Spatial Multicriteria Evaluation

Identifying Alternatives

Most spatial MCE procedures begin by identifying a set of alternatives or possible solutions to the problem of interest using standard GIS buffer, overlay, and selection functions. Depending on the problem, an alternative may be considered feasible only if it passes all constraints (conjunctive screening) or if it scores sufficiently well on one or more constraints (disjunctive screening).

Selecting Criteria and Standardizing Data Values

A criterion is a measurable characteristic or attribute that is used to judge the performance of each alternative relative to a normative objective (i.e., statement of what is desired). Criteria are represented in spatial MCE as either raster map layers or database fields associated with vector GIS objects. Criteria should be

• complete and minimal (representing all relevant aspects of the problem with the fewest criteria),
• nonredundant (no “double counting” of a specific variable), and
• independent (no autocorrelation of criteria values).
• Once the criteria have been selected, their data values are standardized to a common scale that ranges from 0 (worst) to 1 (best) to permit trade-offs between criteria. This scaling can be done relative to a maximum data value, a range, or a more complex nonlinear function.

Generating Criteria Weights

Criteria weights allow stakeholders with dissimilar objectives, interests, or experience to express how important they judge a criterion to be relative to other criteria. Criteria weights are expressed as real numbers that range from 0 (no importance) to 1 (highest possible importance), subject to all weights summing to 1.

Techniques to derive criteria weights vary considerably in mathematical sophistication, theoretical rigor, and cognitive demands. The procedures used most frequently are as follows:

• 1.
  Ranking (ordering criteria from least to most important)
• 2.
  Rating (distributing a fixed budget of “points” [usually 100] among criteria)
• 3.
  Pairwise comparison (comparing all pairs of criteria and indicating the relative importance of criteria through integer ratios)
• 4.
  Trade-off (deriving weights from users’ responses to questions of how many units of change in Criterion A they consider equal to a unit of change in Criterion B, and so on)

Applying a Decision Rule

A decision rule is the mathematical approach used to rank alternatives by combining criteria values and weights. Weighted summation is the most frequently used decision rule. It calculates a score for each alternative by multiplying standardized criterion values by their corresponding weights. The results are summed, and alternatives are ranked from the highest score to the lowest. The analytical hierarchy process (AHP) is also popular. It decomposes a problem into a hierarchy of criteria and subcriteria. Alternatives are evaluated in a pairwise manner for each criterion and assigned a score ranging from 1 (A and B are equally preferred) to 9 (A is very strongly preferred to B). The criteria-based scores are summed to produce overall scores and ranks for each alternative.

...