Satisfaction with urban trees associates with tree canopy cover and tree visibility around the home

Urban trees offer significant environmental, economic, and social benefits, making their expansion a priority for many cities. However, successful urban tree initiatives require understanding public perception and experiences. This study investigates the relationship between community satisfaction with urban trees and their management, and objective measures of tree abundance and visibility at various spatial scales. While increased tree cover intuitively suggests greater satisfaction, this hasn't been rigorously examined across multiple scales of community experience. Previous research has shown a weak association between tree presence and satisfaction, often employing city-wide averages. It remains unclear if similar associations exist at finer scales, such as the neighborhood level, and how these relate to various methodologies for measuring tree abundance and visibility, each with different spatial scales. Traditional measures like NDVI and canopy cover provide a "bird's-eye view" of greenness, potentially failing to reflect what people actually see at eye-level. Newer measures like VGVI offer an "eye-level" perspective, representing actual visibility more accurately. The research addresses three questions: (1) Is there an association between resident satisfaction with urban trees and greenness measures? (2) Is there an association between resident satisfaction with urban tree management and greenness measures? (3) Does the strength of these associations vary with neighbourhood size?

Numerous studies emphasize the multifaceted benefits of urban trees, including ecosystem services such as air pollution regulation and heat mitigation. However, simply increasing tree numbers without considering community perception may yield less-than-optimal results. Public satisfaction, reflecting the alignment between expectations and experiences, is a crucial factor in securing community support for tree planting initiatives. Prior work has shown a weak association between overall satisfaction and average urban tree canopy cover at a city scale. This study aims to address the gap in understanding how people perceive urban tree abundance by using a more granular scale analysis and considering several different measures for calculating greenness. A critical review of prior literature highlights the diversity in public perception of urban forests, with personal identity factors influencing preferences and attitudes. This study controls for cognitive, social-ecological, and demographic influences on these perceptions to better understand how daily experiences with trees shape satisfaction, independent of personal identity factors. The study examines three measures: NDVI (a top-down measure of vegetation abundance), tree canopy cover (a two-dimensional measure of tree abundance from a bird's eye view), and VGVI (an eye-level, three-dimensional measure of tree visibility). The study will utilize a survey of Toronto residents and compare subjective satisfaction with different neighborhood-level measures of greenness.

This research employed a mixed-methods approach, combining survey data and spatial analysis. Ethical approval was obtained from the University of Toronto Ethics Review Board. Data were collected from a demographically and geographically representative online panel survey of 223 residents in Toronto, Canada. The survey, part of a larger research program, used existing validated measures of satisfaction with urban trees and their management (each based on an 8-item, 5-point Likert scale), nature relatedness (NR6 scale), tree knowledge, and various demographic and socio-ecological factors. Postal codes were voluntarily shared by respondents. Three neighborhood-level greenness measures—NDVI, percent tree canopy cover, and VGVI—were calculated using different buffer sizes (100 m, 300 m, and 500 m) around each respondent's postal code. NDVI was derived from Sentinel-2 satellite imagery (June-September 2020), while canopy cover data were obtained from a high-resolution (2 m) land cover dataset from the City of Toronto's "Automated Land Cover Analysis-2018 Tree Canopy Study." VGVI, calculated using high-resolution (2 m) digital elevation data and canopy cover, represented eye-level tree visibility. Data analysis was performed in R v. 4.2.1. Confirmatory factor analysis and reliability measures were used to validate survey scales. Generalized linear models (GLM) with Gaussian error distribution were used to assess the relationship between satisfaction measures (satisfaction with urban trees and management) and greenness measures, while controlling for cognitive, social-ecological, and demographic factors. Individual models were run for each greenness measure at each buffer size, resulting in a total of 18 GLMs. The assumption of normality was checked using VIFs and residual plots.

The survey responses (n=223) were geographically distributed across Toronto, representing diverse greenness conditions. On average, respondents were somewhat satisfied with their trees (M=3.68, SD=0.72) and slightly less satisfied with tree management (M=3.23, SD=0.84). GLM analyses revealed a positive association between satisfaction with urban trees and both canopy cover and VGVI, after controlling for cognitive, social-ecological, and demographic factors. The association with VGVI was consistently stronger than with canopy cover. Notably, there was no significant association with NDVI or satisfaction with tree management. The strength of the associations between satisfaction and both canopy cover and VGVI increased with neighborhood size (100m, 300m, 500m buffer zones), strongest at 300 and 500 m (Table 1). Visual inspection of scatter plots did not reveal clear patterns, emphasizing the importance of statistical analysis in identifying these relationships. Table 1 shows detailed results of GLM analysis including estimates, confidence intervals, p-values, and model fit statistics. Figure 2 displays scatter plots and linear trend lines with 95% confidence intervals showcasing associations between satisfaction with urban trees and the three greenness measures at different buffer sizes.

The findings highlight the importance of considering both the abundance and visibility of urban trees when assessing community satisfaction. The weak relationship with NDVI, which includes all vegetation, suggests that simply increasing overall greenness isn't sufficient for enhancing satisfaction with trees. The stronger association with VGVI compared to canopy cover underscores the significance of eye-level visibility in shaping people's experience and perception of urban trees. The increased strength of associations at larger spatial scales likely reflects the cumulative effect of more trees within a larger area and the increased likelihood of daily interaction with them during normal activities. The lack of association between greenness and satisfaction with tree management may be attributed to more abstract and established perceptions related to local governance and decision-making, rather than direct daily experiences with trees. Future research might benefit from exploring cognitive, social-ecological, and demographic factors that contribute to satisfaction with tree management.

This study demonstrates a positive association between community satisfaction with urban trees and both tree canopy cover and, more strongly, eye-level tree visibility (VGVI). This underscores the need to consider diverse measures of greenness and multiple spatial scales when planning and managing urban trees. Further research should explore the complex interplay of factors that shape satisfaction with tree management and investigate these relationships in different urban contexts with larger datasets.

This study's sample size (n=223) is relatively small, limiting generalizability. Data availability was constrained to the City of Toronto, which prevents assessment of these relationships in diverse urban settings. While various cognitive, social-ecological, and demographic factors were accounted for, other neighborhood characteristics, interactive effects, and levels of urbanity could influence these associations and could not be included due to limitations in the analysis methodology. The use of linear models assumes a linear relationship. Exploring non-linear relationships or a wider range of multivariate data might provide further insights.