Descriptive Title:

Pedestrian Environmental Quality Index Approximation Score

Geographic Unit of Analysis:

Street segment

Pedestrian Environmental Quality Index Approximation Score (2015)
NeighborhoodTotal Road MilesMean of PEQI Approximation ScoreMinimum of PEQI Approximation ScoreMaximum of PEQI Approximation ScorePercent of Streets with PEQI Approximation Score
Bayview/Hunter's Point 96 67 37 86 49%
Bernal Heights 45 66 37 86 57%
Castro/Upper Market 29 69 40 86 64%
Chinatown 12 59 27 83 67%
Excelsior 43 70 41 86 68%
Financial District/South Beach 39 57 32 77 55%
Glen Park 19 63 30 89 64%
Golden Gate Park 30 64 20 83 37%
Haight Ashbury 18 70 34 86 67%
Hayes Valley 18 63 32 86 66%
Inner Richmond 22 67 20 86 76%
Inner Sunset 37 68 35 86 58%
Japantown 5 62 37 81 76%
Lakeshore 26 61 35 81 13%
Lincoln Park 4 74 62 76 37%
Lone Mountain/USF 18 66 37 86 71%
Marina 29 70 27 89 67%
McLaren Park 7 70 60 86 28%
Mission 64 66 27 89 62%
Mission Bay 23 61 30 76 33%
Nob Hill 16 60 34 84 73%
Noe Valley 31 70 38 86 65%
North Beach 15 65 36 86 70%
Oceanview/Merced/Ingleside 34 70 41 86 61%
Outer Mission 41 65 40 86 44%
Outer Richmond 45 71 20 86 82%
Pacific Heights 24 69 34 86 57%
Portola 36 68 44 81 39%
Potrero Hill 37 70 36 86 42%
Presidio 45 75 73 76 0%
Presidio Heights 17 69 45 86 48%
Russian Hill 18 67 39 86 58%
San Francisco 1,117 66 20 89 62%
Seacliff 9 75 69 76 17%
South of Market 36 53 22 78 53%
Sunset/Parkside 107 70 35 88 79%
Tenderloin 14 56 31 81 69%
Treasure Island 24 n/a n/a n/a 0%
Twin Peaks 18 63 39 76 36%
Visitacion Valley 20 69 46 86 52%
West of Twin Peaks 91 70 32 89 20%
Western Addition 20 67 37 86 52%

Why Is This An Indicator Of Health and Sustainability?

Environments that support walking, both as an alternative to driving and as a leisure activity, have multiple potential positive health impacts.  Quality, safe pedestrian environments support a decreased risk of motor vehicle collisions and an increase in physical activity and social cohesion with benefits including the prevention of obesity, diabetes, and heart disease as well as stress reduction and mental health improvements that promote individual and community health.  Environments that encourage walking while discouraging driving can further reduce traffic-related noise and air pollution – associated with cardiovascular and respiratory diseases, premature death, and lung function changes especially in children and people with lung diseases such as asthma.a

Interpretation and Geographic Equity Analysis

As with most cities, roadways in San Francisco have historically been optimized for efficient travel by car.  While the historic precedent is evident across the city, downtown neighborhoods and neighborhoods in the eastern half of the city are disproportionately burdened with high volume arterials. The PEQI Approximation provides a method to quantify and demonstrate these patterns.  As evident in the map and table, the neighborhoods of South of Market, Tenderloin, Financial District/South Beach, and Chinatown have the poorest Pedestrian Environments based on the proxy measures used. However, these neighborhoods also have some some of the highest pedestrian activity in the city, due to proximity to job centers, housing density, regional/local transit, and retail. The neighborhoods with the best Approximation scores include Seacliff, Outer Richmond, Sunset/Parkside, Excelsior, Haigh Ashbury, Marina, Noe Valley, Oceanview/Merced/Ingleside, Potrero Hill, and West of Twin Peaks, which all have scores of 70 or above.

Methods

The original PEQI survey was designed to be used by a trained observer  to quantify, describe and summarize street and intersection environmental factors known to affect people's travel behaviors (i.e., whether they walk) and pedestrian safety.   Once collected, data is entered into a database that automatically scores the data.  A PEQI score, reflecting the quality of the physical pedestrian environment, is created for each street segment and intersection in a defined area. 

