Evaluating GIS for Disaster Management
Geography 596B
Bruce Kinner
Evaluating GIS for Disaster Management
Abstract
Most geographic information systems (GIS) are not designed specifically for the purpose of disaster management. This paper proposes a systematic methodology for evaluating an existing GIS to rate its level of preparedness for a disaster situation. First, a literature review surveysexisting methods of building a GIS and the use of GIS in disaster management are described. From this review a set of system criteria for evaluating a GIS for disaster management are identified. Using heuristic and cost-benefit analysis to explore the nature of an existing system it is possible to identify gaps and detail needed improvements. A case study of a power utility company is described to evaluate the heuristic scoring system. Armed with the knowledge generated from the proposed methodology, GIS practitioners can work to improve their systems to better withstand disaster events.
______
As long as people have been building Geographic Information Systems, there has been debate and discussion on the best way to construct them and what characteristics they should have. Numerous methodologies have been developed for establishing the goals and overcoming the challenges of creating a GIS. Likewise, since the beginning of human enterprises, people have looked for ways to manage disaster situations. GIS is now entrenched within many organizations - both public and private. The focus of these systems is increasingly moving beyond the initial goals to incorporate diverse spatial data and perform a wider array of spatial analysis tasks for an institution.
At the same time, world-wide attention to major disasters such as 9/11, Hurricane Katrina, the Indian Ocean tsunami, and the 5/12 Sichuan earthquake in China has raised disaster management to a higher level of human consciousness. GIS is now being applied to tasks throughout the disaster management cycle (Leitinger, 2004; Miller, 2006). A new evaluation methodology is needed to characterize the ability and readiness of a GIS to contribute to disaster management within its area of operational control.
In the GIS literature there is extensive discussion of both: 1) appropriate methodologies for establishing a GIS and 2) taxonomies of the features of a GIS. Similarly there are a variety of case-study applications of GIS in disaster situations, typically focused on handling a particular type of disaster. What is lacking is any synthesis of this work to develop a systematic methodology for evaluating an existing GIS to rate its level of preparedness for a disaster situation.
This paper outlines a new methodology for rating a GIS’ ability to withstand a disaster situation. Through heuristic and cost-benefit analysis a GIS can be critiqued to provide direction on system weaknesses before a disaster situation occurs. In order to address this gap in the literature, existing methods of system building and GIS disaster management are reviewed and a gap analysis is performed. Based on this analysis, a set of heuristics are proposed to supply a testable set of system criteria. This framework is analyzed through a case study to explore its effectiveness and direct further refinement. The case study evaluates a utility enterprise GIS at Alliant Energy, a mid-western US electric and gas utility. Recently, Alliant Energy has dealt with major ice storms across Iowa and historic flooding in Cedar Rapids. The experiences of this GIS will be used to improve understanding of how to evaluate the readiness of a GIS before a disaster actually occurs. Lastly, possibilities for future research are addressed.
A. Functional Taxonomies for GIS Design
As GIS became a widespread technology, significant research effort was focused on designing GIS and categorizing the components of a successful system. Among the first to present a holistic GIS taxonomy, Calkins' (1991) focus is recognizable as the core of modern GIS: effective and successful uses of GIS within an organization that deliver the benefits of geographic data. Similarly, a study by the SUNY (Center for Technology in Government, 1995) attempts to identify the value of a GIS and define the scope of the cost and benefits of a public sector GIS in order to minimize the costs and maximize the benefits. Building a departmental organization to manage the GIS and using cost-benefit analysis to justify its existence is at the heart of this style of GIS building.
From these and other studies, such as (Albrecht, 1996; Behr, 1995; Gillespie, 1994; Guptill, 1989; Harvey, 2001 and 2004; Nedovic-Budic, 1999; and Yuan, 1995) common themes of GIS roles, benefits, and components can be distilled into five main categories (Figure 1):
- Geographic data
- and a (GIS) technical infrastructure
- managed by a departmental organization represent the costs to providing the
- benefits of effective
- and successful use of GIS to the larger Organization.
Geographic data represents not only the data but also the activities taking place within the GIS using this data: spatial analysis, representation, and products (maps) are the primary functions covered under this heading. Base data and data sharing efforts are also included here.
The technical infrastructure is not only the hardware and software to run the GIS but also maintenance contracts, upgrades, and in-house programming and support.
The organization is a department level bureaucracy responsible for coordinating the activities of the GIS and the personnel necessary to run and maintain it. The costs of salaries, supplies, training, and overhead are generally subsumed at this level.
These three categories represent the costs of the GIS for cost-benefit analysis purposes. Methods for establishing a GIS weigh these costs against the expected benefits in the remaining categories.
