Multiple-criteria decision analysis

Multiple-criteria decision analysis (MCDA) considers both quantitative and qualitative criteria when comparing alternative solutions to a problem. Decisions can be made by comparing various options and removing those that do not meet certain criteria. Of those that remain, possible interdependencies among alternatives are identified, sorted, and prioritized based on factors unique to a given decision. Ultimately, the MCDA process leads to selection of the best option meeting the decision criteria.

To collect air pollutant data in a near-road setting, the proper site must be chosen based on a set of relevant criteria. The purpose of any site selection process is to gather and analyze data that would lead one to draw informed conclusions regarding the most appropriate location of monitoring instruments.

As shown in Table 1 and Fig. 1, the site selection process is a series of steps, with each step having varying degrees of complexity as a result of real-world issues. The first step is to develop a set of site selection criteria. For example, if the project is attempting to collect air pollutant emissions from highway vehicles, then a criterion might be to locate a section of highway that has significant traffic volume. Ideally, one would develop a set of site selection criteria that would maximize positive and minimize negative impacts on the goals of the data collection activity. Maximizing a positive impact might be the inclusion of meteorology (that is, wind speed and wind direction) as a criterion. One would not want to select a location in which the air pollutant emissions would not be “blowing” toward the monitoring instruments. Minimizing a negative impact might be to ensure that the site is located away from other nearby sources of emissions, such as an industrial source or power plant.

Fig. 1 Decision process schematic.

Additional steps in the site selection process include (1) developing a list of candidate sites and supporting information, (2) applying site selection filters (coarse and fine), (3) visiting the site(s), (4) selecting candidate site(s) via team discussion, (5) obtaining site access permission(s), and (6) implementing site logistics.

Candidate sites

A list of candidate sites is developed using the site selection criteria established in Step 1. Geographic information system (GIS) data, tools, and techniques, as well as on-site visits by project team members, are the means of developing supporting information regarding potential sites. Other types of spatial data (for example, a street network) may be located or downloaded from GIS and other relevant Web sites. Nonspatial data (for example, meteorological data) may be downloaded from the National Climatic Data Center or other relevant Web sites. Site visits will provide information that is not readily available or provided or that is not easily gained from site maps.

Coarse filter

Team meetings are necessary to choose the most promising sites. This step involves a review of all the candidate sites, eliminating those that obviously do not meet the minimum criteria requirements. After applying the selection criteria as a set of filters, it may be possible to eliminate most of the candidate sites. For example, an obvious filter for a highway site would be to eliminate those sites with low traffic volume. Other filters should be applied to eliminate even more sites. Ideally, through the use of these filters, unsuitable sites are eliminated and the most suitable ones remain.

Ground truthing

An important component of ground truthing, or site visit, is to obtain information from local sources. Depending on the nature of the air-monitoring project, team members may need to meet with state/local air-quality staff, traffic engineers, or other relevant parties. All of these groups can provide valuable information regarding local conditions that would be difficult, if not impossible, to obtain otherwise. Too often, local resources are overlooked, and this can lead to poor decision making.

Site access

Another important consideration is site access. While site access may not be explicitly included in the selection criteria, it is critical. Even though a project may not directly affect property owners, they may be very reluctant to allow access to their property. While we would like to believe that property owners would permit access out of a sense of corporate or civic responsibility, they sometimes have a different perspective. Property owners may be reluctant to grant access for a variety of reasons, including liability, financial issues, suspicion of government activities, and so on. Therefore, access to any given property is not guaranteed, and researchers should be prepared for a long, involved process.

Logistics

Site logistics includes, but is not limited to, gaining access to electric power and communications connectivity, and arranging for security fencing. Any location where site logistics is restricted or prohibited because of administrative or physical issues is highly problematic and should be eliminated from further development.

Spatial tools

The use of spatial tools in decision processes is increasing and will continue to increase. Historically, the use of spatial tools (GIS) in decision processes has been somewhat problematic because of the magnitude of the data required by a GIS, the perception and reality of operating GIS software, and the level of knowledge required by end users to manipulate data in a GIS. In the last 15 years, GIS data have become more readily available in both quantity and quality, while the availability of a GIS in the Microsoft Windows® operating system environment, the availability of low-cost computer hardware, and the development and implementation of easy-to-use GIS tools (in a Windows environment) has made implementation of GIS-based decision-support tools more practical.

Typically, quantitative weighting criteria are associated with siting criteria, as well as elements of the GIS data layers (for example, certain types of highways would be more suitable than others and thus would have applicable quantitative values). It may not always be possible to explicitly assign quantitative values to selection criteria. However, as the site selection process proceeds, it may become apparent that quantitative values are being implicitly assigned by team members. For example, sites with high traffic volume are more highly valued than sites with lower traffic volume. Thus, while it may appear that the decision process is based on a high degree of subjectivity, given that one might not explicitly assign quantitative values to selection criteria, the process of intrateam communication and application of selection criteria does lead to the selection of an appropriate site for the project.

GIS software may be used to create the maps for the site selection process. Spatial data may be located and downloaded from GIS or other relevant Web sites. Table 2 shows some typical sources of data that may be used for the site selection process. The use of maps for the site selection process is only a tool.

Selection

Following the application of the selection criteria, candidate sites may be further prioritized during team discussions, with the pros and cons of each site being debated. Typically, a primary site is chosen along with a backup site. This is to ensure that if something unforeseen occurs regarding the primary site during the early stages of project implementation, then the backup site is available for deploying the air-monitoring instruments.

In an air pollutant monitoring project, numerous organizations may be involved in the process. Some of these may be policy/regulatory groups that have a different mission and perspective pertaining to overall project implementation and more specifically site selection. Under certain conditions, policy requirements may supplant the researchers' inherent quest for knowledge as a reason to investigate certain phenomena. However, policy must never be divorced from sound science.

There is no perfect air monitoring site, as compromises will have to be made for any environmental study conducted within any area. It is a question of balancing benefits with risks and costs. The selection may be further complicated by external constraints. A principal constraint may be the legal mandate of a regulatory policy. Few, if any, design decisions can be made exclusively from a single perspective. These decisions can be visualized as attractions within a force field, where the center of the diagram represents the initial condition with a magnet placed in each sector at points equidistant from the center of the diagram as shown in Fig. 2 a. If the factors are evenly distributed and weighted, the diagram might appear as in Fig. 2 b. But as the differential in magnetic force increases, the stronger factor(s) will progressively drive the decision. In a particular case study, the decision might be most directly influenced by legal requirements, but it also needs to be scientifically credible and economically feasible (Fig. 2 c).

Fig. 2 Decision force field where (a) the initial conditions will be driven toward different influences. The stronger the influence of a factor (for example, high traffic volume), the greater the decision will be drawn to that perspective. Decision force field where (b) a number of factors have nearly equal weighting in a design decision. For example, if the site selection criteria are somewhat ambiguous, a number of alternatives are available, costs are flexible, and scientific credibility is minimally affected, the design has a relatively large degree of latitude and elasticity. Force field for (c) a decision that is most strongly influenced by policy/regulatory constraints and drivers. Note that all factors drive the decision, but the site selection criteria and other policy and regulatory issues have the greatest pull.

[Disclaimer: The U.S. Environmental Protection Agency through the Office of Research and Development funded and managed the research described here. The present article has been subjected to the agency's administrative review and has been approved for publication.]

See also: Air pollution; Decision analysis; Environmental engineering; Geographic information systems

Sue Kimbrough

Dan Vallero