Objective 2: To develop a GIS based multi-criteria evaluation (MCE) model to determine the best site for UAV launch port
For this objective, a multi-criteria evaluation (MCE) is formulated and utilized through the combination of the factors identified in objective 1. MCE are often used for the purpose of site selection for multiple uses (Charabi & Gastli, 2011; Latinopoulos & Kechagia, 2015). Particularly, this method has been used in the selection of sites for airports which for the purpose of this study is considered analogous to a drone launch port (Ballis, 2003). MCE uses raster data which is “a matrix of cells (or pixels) organized into rows and columns (or a grid) where each cell contains a value representing information” (ESRI, 2016). In this way, MCE's work to arrange complex problems that deal with multiple criteria into an index of finest ranking for best scenarios through the manipulation and combination of the values applied to raster cells in multiple layers so that the final composite presents an option that fulfills the criteria most optimally (Charabi & Gastli, 2011; Ahmed et al., 2016).
First, using Boolean operations, sets of constraints are combined to create a land use constraint map distinguishing unsuitable areas from suitable areas. Land use factors identified in objective 1 are reclassified using the reclassify tool so that raster cells representing unsuitable area are assigned a value of 0 representing unsuitability and the rest are assigned a value of 1 for suitability. In this case, any cell with a zero when overlaid remains a 0 while only if both overlaid cells are non-zero are the assigned a value of 1 (ESRI, 2016).
Next, a set of criteria maps will be created standardized on a scale from 1 (least suitable) to 100 (most suitable). Evaluation criteria are measured along a continuous scale with the aim to either "enhance or to detract from the suitability of a specific alternative location" (Latinopoulos & Kechagia, 2015). These variables include land cover classification in the land cover suitability map and proximity factors discussed in objective 1. Within the land cover suitability map, values of either 0, 50, or 100 are assigned according to literature based justification of factors from objective 1. Due to the heavily forested nature of Northern Ontario, forests could not be entirely avoided and so heavily treed areas are given a value of 50 that will encourage their avoidance in favor of more suitable land cover for construction but will ultimately permit construction if more suitable area is unavailable and sparsely treed areas are considered highly suitable and given a value of 100. A python script was created with assistance to create multiple geodesic buffers around target remote communities, preexisting transportation infrastructure, and Thunder Bay. Criteria for the MCE are standardized on a scale of 0 to 100 using equation 1:
Standardization = (input value of cell - minimum value of criterion data assessed)/(maximum value of criterion data assessed-minimum value of criterion data assessed). (Bonnycastle, Wanhong, & Mersey, 2017).
Then, the analytical hierarchy process (AHP) and pairwise comparison matrix were used to define weights which allows for the ranking of factors above others in a way which permits the prioritization of certain factors over others (Ballis, 2003; Saaty, 2008). To do this, a scale of absolute numbers was used from 1 to 9 to indicate how many times more dominant one element is over another with 1 representing equal importance and 9 representing extreme importance of one variable over another as is outlined in Table 5.2. (Saaty, 2008; Ahmed, Shariff, Balasungram, & Abdullah, 2016). These comparisons combined into a nine-point comparison matrix as shown in Table 5.3 to determine relative weights for selected criteria based that prioritizes appropriate factors as were identified through our research as well as by the purpose of our research project (Saaty, 2008).
Weights are then assigned and applied to their associated constraint or criteria map in our final suitability calculation using the following formula in the Map Algebra calculator in ArcGIS as can be seen in equation 2:
SUIT = C1*C2*...Cn*(W1F1+W2F2+ ... WnFn) where C1 refers to constraint one and W1 to weight of factor one (F1) (Latinopoulos & Kechagia, 2015; Bonnycastle, 2018).
Finally, from the final suitability map, an area that meets the size requirement is identified. To do this, the mean Focal Statistics tool was used in ArcGIS to determine clusters of cells with the highest suitability. From here, cells with the highest suitability score were extracted and converted from raster to polygons. In the atrribute table for the polygons representing suitable area, a field was added and feature geometry calulated to see which had areas larger or equal to 670ha. Three sites were suitable and from here the LCP analysis in Objective 3 is used to determine the site that had the lowest cost to connect Thunder Bay via pre-existing roads since one of the most important factors for this study is reducing transportation costs.