Methodology | Geob370

The first step in our analysis was finding the proposed locations of the nuclear repository. This was accomplished by selecting and extracting them from a populated place names layer provided by DMTI. The location of the Pickering power plant was determined then drawn into the map with the editing tool (See map 1).

Methodology

Step 2: The Infrastructure Layer

To create the infrastructure component of the cost surface, he railway vectors for each of the three provinces were joined together and displayed on the map (Map 2). They were considered with respect according to their track type codes, designating them as normal track, bridges, or tunnels. These three were selected to be weighted differently to represent important infrastructure types, affecting the safety of the cargo.

Step 3:The Population Density layer

Figure 3: Analysis Flowchart

Step 4: Combining the Layers

Raw population data was taken from the 2011 Canadian Census which was retrieved from the Statistics Canada website along with a shapefile of the dissemination blocks in Canada. Data for the provinces of Saskatchewan, Manitoba, and Ontario was isolated in both datasets then joined together. This data was turned into point data and reprojected. This was done so that when the population density map was created there would be no double counting of the population and gridded cells were a constant size. The kernel density tool was then used to create a population density map.

Step 1: Locating Proposed Sites

To create the final cost layer, the infrastructure and population density layers were normalized on a linear scale. We then combined the normalized rasters through an equally weighted sum (See figure 1).

The output layer was subsequently reclassified to reflect all 5 track types, which were assigned their respective friction value (table 2). During reclassification, care was taken to ensure that bridges and tunnels bypassing intersections did not result in a mismatched patchwork of of cells (see map 3).

Map 1

We wanted to account for the neighboring effect of population, i.e. to extent the impact of density populated areas below the constraints of their cell boundaries. To do this we used the focal statistics tools to survey the ares surrounding the tracks. This tool calculated the mean of all cells within a 20km radius of the cell. The rail lines were then used to extract a mask from the output, resulting in a layer that showed population density values along the tracks (see map 4).

We also chose to include intersections between roads and the track as points of potential risk. These points were considered to limit public interaction with the waste carrying trains, and to reduce contact between motorists and railroads as much as possible. Intersections were broken down into two groups, minor and major, dependent on the classification of the roads (either local roads or highways). Their locations were determined by intersecting the roads with the rail, producing points. These points were buffered to ensure overlap with the railroads, and were converted to raster. Once all the track types and intersections had been rasterized, they were added together through the raster calculator (see map 3).

This operation was repeated with the weights of the cost surface adjusted. We chose weights which reflected policies avoiding contact with the population, in strong favour of distancing the waste trains from populated areas. The AHP weights were calculated with a Collaborative Decision maker, provided by MakeitRational.com, and are displayed in figure 5.

Map 1: Source and Potential Sites

Map 2:

Using the cost distance tool, two cost surfaces were calculated for the different weights. Finally, the cost paths were drawn. Refer to figure 3 for a visual summary of all GIS operations performed.

Figure 2: Relative Weights

Map 3: Vector to Raster

Table 2: Track Friction Values

Map 4: Population Density

Figure 1: Weighted Sum

Source: DMTI Spatial

Source: DMTI Spatial, StatsCan

Source: DMTI Spatial

Source: MakeItRational.com