Tree canopy in a city is not decoration. It cools streets, filters air and tracks who gets those benefits, which makes its distribution an equity question rather than a landscaping one. The usual assumption is that canopy follows race. So seven American cities were analyzed with high-resolution land cover data against census data, using correlations and two kinds of regression. What held up?
Income held up and race largely did not. Across all seven cities there is a strong positive correlation between cover and median household income. Negative correlations between race and canopy exist in bivariate models for some cities, and they are generally not observed in that also include income, education and housing age. A suite of variables contributes, income among them.
Urban tree canopy delivers cooling, air filtering and stormwater benefits, so how it is distributed within a city is a question of distributional equity. This study examined it for Baltimore, Los Angeles, New York, Philadelphia, Raleigh, Sacramento and Washington, using high spatial resolution land cover data together with census data.
The analysis worked at the level, which is fine enough to see within-city variation, and used three approaches: Spearman correlation, ordinary least squares regression, and a . Using more than one is not redundancy here, since block groups near each other resemble each other and ordinary regression assumes they do not.
Across all cities there is a strong positive correlation between canopy cover and median household income. Negative correlations between race and canopy cover exist in bivariate models for some cities, but they are generally not observed in multivariate regressions that include income, education and housing age.
The spatial models produced higher r-square values than the ordinary least squares models across all cities, confirming that spatial autocorrelation matters. Similarities between cities emerged along shared characteristics of climate, race and ethnicity, and size. The authors conclude that a suite of variables including income contributes to canopy distribution, and that the findings can help target strategies for canopy goals and environmental justice at once. Read the correlations as correlations: nothing here establishes which came first.
wat. City tree cover tracked household income across all seven cities, while the apparent link to race mostly vanished once income and housing age were in the model.
wat wat
Which variable survives the controls decides where you plant. Targeting low-income block groups and targeting by race are different programs, and the multivariate result points at the first.
How do we know?
Schwarz and colleagues analyzed urban tree canopy cover against census data at the Census Block Group level for seven US cities, using Spearman correlation, ordinary least squares regression and a spatial autoregressive model.
What is still unknown
Cross-sectional correlation across block groups, which cannot say whether canopy followed income or income followed canopy. Income and race are entangled in American cities, so a disappearing coefficient is a statement about this model rather than about causation.
Learn something?
Where to next
How much greenery in a neighborhood is linked to less mental distress?city
Why do leaves under a streetlight stay more intact?trees
Half an hour in a park. Does it really do something to the body’s numbers?city