Calkins (1991) applies the terms 'effective' and 'successful' to characterize uses of GIS in his taxonomy, although it is not entirely clear what distinction is being made between the terms as they are used somewhat interchangeably to indicate a positive impact of the use of geographic data. In order to provide clarity and incorporate others' categorical distinctions, here the two terms are differentiated between the low-level operational/task level and a higher decision-making/strategic planning realm.
Effective use of the GIS covers lower level benefits at the task or operational level. These small scale benefits can be categorized as: data quality, metrics, which users will benefit, if the benefits are internal or external, what are the effects of the benefits, the level of importance of the benefits to users and decision makers, and does it generate revenue?
These individual benefits are then rolled up to a higher level of successful use of GIS in a performance based web to the larger organization in which the GIS serves. Here the benefits help identify stakeholders, enhance decision-making, and integrate with and support other enterprise systems; corresponding directly to the mission statements of both the GIS and the greater institution.
B. GIS for Disaster Management
In a disaster management situation a GIS must respond quickly in order to provide effective and successful benefits. As a result, research has focused on a slightly different categorical structure for a disaster management GIS (Cutter, 2003) as compared to the cost-benefit structure shown in Figure 1. The main difference is the emphasis on the specific components of a GIS which are critical during a disaster.
First, the GIS technical infrastructure must be able to survive the disaster in order to respond to the event.
First, the GIS technical infrastructure must be able to survive the disaster in order to respond to the event. - The GIS can then communicate information in the form of base data and
- spatial analysis
- with dynamic displays and intuitive interfaces
- through a distributed computing solution.
While the two models appear similar, the foci are different. The GIS technical infrastructure may no longer exist post-disaster and may need to be either re-built from the ground up or restored from back-up in another location. Interoperability between systems may no longer work and data may be difficult to access. The core question becomes: what constraints might exist to prevent practitioners from using the system during a disaster? Identifying these constraints in disaster planning scenarios allows a discussion of contingency planning for the technical infrastructure to emerge in order to identify what is necessary for the system to remain operational. This category also includes the process for the capture of any post-disaster data necessary for response and recovery.
Base data takes on a different meaning in a disaster context. Usually, new data, such as post-event imagery, must be acquired and integrated into the database in order to move forward with an effective response. Yet, important characteristics of the data such as scale, uncertainty, and other measures of quality and quantity must not be overlooked in the rush to deliver information. If real-time field data will be used its quality, timeliness, and effectiveness must be critically evaluated.
The cornerstone of GIS disaster response is spatial analysis. It is these functionally oriented tools and methods performed within the GIS which allow the information of where, what, who, and how much to emerge. Currently most spatial analysis is performed by expert users and is usually specific to the type of disaster.
Communication is at the core of the disaster management process. The role of GIS in a disaster is to produce timely and relevant information for decision-makers in the field and in the command center (Cai, 2006). To maximize the ability of the GIS to communicate with these users the information should be delivered through dynamic displays which allow users to control views. Additionally, user interaction with these displays should be through intuitive interfaces which users can quickly master to maximize their cognition of the information presented through the GIS. This style of data sharing maximizes the human interaction between the providers and users of the GIS data (Zerger, 2003) and increases user cognition of the information being presented. In addition to these technical aspects, the 'soft' skills needed during crisis are often not given enough weight in research. These non-technical issues can be critical to a successful transition of information from the GIS to the users in a people-driven setting.
Finally, a robust GIS solution for disaster management works best when it uses a distributed computing solution. This means users do not have to rely solely on GIS personnel at a singular location for their information and data is available in a service-centered way via multiple portals (Neuvel, 2006) such as mobile devices, command centers, inter/intra-nets, or wireless delivery. Ideally, a distributed solution (such as the IMS example in Lembo, 2008) is integrated directly with the data without intervening systems or middleware. Meeting end-users needs in this way is currently a major practical hurdle.
C. Current Methods for Evaluating a GIS for Disaster Management
Lack of preparation for a disaster increases risk. Likewise, a GIS unprepared to assist in any phases of the disaster management cycle can do little to help in a crisis. It is a false assumption to expect the existing base data and technical infrastructure to be adequate in the face of disaster (Cutter, 2003). A disaster is not the time to identify problem areas. These gaps are often spurred by a common situational factor: disaster management is not usually the primary function of a GIS but the GIS needs to be useful during disaster management. Since the demand for a disaster management GIS is not constant it is relegated to a secondary function by default. This makes the GIS more prone to failures during a disaster - especially ones related to system usability (Zerger, 2003).
One approach to solve this problem is to conduct disaster studies to determine the kind(s) of geographic data and technical infrastructure used to meet the challenges of a specific disaster. However, these studies rarely tie applied experiences into the larger-scale organizational issues influencing the design and use of the GIS. While post-disaster studies are very helpful for identifying appropriate response methods, the lessons learned from such studies are generally limited to a specific unique disaster.
