A Report on the City of Pittsburgh's Existing and Possible Tree Canopy

DRAFT A Report on the City of Pittsburgh’s Existing and Possible Tree Canopy

Why is Tree Canopy Important? Project Background Tree canopy (TC) is the layer of leaves, branches, and stems of trees that The goal of the project was to apply the USDA Forest Service’s cover the ground when viewed from above. Tree canopy provides many Tree Canopy (TC) Assessment Protocols to the City of Pitts- benefits to communities, improving water quality, saving energy, lowering burgh. The analysis was conducted based on 2010 data. This city temperatures, reducing air pollution, enhancing property values, project was made possible by funding from Tree Pittsburgh. providing wildlife habitat, facilitating social and educational opportunities, This analysis of Pittsburgh’s tree canopy (TC) was conducted in and providing aesthetic benefits. Establishing a tree canopy goal is crucial collaboration with the Tree Pittsburgh, Allegheny County, and for communities seeking to improve their green infrastructure. A tree can- the USDA Forest Service’s Northern Research Station. The Spa- opy assessment is the first step in this goal-setting process, providing esti- tial Analysis Laboratory (SAL) at the University of Vermont’s mates for the amount of tree canopy currently present in a city as well as Rubenstein School of the Environment and Natural Resources the amount of tree canopy that could theoretically be established. carried out the assessment.

How Much Tree Canopy Does Pittsburgh Have? An analysis of Pittsburgh’s tree canopy (TC) based on land cover data derived from high-resolution aerial imagery and LiDAR (Figure 1) found that 14,883 acres of the city were covered by tree canopy (termed Existing TC), representing 42% of all land in the city. An additional 33% (11,914 acres) of the city could theoretically be modified (termed Possible TC) to accom- modate tree canopy (Figure 2). In the Possible TC category, 9% (3,436 acres) of the city was classified as Impervious Possible TC and another 24% was Vegetated Possible TC (8,406 acres). Vegetated Possible TC, or grass and shrubs, is more conducive to establishing new tree canopy, but establishing tree canopy on areas classified as Impervious Possible TC will have a greater impact on water quality and summer temperatures.

Figure 2: TC metrics for Pittsburgh based on % of land area covered by each TC type. Key Terms TC: Tree canopy (TC) is the layer of leaves, branches, and stems of trees that cover the ground when viewed from above. Land Cover: Physical features on the earth mapped from aerial or satellite imagery, such as trees, grass, water, and impervious surfaces. Existing TC: The amount of urban tree canopy present when viewed from above using aerial or satellite imagery. Impervious Possible TC: Asphalt or concrete surfaces, excluding roads and buildings, that are theoretically available for the establishment of tree canopy. Vegetated Possible TC: Grass or shrub area that is theoretically Figure 1: Land cover derived from high-resolution aerial imagery for the City available for the establishment of tree canopy. of Pittsburgh.

09/08/11 1 DRAFT Mapping Pittsburgh’s Trees Parcel Summary Prior to this study, the only comprehensive remotely-sensed esti- After land cover was mapped city-wide, Tree Canopy (TC) metrics mates of tree canopy for Pittsburgh was from the 2001 National were summarized for each property in the city’s parcel database Land Cover Database (NLCD 2001). While NLCD 2001 is valuable for (Figure 4). Existing TC and Possible TC metrics were calculated for analyzing land cover at the regional level, it is derived from relative- each parcel, both in terms of total area and as a percentage of the ly coarse, 30-meter resolution satellite imagery (Figure 3a). Using land area within each parcel (TC area ÷ land area of the parcel). high-resolution aerial imagery and LiDAR acquired in 2010 and 2006 respectively (Figure 3b), in combination with advanced auto- mated processing techniques, land cover for the city was mapped Parcels in much greater detail (Figure 3c). NLCD 2001 substantially under estimated Pittsburgh’s tree canopy, at 24%, largely because it failed to capture smaller patches.

a. NLCD 2001 Percent Tree Canopy (30 meter)

Existing Tree Canopy (TC)

b. 2008 Aerial Imagery (1 meter)

Possible Tree Canopy (TC)

c. Land Cover Derived from 2010 Aerial Imagery

Tree Canopy Grass/Shrub Bare Soil Water Buildings Roads/Railroads Other Paved

Figure 3a, 3b, 3c: Comparison of NLCD 2001 to high-resolution land Figure 4a, 4b, 4c: Parcel-based TC metrics. TC metrics are generat- cover. ed at the parcel level, allowing each property to be evaluated ac- cording to its Existing TC and Possible TC.

