Our second analysis mapped school locations in relation to public parks in order to identify Toronto schools that have more or less access to green space. We choose school locations in our analysis because they indicate where the daytime populations of children are. For the purpose of this map we were only interested in looking at public parks in relation to the locations to schools. Therefore, the first step in this analysis was to select relevant park polygons from the Toronto green space layer before analysis. Our net green space layer used in the previous analysis would not be appropriate here as it includes green space other than public parks which are not relevant spaces for children to enjoy the mental and physical benefits of green space.
The parks shapefile from the City of Toronto Open Data Catalogue included many green spaces that were not parks, such as traffic islands and golf clubs. However, non-park green areas such as these are not appropriate for children to play in, so the data had to be filtered. Using Select by Attribute in ArcMap, we selected polygons with the word ‘Park’ in it or ending in ‘Park’, using the query NAME LIKE ‘%PARK %’ OR NAME LIKE ‘%PARK’. The space after ‘park’ was to exclude Parkettes, which in Toronto are parks that are less than 0.5 hectares, and due to their size they are not an adequate space for children to play. Some parks have a word after the word ‘park’ (e.g. Humber Bay Park West), so we used the second percent sign to allow those parks to be included. After this, there were still some remaining polygons that contained the name park but were not parks (e.g. Detonia Park Golf Course). To remove these, we went through and manually deleted them. The possibility of error in this selection process is discussed in the Error and Uncertainty section.
After we created a parks layer, we performed a buffer analysis to find the total area of green space within a specified distance from Toronto schools. For our analysis we choose a 500 meter radius around schools as our buffer distance. This value is arbitrary, however we believe that 500 meters would be an appropriate walking distance from schools to parks. This distance would be short enough for teachers to take their classes outdoors, for example. After our 500 meter buffer was applied we intersected the parks layer by the Toronto schools buffer layer. This filtered out all the park area that was not within the buffer, to minimize the amount of data and make it easier to work with. Then, we had to union the intersected park layer with the buffer layer to ensure that the area of a park polygon that extends beyond a buffer is not included in the final count of park area within that buffer. This allowed us to summarize the sum of public green space contained within each cell buffer by school name, producing a table output with the total park area within 500 meters of the school. Finally, we joined the output table to the original school layer based on the school names to add the area of public green space attribute field to the layer. To display the data, we quantified the schools by park area using Natural Breaks (Jenks) classification with six classes and a separate class for schools with no park area within the buffer.