Another method for identifying gaps between standard GIS operations and disaster management tasks is to describe what not to do (Zerger, 2003). A separate disaster management GIS is of limited utility compared to one used on a regular basis. Non-expert users do not have the time during a disaster to learn new systems. Therefore, while a different data model may be used to represent disaster data for technical reasons, the interfaces and displays for the user must be familiar. During a disaster the GIS must be delivered in a manner that does not impede operational efficiency of emergency tasks.
The system diagrams from Figures 1 and 2 show the base components of GIS do not change with a disaster. Instead, they show a shift to specific data sets and analysis techniques delivered in near/real-time. This shift of focus means there are different user requirements from traditional GIS. Not only are the requirements of a GIS during a disaster different from standard operations but the requirements of the GIS are different at different phases of the disaster management cycle (Zerger, 2003). However, the shift of focus is not so radical as to invalidate methods traditionally used to establish or evaluate a GIS. As long as the special criteria of disaster management GIS are considered, such studies can still be helpful in evaluating GIS preparedness.
For example, Guptill’s (1989) study describes a method for evaluating GIS technology through user requirements. Extrapolating this method to disaster management - if the benefits of supplying the GIS data for a disaster outweigh the costs, a GIS solution makes sense. Guptill provides a list of steps for defining user requirements – all of which are relevant in an emergency context. From user requirements, applying cost-benefit analysis is an appropriate next step and, “...one way to facilitate the measurement of effectiveness is to focus on specific applications…” (Gillespie, 1994): disaster management in this context. As a compliment to this work, four categories of GIS benefits have been suggested (Behr, 1995): operational (task completion), efficiency (time), strategic (market share/department goals/public service), and external (customers/public).
By understanding the specific user requirements, identifying the benefits, and determining the costs, cost/benefit analysis can show if: 1) the standard GIS applications in place are sufficient, 2) applications previously considered out of reach (from a standard GIS point of view) are now more practical (in a disaster management context), or 3) if entirely new solutions are needed. A method which evaluates a GIS for user requirements, costs, and benefits associated with each phase of the disaster management cycle would provide the necessary information to discover the readiness of a particular GIS and could provide more information to support detailed gap analyses. The next section outlines a proposed framework intended to provide the needed details.
D. A Heuristic Method for Evaluating a GIS for Disaster Management
Disaster management GIS needs to move beyond data-centered operations to functionally oriented ones. Data-centered operations (collection, organization, spatial data handling and manipulation) are those which can be identified and completed before a disaster (Goodchild, 2003). If the data is to be collected during the disaster: the methods and processes (the functional operations) for doing so should be designed, tested, rolled-out, trained on, and critiqued ahead of time. The disaster data model does not need to be the same as the standard operational GIS, but must “reflect the mental models of both the GIS experts” (analysts/technicians, ie. system designers) and the first responders (users) (Yuan, 1995). Spatial views can be used to provide the specific database fields needed for disaster operations. It is these more specific criteria of disaster management GIS that must be included in any evaluation of the system.
Unfortunately it is common for a GIS to not be well integrated into an organization's disaster management cycle until after a disaster strikes. This results in little contribution of the GIS to the preparation and mitigation phases. Yet both of these stages provide many opportunities for GIS to deliver benefits through topics such as modeling and risk maps (Neuvel, 2006). In order to avoid a lack of effort in these stages of the disaster management cycle, an evaluation of the GIS should being with general hazard analysis in which the GIS is emphasized. As material is developed during this process, the nature and form of the appropriate GIS response will materialize.
Designing a GIS to function during disasters should employ test simulations where there is adequate time for discussion, instead of in a real disaster when tensions are high and decisions must be made quickly. Hazard analysis using scenarios (such as FEMA's Emergency Planning Course) is an evaluation method allowing for extensive information gathering and performance testing to be completed and is one of the best ways to discover potential weak points in a system. These analyses should be done in concert between GIS staff and emergency personnel in order to involve decision-makers as early as possible (Cai, 2006). A small team of GIS and emergency managers along with experts from both the GIS side and first responders is ideal. The idea is to capture the knowledge of the first responders in such a way the GIS staff can begin to frame and visualize the data and processes needed to supply critical information. Preferably, there should be significant and seamless integration between the GIS and the distributed solution (Lembo, 2008).
These steps are similar to the initial design steps of Tomlinson's GIS planning methodology (Tomlinson, 2003). The first four steps of his process: 1) Consider the strategic purpose, 2) Plan for the planning, 3) Conduct a technology seminar, and 4) Describe the information products: are useful for bridging the gap between an existing GIS and a disaster management team. Similarly, the purpose of processes defined here can help develop tangible goals and identify key issues for implementation. The business-oriented approach of Tomlinson's process also meshes well with a cost-benefit perspective of GIS design.