09/08/11 2 DRAFT Land Use Allegheny County maintains a comprehensive land use layer for the region. Existing and Possible Tree Canopy (TC) were summarized for the twelve aggregated land use classes (Figure 5, Table 1). For each land use class, Tree Canopy (TC) metrics were calculated as a percentage of all land in the city (% Land), as a percentage of land area in the specified land use category (% Category), and as a percentage of the area for a given TC type (% TC Type). Residential land is the largest land use type within the city. The fact that 43% (% Category) of residential land is covered by tree canopy results in residential land being the largest contributor to the city’s overall tree canopy (35%). With 98% of its land covered by tree canopy, agricultural land it is the clear leader in the % Category field for Existing TC, but the amount of agricultural land in the city is so small that the overall contribution is insignificant (% TC Type). Government land use also had a high percentage of land covered by Existing TC (65%). Residential land also accounts for the largest single amount of Possible TC in the city, at 38% (% TC Type), but industrial land and areas where the land use is either not available (N/A) or classified as other all have nearly 50% of their land area available for tree canopy (% Category). The challenge in establishing tree canopy on industrial land is that much of the available space for planting trees falls into the Possible TC Impervious category (paved surfaces that are not buildings or roads). Establishing new tree canopy on lands considered to be residential or other should theoretically be easier as they have relatively high percentages of land in the Possible TC Vegetation category.

Figure 5: Tree Canopy (TC) metrics summarized by land use.

Existing TC Possible TC Vegetation Possible TC Impervious Land Use % Land % Category % TC Type % Land % Category % TC Type % Land % Category % TC Type Agricultural 0% 98% 0% 0% 2% 0% 0% 0% 0% Commercial 6% 36% 14% 3% 19% 13% 3% 21% 35% Government 11% 65% 26% 3% 19% 13% 1% 8% 13% Industrial 0% 17% 1% 0% 16% 2% 1% 33% 9% N/A 0% 33% 0% 0% 22% 1% 0% 26% 2% Other 2% 41% 6% 2% 37% 9% 1% 11% 6% Residential 15% 43% 35% 11% 32% 47% 2% 5% 17% Rights of Way 6% 31% 15% 3% 14% 12% 1% 6% 12% Utilities 1% 37% 3% 1% 23% 3% 1% 17% 5%

Area of TC type for zoning district Area of TC type for zoning district Area of TC type for zoning district % Land = % Category = % TC Type = Area of all land Area of all land for specified land use Area of all TC type

The % Land Area value of 15% indicates that 15% of Pitts- The % Land value of 43% indicates that 43% of land in the The % TC Type value of 35% indicates that 35% of all tree burgh’s land area is covered by tree canopy in the Residen- Residential land use class is covered by tree canopy. canopy is in Residential land use. tial land use class.

Table 1: Tree Canopy (TC) metrics were summarized by land use. For each land use class, TC metrics were computed as a percentage of all land in the city (% Land), as a percentage of land in the specified land use class (% Category), and as a percentage of the area for a given type (% TC Type).

09/08/11 3 DRAFT Parks Analysis The Existing and Possible Tree Canopy were analyzed for all parks within the Pittsburgh boundary. There are many outliers on both ends of the spectrum with some parks in forested areas having 100% tree canopy and others having no tree canopy. The parks with 0% tree canopy tend to be less than an acre and are playgrounds or sports fields. As a result, it is not socially desirable to plant trees on these parks regardless of the Possible TC percentages. Pittsburgh’s four largest parks in the area range from 63% to 90% Existing Tree Canopy. The same four largest parks ranged from 8% to 32% Possible Tree Canopy.

