Environmental Rankings of Toronto Schools: A Pilot Project

Introduction:

As part of a professional experience course at the University of Toronto, undergraduate students Kevin Yuan, Kasandra Tworzyanski, and Nick Chan conducted a pilot project for CANUE over the academic year. The project aimed to assess the environmental conditions within schoolyards and surrounding school streets by developing a ranking based on five key environmental variables: air pollution, noise pollution, tree canopy coverage, street trees, and access to healthy food outlets. Rankings were determined using the values of these variables within a defined area. The analysis focused on 10 randomly selected public schools in Toronto. Each school received a ranking reflecting the quality of its environmental conditions within and around the school.

To understand the environmental exposures students may encounter on their way to school, an analysis was conducted in two parts. In part A, the schoolyards themselves were examined, where four of the five environmental variables, excluding access to healthy food outlets, were summarized based on their values within the schoolyard boundaries. In part B, environmental conditions around school streets were evaluated by summarizing values for all five variables within a defined distance from each school.

 

Methods: 

For the project, we have included both vector and raster data layers. The vector data layers are points indicating the locations of street trees and healthy food outlets, and were obtained from OpenStreetMap and the City of Toronto Open Data. A vector layer from the Toronto District School Board (TDSB) outlining school boundaries was used as a guideline for defining the school yards. The raster data layers we used include noise pollution, air pollution, and tree canopy rasters, obtained from CANUE’s public-facing platform, HealthyPlan.City. Figure 1 shows the three raster data layers mapped to the city of Toronto boundary.

Figure 1: Raster data layers depicting noise pollution, air pollution, and tree canopy coverage. In each map, darker shades indicate higher values, either more pollution or denser tree canopy.

 

Spatial analysis methods:

For part A, we used ArcGIS Pro to revise the 10 polygons representing school yards obtained from the TDSB. For part B, we used the network analysis method in ArcGIS Pro to create three walking isochrones for each school. Three isochrones indicate areas that can be reached by walking within 5, 10, and 15 minutes of the school. Figure 2 shows a sample of the isochrones created and an outline of one of the school’s boundaries, with the tree canopy raster overlaid.

For point data layers, we used the Spatial Join tool to count how many points are inside each polygon. For raster data layers, we use zonal statistics to summarize the average air and noise pollution levels and tree canopy coverage in each polygon. These averages and count values were obtained for each school.

Figure 2: The image on the left displays three walking isochrones surrounding a school, overlaid with a tree canopy raster where darker green indicates denser tree cover. The image on the right shows the polygon outline of a school boundary.

 

Ranking method:

The R programming language was used to rank the schools based on their score for each environmental variable, and the raw values of averages and counts were converted into ranks from 1 to 5 for each variable. Two separate rankings were created for parts A and B. An overall score for parts A and B was then calculated using a weighted ranking system. For Part B, the raw values were derived by averaging or totaling each environmental variable across the three isochrones.

The weights applied to each environmental variable were as follows: air pollution (40%), noise pollution (30%), tree canopy (25%), and tree points (5%). When healthy food outlets were included for part B, they were weighted at 5%, and the remaining percentage was distributed similarly among the other variables. These weightings were informed by external sources that highlight air pollution and noise pollution as the most significant determinants of adverse health outcomes (Carrier et al., 2016). 

 

Results:

A web map was created, which visualizes the results and final rankings of the analysis for all 10 schools. By clicking through the map, users can view the raw values of each environmental variable and see how each school ranks (from 1 to 10) for both the part A and B analyses. Images are also included to illustrate the physical conditions of the schoolyards and surrounding streets, helping to understand the environmental exposures. 

 

Figures 3 & 4 show a sample of these images for the best and worst-ranked schools, respectively.

Figure 3: Images of the highest-ranked school, ranked highly due to its large green space, tree coverage, and quiet residential surroundings, indicating low noise pollution.

 

Figure 4: Images of the lowest-ranked school in the assessment. The schoolyard appears significantly smaller with limited tree coverage. Its location near busy streets and close to downtown suggests higher levels of air & noise pollution.

 

Conclusion:

This project demonstrated how geographic data can be used to evaluate the environment in and around schools. Scaling up this work could serve as a tool for identifying inequities in key environmental factors within and around school environments across the city of Toronto or the entire province.

 

References: