Objective 3 - To apply the model to the city of Guelph to determine the most suitable location for marijuana dispensaries.
The model is applied to the City of Guelph to recommend the most suitable location of the marijuana retail store scheduled to open in July, 2018. The model's application is detailed in the flow in Figure 2. As the majority of data are in compatible shapefiles with ArcGIS, little conversion is required to apply the model. The census demographic data are added to the attribute table of the layer Cartographic Boundary Files (CBF), 2016 Census – Dissemination Areas, described below in order to apply the model. The model is only as accurate as the data used, however, the census data being used is frequently updated and produced in high quality to dismiss any possible errors.
Figure 2: ModelBuilder illustrating proximity constraints.
This portion of the model creates 6 outputs, 4 constraints which the marijuana storefront cannot be located in, and 2 areas which the storefront must be located within. The buffer tool was used to create the areas which the store could not be constructed in for each of these 4 constrinats, creating 4 non-allowable constrained areas. The 4 non-allowable land constraints, schools, park boundaries, are merged to create a final dissolved constraint layer, "constraintnogo_DISSOLVEDFINAL", which constrains out all areas which the store cannot be located within. The merge and dissolve tools were used to create this file as the total area which is constrained out is the only information of interest. Next, the two areas which the store must be located within, a commcercial zone, and a 400-metre buffer of a bus-stop are created and intersected to form a final allowable land area shapefile, "allowable_land_const" from which the intersected commercial and busstop constraints are erased. This portion of the model then creates a final constraint layer of all allowable land in each dissemination area which will be ranked based on the criteria described in Objective 1.
Figure 3: ModelBuilder illustrating proximity criteria of schools, parks, LCBOs, and bus stops.
The portion of the model found in Figure 3 demonstrates the construction of zonal statistics tables for each criteria, so that the average distance from schools, parks, bus stops, and LCBOs/Beer Stores may be calculated and standardized on a 1-100 scale for each dissemination area. These four vector shapefiles are converted to raster using the eucliddean distance tool, so that the distance from each of these attributes are calculated. Then, this rasterized data is converted back to vector, using the zonal statistics tool, averaging this distance in a vector attribute table, from which a simple field calculation can be ran to standardize the average distance of each dissemination area from each of the criteria on a 1-100 scale. The other two shapefiles, representing the denstiy of underage children, and population density are included in this portion of the model to visualize that these shapefiles were included in criteria formation, however no rasterizing tools such as euclidean distance were used as these layers are standardized based on area density which is accomplished using the field calculator. The output is 6 standardized attribute tables which are joined to the final shapefile of allowable areas, shown in Figure 2.
To apply the model, the specific weights of the factors were calculated using a pairwise ranking method that accounts for the significance of each factor. The order of importance of factors indicates proximity to schools being the heaviest weighted factor followed by accessibility. Proximity to sensitive locations, parks, and existing LCBO's received the same weight, while the age of surrounding demographics is considered to be of the least importance.Table 2 describe the process of determining the weights of each factor.
Table 2: Pairwise Rankings and Weigted factors for multi-criteria evaluation.
These tables provide the necessary information to construct the specific equation that will be applied to the MCE model. The consistency ratio (CR) is calculated to evaluate the logic of the pairwise ranks to assure that there are no flaws while determining how factors should be weighted compared to one another. Since the CR for our model is much less than 0.10, it has an appropriate level of consistency (Bonnycatle, 2018).
Using the structure of the equation outlined in Objective 2, the following equation will be calculated in an added field in the attribute table where each of the standardized suitability are joined.
MCE = ([SchoolStandard.SchoolStand]*0.414573878)+ ([ChurchStandard.Church_Standard]* 0.095725752)+ ([AlcStandard.AlcStandard]*0.095725752)+ ([ParkStandard.ParkStandard]*0.095725752)+ ([BusStandard.BusStandard]*0.22027846)+ ([Population_density.StandardPo]*0.038985203)+ ([GuelphDAAge.U19_standa]*0.038985203)
Running this calculation will output an attirbute field in this shapefile that highlighs the average sutaibility of the commercially zoned land in each dissemination area that satisifies the constraints.