The PEQI Approximation uses secondary data related to the key factors from the original PEQI survey to estimate street and sidewalk conditions. The table below illustrates the original PEQI measures as they relate to those used in the Approximation Score:

Original PEQI Category Original Weight Proxy Dataset Adjusted Weight Percent of City Streets w/ Proxy Proxy Tier 
Number of Lanes 8.06% Number of Lanes 13.98% 100.00% 1
Posted Speed Limit 7.66% Speed Limit 13.29% 100.00% 1
Traffic Volume 6.05% Traffic Volume 10.49% 97.50% 1
Width of Sidewalk 8.87% Sidewalk Width (ft.) from AECOM's Sidewalk Nexus Study and Urban Watershed Assessment of the
Sewer System Improvement Program
15.38% 83.28% 1
Width of Throughway 8.87% Width of walkable througway from AECOM's Sidewalk Nexus Study 15.38% 66.47% 1
Trees 3.63% Count of Street Trees 6.30% 100.00% 1
Driveway Cuts 6.05% Parking Lots and Garages 10.49% 100.00% 1
Presence of a Buffer 5.24% Protected Bike Lanes and On-Street Parking 9.09% 100.00% 1
Planters or Gardens 1.61% Community Gardens, Plazas, Parks, and Patios 2.80% 100.00% 1
Empty Spaces 1.61% Empty Lots, Vacant Buildings, and Parking Lots 2.80% 100.00% 1
Total PEQI Approx Var. % 57.66%   100.00%    
Street Traffic Calming Features 4.03% Speed Cushions, Speed Humps, Traffic Islands, Chicanes, and Speed Radar Displays 0%* 100.00% 2
Sidewalk Impediments 9.68% All complaints of sidewalk conditions to DPW or 311 (e.g. cracked and lifted sidewalks) from summer 2012 to summer 2014 0%* 26.61% 2
Illegal Graffiti 0.81% Complaints of graffiti to DPW or 311. 0%* 66.96% 2
Litter 4.44% DPW actions: packer trucks for illegal dumping, abandoned waste, bulky items; steam cleanings of any time (e.g. urine, feces, general); street cleaning and general litter pickup; and litter from overflowing cans and can tip overs from summer 2012 to summer 2014 0%* 74.39% 2
Total Subjective Overlay % 27.83%   0%*    
Sidewalk Obstructions 8.87% None 0% 0% 3
Public Seating 1.61% None 0% 0% 3
Public Art or Historic Sites 1.61% None 0% 0% 3
Retail Use or Public Places 4.44% None 0% 0% 3
Street Lighting 6.85% None 0% 0% 2
Total Dropped % 23.38%        

*Overlay subset variables – These variables should be used as an overlay to prioritize segments with low PEQI approximation scores.

 

the weights for the variables not included in the PEQI-Lite score calculation were redistributed to preserve the relative weight between
variables for the approximation score.

The weights for the variables not included in the PEQI Approximation Score calculation were redistributed to preserve the relative weight betweenvariables for the approximation score. The original PEQI surveyed each side of the sidewalk and computed a unique score for each side of the street. The Approximation uses an aggregation of data from both sides of the sidewalk (where data was available) to create a summarized score per street segment and not per side. Google Street View was used to check a sample of streets to determine whether the Approximation scores accurately reflected the pedestrian environment.

Limitations

An important caveat realted to the neighborhood averages presented in the table is that at not all street segments have available secondary data to assess. Data was most complete for the Outer Richmond where 82% of street segments had available data. However, no data was available for the Presidio or Treasure Island, and only 13 % of street segments in Lakeshore had availbale data.

Additionally, many of the data sets used as proxies are estimates themselves; including, traffic volume, sidewalk widths and througways, trees. This means there is inherant error in the values used to calculate the approximation, and thus, the PEQI Approximation is a "best guess" that should be ground truthed before using for decision making processes.

