.

Four types of data have been collected in the development of the IISIS prototype. These types include data regarding the existing conditions of University operations, as well as the organizational, technical, and operational characteristics of those functions. Data have been collected using several methods. First, data regarding the current state of the University=s operations were gathered to provide an initial profile of its geographic location, academic mission, size, structure and population from a review of documentary materials and direct observation. Second, eight administrative units were identified as having primary and supporting roles in risk assessment and response for the University community. These units include: 1) Facilities Management; 2) Campus Police; 3) Environmental Health and Safety; 4) Emergency Medicine;

5) Telecommunications; 6) Registrar=s Office; 7) Human Resources; 8) Computing and Information Systems.

With the assistance from staff of Computing and Information Systems, key types of information essential for response to a campus emergency were identified in their respective databases. Permissions were sought to access those databases, with proper authorization, for specific purposes of emergency management. The requisite permissions were granted by seven of the eight administrative units. Campus Police has a separate, proprietary database that can only be accessed through its vendor. The department, further, has legal requirements for confidentiality of information stored in its database.

With clear understanding that the data would be used only for risk assessment and emergency response, database links were created among the specified databases that support operational decisions by emergency response personnel. These databases included data on faculty, staff, and student personnel; class registration, pedestrian traffic on campus, assets, equipment, and types and quantities of hazardous materials stored on campus. The database links created a distributed system among the designated University units.

Third, technical data were gathered to provide accurate information on the technical infrastructure of the University. These data included a digital map of the campus with its buildings, floor plans for selected buildings on campus, detailed plans for the distribution networks of the lifeline systems: water, gas, electricity, sewage, telecommunications, and locations of the major points of intersection among the lifeline systems. Fourth, patterns of information search, strategies for damage assessment, procedures and means of transmitting information in response to risk were identified through interviews with managers of each of the eight administrative units responsible for emergency action. These patterns, identified separately for each administrative unit, serve as the basis for identifying the threshold points of change and limits of a self organizing system for disaster reduction and response.

Although the University is a legal jurisdiction responsible for life and property within its boundaries, it is dependent upon the City of Pittsburgh for critical fire and medical services. Since the University is located within the jurisdiction of the City of Pittsburgh, two of its primary emergency response functions, by law, report directly to City of Pittsburgh Emergency Services. Any incident that involves fire is reported immediately to the City of Pittsburgh=s Fire Department. Similarly, any incident requiring emergency medical service is reported immediately to the City of Pittsburgh=s Medic Command, which is located on the University campus but operated as a city-wide service. This relationship requires the University to coordinate its emergency response services closely with the City of Pittsburgh. The IISIS knowledge base will be used to support coordinated action among the professional participants in an evolving response system at the University of Pittsburgh and, in event of escalation of danger, between the University and City emergency services. In later development, IISIS will have a public browser that may be accessed by the wider community and the nation.

The set of expert interviews characterized the first level of information search, transfer and organizational learning in response to a threatening event for each administrative unit. Based upon these interviews, we have developed a map of potential information flow among the administrative units under emergency conditions. In one interesting finding, three managers described their risk assessment practices in similar terms. Each reported independently that he formed a mental picture of Agood performance@ in his area of operations. He judged the degree of risk or danger to operations in a given situation by the degree of discrepancy he observed between the actual situation and his previously developed mental picture. Each built his model of good performance on previous experience with the building, equipment, machines, or staff involved in operations. This finding confirms a theoretical model of problem solving (Weick, 1993) that shows human beings are able to assess risk or danger more quickly through visual and aural clues than by following procedural rules.

In continuing development, a second level of detail would be added for each of the administrative units. This effort would create a departmental knowledge base that would guide the daily operations of each administrative unit, so the data would be current and maintained by unit staff. In an emergency, with proper authorizations, these departmental knowledge bases would be accessed to provide timely, valid information to the campus emergency coordinator through the IISIS distributed network. Specifically, eight administrators with responsibilities for different types of operations defined under the University plan were interviewed to elicit their judgment regarding the critical points of decision within their units that activated emergency response, as well as the points of coordination among the different University units in response to threat. These judgments indicate the threshold points of change in the evolution of a response system, as well as the points at which the system would lose coherence.

