Assumptions

To allow us to conduct our analysis, we made several assumptions. Firstly, we assumed that cherry blossom trees are a suitable proxy for “aesthetically pleasing” street trees. We made this assumption because Vancouver is known for its cherry blossom trees and people visit every year to see these trees in full bloom. There is a Vancouver Cherry Blossom Festival that is hosted annually by the Vancouver Board of Parks and Recreation. This festival includes over 20 events that celebrates the 43,000 cherry blossom trees that bloom across Vancouver each spring (Vancouver Board of Parks and Recreation, 2016). Secondly, we assumed that all cherry tree species in the City of Vancouver street tree data that contain the word “cherry” in the common species name are aesthetically pleasing cherry blossom species.

Based on the body of research on the social benefits that street trees provide, we made the assumption that planting aesthetically pleasing street trees near greenways, cultural spaces, healthcare facilities and low-income, low-tree density areas would benefit society. We thought that street trees near cultural spaces and greenways would provide social benefits as these are public spaces that are enjoyed by all, regardless of income or socioeconomic status. We made the assumption that locating street trees near healthcare facilities would benefit society based on a study that found that window views of nature reduced the number of post-operative hospital stays in patients that have undergone survey (Ulrich, 1984). The study recommends taking window views of natural scenes into account when designing hospitals. Therefore, we chose healthcare facilities as one of the criteria for our MCE analysis.

We assumed that setting the visibility of the street trees at 50m from greenways, 100m from healthcare facilities and 100m from cultural spaces was appropriate. This is because one would be viewing the trees from ground level at greenways, and so the trees would need to be closer.  For healthcare facilities, the trees would need to be further away for maximum visibility by patients looking out the windows of multiple floors of a hospital or care home. For cultural spaces, we assumed that a distance of 100m was appropriate given that these are places that people gather, so having lots of street trees very close to the space would reduce the ability of large volumes of people to pass through.

When conducting the regression analysis, we did not consider the potential effects of other trees that could already be planted, because within the scope of our project this was not relevant for assessing if there was correlation between cherry blossom tree density and population density, average income and property value.

Throughout our analysis we assumed that income is a social determinant of health and general well-being. This assumption was made because income is one of the many determinants of health that the World Health Organization uses to conduct Health Impact Assessments (WHO, 2016).

Preliminary Data Modification 

In the preliminary data modification stage, we imported a map of the City of Vancouver as our base layer as well as roads, health care facilities and census tracts. The base layer files were clipped to the Vancouver boundary. We obtained income, population density, and property value data for all census tracts in Vancouver from the University of Toronto’s CHASS database, and imported them to our map

We obtained street tree data from the City of Vancouver and applied a filter to the Excel sheet to isolate trees containing the word “cherry” in their common name. The street tree data included addresses but not geocordinates so we geocoded the addresses to obtain longitude using the website http://www.gpsvisualizer.com/geocoder/. We then imported and joined the tables to the census tracts, set the XY values to longitude/latitude NAD (1983), then converted these to feature classes.

Statistics & Grouping Analysis 

To conduct the statistical analysis, we first calculated tree density for each of the census tracts by using a spatial join to determine the number of trees per census tract and then dividing the number of trees by census tract area using the “Zonal Geometry” spatial analyst tool.

We then performed a linear regression using the Ordinary Least Squares tool. The independent variable for the regression was tree density and the dependent variables were population density, income and property value. The next step was to calculate the spatial autocorrelation using the Moran’s I test. Based on the results of the regression analysis and the Moran’s I test, we conducted a grouping analysis of the average income, tree density, and population density. The results of the grouping analysis allowed us to identify areas with low income and low tree density.

MCE Analysis 

Prior to conducting the MCE analysis, we created buffers around three of the criteria: greenways (50 m), health care facilities (100 m) and cultural spaces (100 m). We used the polygons of low-income, low-tree density areas obtained from the grouping analysis. We converted these buffers/polygons into rasters, then reclassified them for each criteria with values of 1 for cells that fit the criteria, and values of 0 elsewhere. We then created a new raster with raster calculator, which added up all the values, and resulted in a map which showed which areas met our criteria and were suitable for planting additional street trees.