Existing Tree Canopy Possible Tree Canopy

Figure 6 Existing TC (left) and Possible TC (right) as a percentage of land area by park. Decision Support Parcel-based Tree Canopy (TC) metrics were integrated into the city’s existing GIS database (Figure 7). Decision makers can use GIS to query specific tree canopy and land cover metrics for a parcel or set of parcels. This information can be used to estimate the amount of tree loss in a planned development or set tree canopy improvement goals for an individual property.

GIS Database

Figure 7: GIS analysis of parcel-based TC metrics for decision support. In this example, GIS is used to select an individual parcel. The attributes for that parcel, including the parcel-based TC and land cover metrics, are displayed in tabular form providing instant access to relevant information.

09/08/11 4 DRAFT Neighborhood Analysis Despite Pittsburgh’s relatively high Existing Tree Canopy overall, the tree canopy is not evenly distributed. More than half of the city’s neighborhoods having less than mean Existing TC for the city (42%) and nearly 60% of the total population lives in these neighborhoods. The neighborhoods of Hays and Glen Hazel have the highest amount Existing Tree Canopy at 81% and 80% respectively. Chateau and North Shore have the least amount of Existing Tree Canopy at 5% and 6% respectively. Chateau and Terrace Village have the largest amount of Possible Tree Canopy at 47%. Glen Hazel and Hays also have the least amount of Possible Tree Canopy at 12% and 13% respectively. Existing Tree Canopy Possible Tree Canopy

Figure 8. Existing TC (left) and Possible TC (right) as a percentage by neighborhood. Population Density Poverty Rate The neighborhoods of Chateau and South Shore have the lowest The neighborhoods of Bluff and Chateau have the highest poverty population densities at 29 and 87 people per square mile respec- rates at 80% and 100% respectively, while North and South Shore tively. Chateau also has the least amount of Existing Tree Canopy at have the lowest poverty rates at 0%. It should be noted that 5% and the 2nd most Possible Tree Canopy at 47%. The neighbor- North and South Shore are ranked 2nd and 11th respectively for hoods of Central Oakland and North Oakland have the highest pop- Existing Tree Canopy. The areas where the poverty rates are high ulation densities at 21,742 and 21,150 people per square mile re- exhibit low amounts of tree canopy and high amounts of possible spectively. The percent area of Existing Tree Canopy in Central Oak- tree canopy. Bluff and Chateau have an existing tree canopy of 5% land is 20% and the percent area of Possible Tree Canopy is 33%. and 12% and a possible tree canopy of 35% and 47% respectively. There is a divide between the least densely populated neighbor- The relationship between tree canopy and poverty is not con- hoods, with some having a large amount of Existing Tree Canopy sistent. Central Oakland and Arlington have similar poverty rates, and others having a very small amount of Existing Tree Canopy. It is but vastly different percentages of Existing TC. worth noting that there is no population data for the Mt. Oliver borough.

Figure 9. Tree Canopy per capita in Pittsburgh neighborhoods. Areas Figure 10. Tree Canopy compared to crime per capita in Pittsburgh in black denote no data. neighborhoods. Areas in black denote no data.

09/08/11 5 DRAFT Neighborhood Analysis Continued Vacancy Crime Per Capita

The vacancy rate in Pittsburgh seems to relate to some extent to the There appears to be a strong relationship between crime per capita amount of tree canopy in an area. The Strip District has a vacancy and the percent Existing Tree Canopy. There are strong contrasts rate of 14.8% and only 11% Existing Tree Canopy. St. Clair has a between neighborhoods that are close geographically. Highland vacancy rate of 50% and 69% Existing Tree Canopy. Vacancies, par- Park has a crime rate per capita of 3, is 66% white, has a poverty ticularly if they result in abandoned lots, can result in increased tree rate of 9%, and 49% Existing Tree Canopy. Directly to its south, canopy. Unless this land is cared for properly, the combination of Larimer has a per capita crime rate that is four times higher, is 9% vacant properties and vegetation can result in unsightly conditions. white, has 22% of the population living at or below poverty, and has less than half as much Existing Tree Canopy at 22%.

