This post is part of a series of articles written by the 2020 Summer of Maps Fellows. Azavea’s Summer of Maps Fellowship Program is run by the Data Analytics team and provides impactful Geospatial Data Analysis Services Grants for nonprofits and mentoring expertise to fellows. To see more blog posts about Summer of Maps, click here.
During the hottest months of the summer, you’ll find Philadelphia residents doing just about everything to keep cool – cramming inflatable pools into even the smallest backyards, opening fire hydrants, and traveling halfway across the city just to get to a park with some shade.
All of these are temporary ways to keep cool, but what about solutions for the long term, like having shade everywhere throughout your neighborhood? Why should some residents have to travel halfway across the city just to cool down?
Ideally, all Philadelphia neighborhoods will have at least 30% tree canopy coverage in the future, but the current reality is that the tree canopy is unequally distributed across the city, and the neighborhoods with the most tree canopy are typically those with wealthier residents.
This means that large parks such as Fairmount and Wissahickon contain acres of dense forest, which provide an abundance of shade and essential ecosystem services, whereas neighborhoods such as Hunting Park in North Philadelphia have very few trees at all.
Philadelphia’s Beat the Heat Plan recognizes that Black, Hispanic, and other residents of color are more likely to live in neighborhoods with sparse tree canopy as well as heat-vulnerable neighborhoods, or neighborhoods where residents are more susceptible to illnesses that are related to hot temperatures. This pattern can be traced back to decades of redlining, when the federal government actively dissuaded lenders from investing or giving home loans to neighborhoods that have more immigrant residents or residents of color.
The effects of redlining and past disinvestment within Hunting Park can be seen through the various physical characteristics of the neighborhood, such as sparse tree canopy, an aging housing stock, and a high percentage of impervious surfaces. All of these characteristics directly contribute to heat vulnerability, making Hunting Park one of the more vulnerable neighborhoods in the city.
Philadelphia’s Greenworks Plan aspires to attain a 30% tree canopy for every neighborhood to reduce heat vulnerability and inequity within the city, but to meet this goal, first focusing on increasing tree canopy in historically underserved neighborhoods is essential. Hunting Park’s adoption of the city’s first-ever neighborhood forestry plan is a huge step toward focusing on greening in an equitable way.
The current structural makeup of Hunting Park—sparse tree canopy, a high percentage of impervious surfaces, and an aging housing stock that is unequipped to effectively cool down during summer months—combines the perfect conditions for a heating effect called the urban heat island.
As the figure above illustrates, urban heat island occurs in cities rather than in rural areas. Cities tend to have far more impervious surfaces such as roads, sidewalks, and buildings than suburban and rural areas; this in turn inhibits soil, grass, and other vegetation from effectively absorbing heat. Trees play a significant role in reducing the urban heat island effect, because tree leaves absorb heat and sunlight, which in turn creates shading from the sun.
As a result, neighborhoods with denser tree canopy tend to have cooler temperatures in the summertime, while neighborhoods with sparse tree canopy are more prone to dangerously high temperatures. In neighborhoods with an aging housing stock, many homes lack air conditioning or easily accessible cooling methods, which can lead to higher rates of heat-related illnesses and hospitalizations.
The adoption of Philadelphia’s Urban Forest Strategic Plan and Hunting Park’s Forestry Plan presents a unique opportunity to address the need for equitable tree planting and the reduction of heat vulnerability, as these two plans are the city’s first cohesive attempts to prioritize and manage its urban forest.
In collaboration with Esperanza (a non-profit serving the residents of Hunting Park), the Pennsylvania Horticultural Society, Philadelphia Parks & Recreation, the Philadelphia Water Department, and the City of Philadelphia Office of Sustainability, my project seeks to accomplish the following:
- Analyze the distribution of Hunting Park’s current tree canopy.
- Find suitable sites for additional tree plantings on Hunting Park’s residential, commercial, and industrial land.
- Determine which Hunting Park census block groups and streets should be prioritized for additional tree plantings.
The results of this summer’s project aim to inform Hunting Park’s Forestry Plan, but they may also serve as a model for Philadelphia’s future neighborhood-level forestry projects.
In this blog, I present some of my project’s key findings.
The first phase of my project examined the distribution of tree canopy in Hunting Park using the University of Vermont 2018 tree canopy dataset. This dataset examines change in tree canopy from 2008 to 2018 and classifies this change into 3 categories: loss (removed trees), gain (new tree plantings or tree growth), and no change.
I found that Hunting Park has only an 8.6% tree canopy cover, with most of its tree canopy being clustered in the actual park rather than in the surrounding residential, commercial, and industrial street blocks.
The map above depicts the neighborhood’s tree canopy change patterns between the years 2008 and 2018. The majority of tree canopy in the neighborhood did not change significantly during these 10 years, but 1.3% was lost overall during this period.
Given that the city aspires to attain 30% tree canopy coverage for Hunting Park, tree planting efforts need to drastically increase. Additionally, existing tree canopy needs to be well-maintained so that in the future, gained tree canopy outweighs lost tree canopy.
Want to work on projects with a social and civic impact? Learn what it’s like to work at Azavea.
The goal of the second phase of the project was to determine an equitable, prioritization plan for future planting
How does one go about prioritizing parts of the neighborhood for tree planting when the entire neighborhood needs more trees? Many different factors can go into a tree planting prioritization model, which makes determining prioritization methodology a bit daunting.
Based on existing projects and known patterns within the neighborhood, we determined a list of key criteria to represent the neighborhood’s needs. These factors include average surface temperature (which directly represents the urban heat island effect), tree canopy, impervious surface area, and socioeconomic variables such as income.
