 | Identifying Corridors |
The technique generates accumulation surfaces from both the Start and End locations of the proposed power line. For any given location in the project area one surface identifies the best route to the start and the other surface identifies the best route to the end. Adding the two surfaces together identifies the total cost of forcing a route through every location in the project area.
The series of lowest values on the total accumulation surface (valley bottom) identifies the best route. The valley walls depict increasingly less optimal routes. The red areas in figure 6.2.2-1 identify all of locations that within five percent of the optimal path. The green areas indicate ten percent sub-optimality
The corridors are useful in delineating boundaries for detailed data collection, such as high resolution aerial photography and ownership records. The detailed data within the macro-corridor is helpful in making slight adjustments in centerline design, or as we will see next month in generating and assessing alternative routes.
6.2.3 Calibrating Routing Criteria
Implementation of the LCP routing procedure provides able room for interpretation and relative preferences. For example, one of the criteria in the routing model seeks to avoid locations having high visual exposure to houses. But what constitutes high visual exposure 5 or 50 houses visually impacted? Are there various levels of increasing high exposure that correspond to decreasing preference? Is avoiding high visual exposure more or less important than avoiding locations near sensitive areas. If so, how much more (or less) important?
The answers to these questions are what tailor a model to the specific circumstances of its application and the understanding and values of the decision participants. The tailoring involves two related categories of parameterizationcalibration and weighting.
Calibration refers to establishing a consistent scale from 1 (most preferred) to 9 (least preferred) for rating each map layer used in the solution. Figure 6.2.3-1 shows the result for the four decision criteria used in the routing example
The Delphi Process, developed in the 1950s by the Rand Corporation, is designed to achieve consensus among a group of experts. It involves directed group interaction consisting of at least three rounds. The first round is completely unstructured, asking participants to express any opinions they have on calibrating the map layers in question. In the next round the participants complete a questionnaire designed to rank the criteria from 1 to 9. In the third round participants re-rank the criteria based on a statistical summary of the questionnaires. Outlier opinions are discussed and consensus sought.
The development and summary of the questionnaire is critical to Delphi. In the case of continuous maps, participants are asked to indicate cut-off values for the nine rating steps. For example, a cutoff of 4 (implying 0-4 houses) might be recorded by a respondent for Housing Density preference level 1 (most preferred); a cut-off of 12 (implying a range of 4-12) for preference level 2; and so forth. For discrete maps, responses from 1 to 9 are assigned to each category value. The same preference value can be assigned to more than one category, however there has to be at least one condition rated 1 and another rated 9. In both continuous and discrete map calibration, the median, mean, standard deviation and coefficient of variation for group responses are computed for each question and used to assess group consensus and guide follow-up discussion.
6.2.4 Weighting Criteria Maps
Weighting of the map layers is achieved using a portion of the Analytical Hierarchy Process (AHP) developed in the early 1980s as a systematic method for comparing decision criteria. The procedure involves mathematically summarizing paired comparisons of the relative importance of the map layers. The result is a set map layer weights that serves as input to a GIS model.
In the routing example, there are four map layers that define the six direct comparison statements identified in figure 51 (#pairs = (N * (N 1) / 2) = 4 * 3 / 2= 6 statements) as shown in figure 6.2.4-1. Members of the group independently order the statements so they are true, then record the relative level of importance implied in each statement. The importance scale is from 1 (equally important) to 9 (extremely more important).
This information is entered into the importance table a row at a time. For example, the first statement in the figure views avoiding locations of high Visual Exposure (VE) as extremely more important (importance level= 9) than avoiding locations close to Sensitive Areas (SA). The response is entered into table position row 2, column 3 as shown. The reciprocal of the statement is entered into its mirrored position at row 3, column 2. Note that the last weighting statement is reversed so its importance value is recorded at row 5, column 4 and its reciprocal recorded at row 4, column 5.
Once the importance table is completed, the map layer weights are calculated. The procedure first calculates the sum of the columns in the matrix, and then divides each entry by its column sum to normalize the responses. The row sum of the normalized responses derives the relative weights that, in turn, are divided by minimum weight to express them as a multiplicative scale.
The relative weights for a group of participants are translated to a common scale then averaged before expressing them as a multiplicative scale. Alternate routes are generated by evaluating the model using weights derived from different group perspectives.
6.2.5 Transmission Line Routing Experience
Figure 6.2.5-1 shows the results of applying different calibration and weighting information to derive alternative routes for a routing application in central Georgia. Four routes and corridors were generated emphasizing different perspectivesBuilt environment (community concerns), Natural environment (environmental), Engineering (constructability) and the Simple un-weighted average of all three group perspectives.
While all four of the routes use the same criteria layers, the differences in emphasis for certain layers generate different routes/corridors that directly reflect differences in stakeholder perspective. Note the similarities and differences among the Built, Natural, Engineering and un-weighted routes. The bottom line is that the procedure identified constructible alternative routes that can be easily communicated and discussed.
The final route is developed by an experienced transmission line siting team who combine alternative route segments for a preferred route. Engineers make slight centerline realignments responding the detailed field surveys along the preferred, and then design the final pole placements and construction estimates for the final route.
The ability to infuse different perspectives into the routing process is critical in gaining stakeholder involvement and identifying siting sensitivity. It acts at the front end of the routing process to explicitly identify routing corridors that contain constructible routes reflecting different perspectives that guide siting engineer deliberations. Also, the explicit nature of the methodology tends to de-mystify the routing process by clearly identifying the criteria and how it is evaluated.
In addition, the participatory process 1) encourages interaction among various perspectives, 2) provides a clear and structured procedure for comparing decision elements, 3) involves quantitative summary of group interaction and dialog, 4) identifies the degree of group consensus for each decision element, 5) documents the range of interpretations, values and considerations surrounding decision criteria, and 6) generates consistent, objective and defendable parameterization of GIS models.
The use of Delphi and AHP techniques provides a structure for infusing diverse perspectives into calibrating and weighting criteria maps used in GIS-based routing. Traditional siting techniques rely on unstructured discussions that often seem to mystify the routing process by not clearly identifying the criteria used or how it was derived and evaluated. The approach described in this paper is objective, consistent and comprehensive and encourages multiple perspectives for generating alternative routes as well as thoroughly documenting the decision process itself. The general approach is readily applicable to other siting applications of linear features, such as pipelines and roads.