Figure 11. Neighborhood tree canopy and percent vacancy rates. Figure 12. Neighborhood tree canopy and crime per capita. Percent White Percent Under 18 Neighborhoods with higher percentages of white residents are large- As tree canopy can help improve air quality and the urban heat is- ly located in the southern area of the city. South Shore and Lincoln land effect, it is desirable to have tree canopy in areas with vulnera- Place have the greatest percentage of white residents with 100% ble populations such as children. Northview Heights and Homewood and 97% respectively, while Homewood South, East Hills, and Home- South have higher than average populations under the age of 18, wood North have the lowest percentage of white residents with 2%, but have very different percentages of Existing Tree Canopy. 2.1%, and 2.1% respectively. There is no apparent relationship between the percentage of white residents and the percent of existing or Possible Tree Canopy by neighborhood.

Figure 13. Neighborhood tree canopy and percent population under 18. Figure 14. Neighborhood tree canopy and percent of the population that is white.

09/08/11 6 DRAFT River Buffer Tree Canopy (TC) metrics were computed for a 100ft buffer surrounding all rivers. Within the river buffer zone 42% of the land is tree canopy and 41% of the land is available for the establishment of tree canopy.

Figure 15: TC metrics summarized by a 100ft buffer around all rivers in Pittsburgh. Sewer Authority Tree Canopy (TC) metrics were computed for the sewer authority area within the city. This can be considered to be the urbanized area and thus the Existing Tree Canopy, at 37% is less than the mean for the city as a whole.

Figure 16: TC metrics summarized by the sewer authority within the city. Building Buffer Tree Canopy (TC) metrics were summarized within a 25ft buffer around all buildings. Existing TC covered 18% of the buffer while Possible Tree Canopy covered 39% of the buffer. This indicates that there is room for additional trees to help reduce energy costs.

Figure 17: TC metrics summarized by a 25ft buffer around buildings. 09/08/11 7 DRAFT Subwatershed Analysis Tree canopy metrics were computed for the portions of the 12 watersheds within the city boundary. The Streets Run watershed and Shades Run watershed have the highest percentage of their land area covered by tree canopy at 69% and 64% respectively. The Jacks Run and Allegheny River watersheds have the lowest percentage of its land area covered by tree canopy at 36% each. However, the Thompson Run watershed has the highest possible percentage of tree canopy at 44%. The Streets Run watershed has the lowest Possible Tree Canopy at 23%. In general, watersheds with a lower percentage of tree canopy are clustered in the downtown area, while those with a higher percentage of tree canopy are on the outskirts of the city. Research in the Mid-Atlantic states indicate that tree canopy percentages of 45% in a watershed are associated with “good” stream health. Of the six major watersheds that intersect the city, three, the Allegheny River, Ohio River, and Monongahela River, all have tree canopy percentages clearly below the recommended 45%. Goetz, S. J., R. K. Wright, A. J. Smith, E. Zinecker, and E. Schaub. 2003. IKONOS imagery for resource management: Tree cover, impervious surfaces, and riparian buffer analyses in the mid-Atlantic region. Remote Sensing of Environment 88, no. 1: 195-208.

Existing Tree Canopy Possible Tree Canopy

South Figure 18: Existing TC (left) and Possible TC (right) as a percentage of area for each subwatershed.

Figure 19. Graphical representation of the tree canopy metrics for the ten largest subwatersheds by land area.