Since numerous studies found that denser tree canopy is correlated to lower crime rates, property and violent crime were added as criteria inputs. This means that areas with higher crime rates could benefit from increased tree canopy, so areas with higher crime rates will be prioritized for additional tree plantings over areas with lower crime rates.
Since trees remove pollutants, air quality and related public health indicators such as traffic volume and the percentage of residents with asthma were also included. Additionally, since more green space and additional tree canopy can contribute to higher rates of exercise among residents, variables such as the percentage of residents with diabetes and obesity were also included.
The possibilities are endless, and the variables that the project partners and I found useful may not be the same as a citywide analysis or other neighborhood analyses.
After reviewing existing literature on past tree planting prioritization projects in three U.S. cities, the project team decided to take a similar approach to that of The Tree Prioritization Plan Developed for New York City: different clusters or groupings of criteria were identified based on varying themes of chosen factors, and each factor within these clusters was assigned a weight according to its perceived importance to the project team.
|Chosen Variable||Data Source||Variable Weight||Cluster|
|Air Toxics Respiratory Hazard Index||EPA EJ Characteristics (2018)||0.5||Air Quality & Noise Pollution (0.16)|
|Traffic Volume & Proximity Index||EPA EJ Characteristics (2018)||0.5||Air Quality & Noise Pollution (0.16)|
|% Impervious Surface||Land Cover Raster (2018)||1||Water (0.16)|
|Max. Average Surface Temperature||Office of Sustainability/NOAA Temperature Raster||1||Urban Heat Island (0.16)|
|% Low-income Population||EPA EJ Characteristics (2018)||0.25||Socioeconomic (0.16)|
|% Linguistically Isolated Population||EPA EJ Characteristics (2018)||0.25||Socioeconomic (0.16)|
|% Homeowner Population||ACS 5-Year Estimates (2018)||0.25||Socioeconomic (0.16)|
|% Crime per Population||PPD Crime Incidents (2010-2020)||0.25||Socioeconomic (0.16)|
|% Tree Canopy||2008 – 2018 Tree Canopy Change||1||Tree Canopy Coverage (0.16)|
|% Population with Asthma||CDC 500 Cities (2019)||0.33||Public Health (0.16)|
|% Obese Population||CDC 500 Cities (2019)||0.33||Public Health (0.16)|
|% Diabetic Population||CDC 500 Cities (2019)||0.33||Public Health (0.16)|
Six different clusters were chosen based on available data as well as the themes and perceived importance of the variables. Equal weights were ultimately chosen for all six clusters, though it should be noted that clusters with fewer variables, such as the Urban Heat Island with only the average surface temperature measurements going into it, will naturally be weighted a bit higher.
Part of what makes the Hunting Park tree planting prioritization effort unique is that it examines potential urban canopy at the neighborhood level rather than at the citywide level. Given that we’re determining prioritization for a neighborhood, certain variables that are more unique to the neighborhood can more easily be incorporated into the criteria.
For example, Hunting Park’s Beat the Heat plan recognizes that primarily Spanish-speaking households are more likely to be unaware of the tree giveaway programs that serve the neighborhood. This distinction provides an opportunity to incorporate more characteristic socioeconomic variables, such as the percent of linguistically isolated households, into the criteria.
While conducting a neighborhood-level analysis allows for more detailed results than what would be offered at the citywide scale, some of the data simply does not exist at a more precise level than what would be used for a citywide analysis. Much of the socioeconomic and public health data is only available at a larger unit of analysis such as the census block group or tract, which isn’t always helpful at guiding tree plantings in on streets or in backyards.
With a roughly 1,100-acre neighborhood and only 29 census block groups (some much larger than others), how do we more closely examine smaller areas within the census block groups?
A possible solution to this problem is to calculate prioritization at the street centerline segment level. Street centerline segments are any streets that are maintained by the city government rather than private property owners.
Since the socioeconomic and public health data is not available at this street centerline segment scale, I had to join all the data that is only available at the census block group or census tract level back to the street centerline segments that are within the census block groups or tracts.
What’s most useful about the street centerline segment unit is that more specific physical characteristics, such as impervious surface percentage or mean surface temperature, can be analyzed at a more precise scale than what would be included in the census block group prioritization analysis.
For a more in-depth description of my methodology for the street centerline segment analysis, check out my lightning talk here starting at 1:15.
The end result includes two different outputs—one is a composite prioritization map at the census block group level, and the other is a composite prioritization map at the street centerline segment level. Both are based on the same prioritization criteria with the same weighting for variables.
The two maps above highlight the composite prioritization scores at both geographic levels. Darker census block groups or lighter street segments indicate a higher prioritization score, which means that geographic unit is most in need of additional tree plantings.
By comparing the two maps, the results for both geographic units are relatively similar. For both, the areas with higher prioritization scores are mostly located in the residential blocks to the east and west of the park.
If you look closely, though, you can see some significant differences between the two maps. Some streets have a much higher prioritization score than what the census block group analysis alludes to, such as the streets in the northeastern part of Hunting Park.
This could be due to a combination of higher surface temperatures, more impervious surfaces, fewer trees, or more crimes around specific streets. Since the street segment prioritization analysis highlights these physical differences at a more precise level than the census block group analysis, it will likely be more useful when guiding in-depth tree planting strategies.
Hunting Park needs to reach 30% overall tree canopy to align with the Greenworks goals, which means tree planting efforts need to be well-informed and planned. The results of my analysis are meant to serve as a guide for future equitable tree planting strategies both within the neighborhood and throughout the city.
All residents deserve to experience the same benefits of trees within their own neighborhood, which is why it’s so important for cities such as Philadelphia to focus on increasing tree canopy in the neighborhoods that have historically been ignored. Hopefully, projects such as this one can inspire other neighborhoods to focus on equitable tree planting strategies so that the city can attain a denser tree canopy that will benefit all residents for years to come.