The intelligent reasoning component for the IISIS is in its initial stages of development. The current design of IISIS uses a blackboard system that allows opportunistic problem solving(Nii, 1986) based upon the specified responsibilities and resources of participating members of the response system. That is, incoming information regarding an actual problem is posted on an electronic blackboard that is accessible to all of the authorized administrative units of the university in their respective offices or stations. As administrators from each unit read the message, they contribute information from their knowledge and experience for the solution of the problem. The evolving knowledge base serves as the focal point for coordinating action among the administrative units, as each unit adjusts its response to the incident, informed by knowledge of the other units= actions, resources and capacity.

This model fits the theoretical design for a self organizing system. Standard operating procedures specified in the University=s Emergency Plan can be adapted as parameters for emergency operations by individual units. During this period of development, we have also explored two other forms of intelligent reasoning as complementary to the blackboard system. One is an intelligent reasoner, Genie, developed by Marek Druzdzel in the School of Information Science, University of Pittsburgh. This program identifies the goal of a system, the component variables of the system, and the relationships among the variables. The reasoner then calculates the probability of risk or action for the whole set of variables, or the system. This program is still in its developmental stage, but the logic is clearly traceable across sets of conditions, interactions, and jurisdictions. The third reasoner that we have explored is NetWeaver, developed by Michael Saunders of Penn State University and Bruce Miller, Rules of Thumb, Inc. This reasoner uses fuzzy logic to approximate the often ill-defined, dynamic conditions of emergency environments. These technical issues are currently under development.

Data from the expert interviews served as the basis for mapping the flow of information through the university administrative units in event of a hazardous threat. This map identifies points where coordination with other units is essential and where gaps in the current communication patterns may exist in University response procedures. Next, we constructed a system diagram to show the current knowledge base used by each unit for decision support, with its existing software for access, storage and format of data, and noted points of compatibility or difference among them. From the interview data and the system diagram, we identified a preliminary set of information requirements that is essential for coordinated action in response to disaster. This set of information requirements will be reviewed, revised and validated by the practicing managers. The result will be a map of the decision process among interacting administrative units, showing the exchange of information essential to support timely, coordinated response by the University system. This analysis follows the organizational audit and analysis techniques developed by Kenneth Mackenzie (1984) for dynamic organizations. This set of methods produces a task/process matrix, or detailed specification of who performs what tasks using what resources within what time frame for the organization. It reveals not only the points of communication and reinforcement in organizational action, but also the gaps in communication and the likely points of vulnerability over time.

An important function of the IISIS prototype is to begin building the knowledge base for a University -wide decision support system, integrating three technologies: interactive communication, GIS, and intelligent reasoning. On campus, interactive communication is conducted through using the University intranet and the Oracle Web server. Security is established by setting clear criteria for authorized use, and extending access only to those professionals with responsibility for decisions during emergency events. If coordination should escalate to the city jurisdictional level, the same security procedures would be followed. Escalation of an event to city jurisdiction would require establishing agreements between the city and the university for city emergency response units to have access to the university knowledge base under emergency conditions. This is the kind of agreement that must be specified in advance, with proper authorizations granted and constraints accepted by the participants in the response system.

We are creating an initial geofile for the campus, which includes detailed plans of selected buildings and lifeline systems. These spatial data files will serve as a basis for a University GIS. We are currently developing examples for the intelligent reasoning component. These examples incorporate criteria for decision specified by the respective administrative units and draw on the University knowledge base to provide a dynamic assessment of risk. An important function of the IISIS Project is to design and monitor the integration of these three components for robust performance under disrupted operating conditions. This work is being done in conjunction with the faculty and industry experts in computation and system integration.

Actual implementation of the IISIS prototype on the University campus or elsewhere will require a clear commitment of both organizational and technical resources. The two components of the system are interdependent. One will not function well without the other to create a vital, self organizing system. Yet, the model indicates that increased efficiency in disaster mitigation and response will accrue from increased coordination among administrative units and members of the University community, facilitated by an interactive, intelligent, spatial information system.

NOTES