09/08/11 8 DRAFT Tree Canopy Opportunity Index In addition to simple descriptive statistics, more sophisticated techniques can help identify areas of the city where tree-planting and stewardship programs would be most effective. One approach is to focus on spatial clusters of Existing and Possible TC. When a 250-foot grid network is superimposed on the city’s land-cover map, it is possible to map regions of the city where high values of Existing TC are tightly clustered (Figure 20a). A similar map was constructed for Possible TC (Figure 20b). A single index was creat- ed by subtracting the percentage of Existing TC per grid cell from Possible TC, which produced a range of values from –1 to 1. When clustered, this tree canopy opportunity (TCO) index highlights areas with high Possible TC and low Existing TC (Figure 20c); these areas theoretically offer the city the best places to strategically expand the city’s tree canopy and increasing itsmany attendant benefits. Different geographies can be overlaid on the TCO index such as neighborhoods (Figure 20d) to target specific regions for establishing tree canopy.

(a) (b)

Figure 20: (a) Spatial clustering of Existing TC in Pittsburgh; dark green areas are highly clustered and have high Existing TC values; (b) Spatial clustering of Possible TC in Pittsburgh; dark red areas are highly clustered and have high Possible TC values.; (c) Spatial clustering of a combined index of Existing and Possible TC; red areas theoretically provide the best opportunities for expanding tree canopy and; (d) neighborhoods where overlaid the combined index to target potential neighborhoods for expanding tree canopy in the city.

09/08/11 9 DRAFT Conclusions  Pittsburgh’s urban tree canopy is a vital city asset that reduces  With Existing and Possible TC summarized at the parcel level stormwater runoff, improves air quality, reduces the city’s car- and integrated into the city’s GIS database, individual parcels bon footprint, enhances quality of life, contributes to savings on can be examined and targeted for tree canopy improvement. Of energy bills, and serves as habitat for wildlife. particular focus should be parcels in the city that have large,  At 42% of the land area, Pittsburgh’s overall tree canopy is a contiguous impervious surfaces. These parcels contribute high higher percentage than similar cities, but majority of the city’s amounts of runoff, which degrades water quality. The establish- residents live in neighborhoods with substantially less tree cano- ment of tree canopy on these parcels will help reduce runoff py. Some neighborhoods have as little as 5% of their land area during periods of peak overland flow. Vacant lands should also covered by tree canopy. be a focus as they present a unique opportunity.  The tree canopy metrics, in combination with neighborhood  Pittsburgh’s residents are the largest steward of the city’s tree indicators such as poverty and crime, can be used to prioritize canopy and have most of the land to plant tees. Programs that neighborhoods for tree planting initiatives. educate residents on tree stewardship and provide incentives  Although this assessment indicates that half of the land in Pitts- for tree planting are crucial if Pittsburgh is going to sustain its burgh could theoretically support tree canopy, planting new tree canopy in the long term. trees on much of this land may not be socially desirable (e.g.  Consideration should be given for establishing tree canopy goals recreation fields) or financially feasible (e.g. parking lots). for individual watersheds. Research by Goetz et al. (2003) indi- Setting a realistic goal requires a detailed feasibility assessment cates that watersheds with 37% tree canopy can be categorized using the geospatial datasets generated as part of this assess- as “fair” in a stream health rating; watersheds with 45% tree ment. canopy can be categorized as “good.”

Figure 21: Comparison of Existing and Possible Tree Canopy with other selected cities that have completed Tree Canopy Assessments. Prepared by: Additional Information Jarlath O’Neil-Dunne Funding for the project was provided by Tree University of Vermont Pittsburgh. More information on Urban Tree Spatial Analysis Laboratory Canopy Assessments can be found at the fol- [email protected] lowing web site: 802.656.3324 http://nrs.fs.fed.us/urban/utc/

Spatial Analysis Lab Tree Canopy Assessment Team: Brian Beck, Terence Bennett, Lauren Demars, Tayler Engel, Samantha Gollub, Ray Gomez, Michael Grobicki, Daniel Hedges, Donald Hefferon, Lindsay Jordan, Dan Koopman, Sean MacFaden, Jarlath O’Neil-Dunne, Keith Pelletier, Eleanor Regan, Anna Royar, Sam Shaefer-Joel, Bronson Shonk, Bobby Sudekum 09/08/11 10