Data Source

PEQI Approximation Data was collected from a variety of sources:

Number of Lanes – SFCTA 2010 CHAMP Model, PM peak. Lanes known to have changed configuration since this data was released (2009) were manually updated to reflect their current configuration (Valencia, Cesar Chavez). This data only represents through traffic lanes and does not include turning lanes or pockets.  

Posted Speed Limit – SFMTA (2013) and includes 15mph school zones.

Traffic Volume – SFCTA 2010 CHAMP Model and represents 24 hour traffic volume (2009). For street segments where this data was missing (2.5%) the value was estimated by taking the traffic volume value from an adjacent touching segment. If no touching segments with traffic volume were present the traffic volume was estimated using an average of nearby streets (parallel or within a block) of the same type (e.g. neighborhood residential, industrial, etc.).

Width of Sidewalk (called Better Street Recommendation in approximation) – This variable was acquired from AECOM’s Sidewalk Nexus Study with the original source of the data being cited as DPW (2013?). A second source was used from AECOM’s Urban Watershed Assessment of the Sewer System Improvement Program for SFPUC to fill in any missing sidewalk width data. Between these two sources, sidewalk width data was available for 82.8% of the city streets. In cases where sidewalk width was available for both sides of the street, the maximum value was used. For the remaining streets where sidewalk data was missing, two approaches were used: 1) an estimation of sidewalk width based off the city curb data file and city right of way file using the near tool in ArcGIS and 2) a visual estimation using Google Earth’s ruler feature. 

Width of Throughway – This variable came from an AECOM analysis of DPW’s sidewalk width data for the Sidewalk Nexus Study (2013?). This study was used to determine if a sidewalk was within the recommended width as specified by the Better Streets Plan for each different street type. The original PEQI study measured the width of walkable throughway on a street. We feel that AECOM’s method is a good estimation of sidewalk capacity and a comparable to width of throughway. For segments that lacked data (33.53%) an estimation was created using sidewalk width data or estimated sidewalk width data and the Better Streets Plan recommendation.

Trees – Tree data came from two different sources: 1) Plannings’s street tree census data (2012) and for streets lacking census data 2) Science Application International Corporation LiDAR estimation of tree locations (2007) using a 150 foot buffer around the street segment. Tree data was then normalized per mile and cut points created based on the averages of existing PEQI survey data and street trees per mile data. The following proxy groupings were used: Less than 5 trees within 150ft per mile (None); 5 to 400 trees within 150ft per mile (Sporadically lined); More than 400 trees within 150ft per mile (Continuously lined).

Driveway Cuts (called Off Street Parking in approximation) – This variable is an estimation of the impact of vehicle traffic crossing over the sidewalk to enter parking lots. A comprehensive curb cut dataset doesn’t currently exist. We used an approximation of the impact of private/public parking lots and garages with data provided by the MTA’s parking census (2013). We felt this was a good estimation since these lots, given their size, probably have a disproportionate amount of cars exiting and entering over the sidewalk right-of-way. The following cut points were used: Off street lots with less than 25 spaces; off street lots with 25 and 50 spaces; and off street lots with more than 50 spaces.

Presence of a Buffer – This variable comes from 3 different sources: 1) the MTA’s bicycle network (2013), 2) the MTA’s on street parking census estimation (2013), and 3) the location of MTA’s parking meters (2013) (count within a 150ft buffer a street segment). The original PEQI survey included non-peak parallel parking which we were unable to find a suitable dataset of and this value was dropped from the approximation score. The following groupings were used: Bike lane (not sharrow) and count on-street parking greater than 5, bike lane (not sharrow) and count metered parking greater than 5, count on-street parking greater than 5, Bike lane (not sharrow), and count on-street parking 5 or less.

Planters or Gardens (called Amenities in approximation) – A comprehensive database of planters and gardens within the city does not currently exist at this time. We used the proxy variables for patio seating permits from Planning (2014); adjacency to parks and plazas (2013); and adjacency to community gardens (2013) – within a 150ft buffer of the street segment being considered adjacent. This variable is scored as yes if the count of any of these amenities is greater than 0. 

Empty Spaces – Empty spaces were approximated using adjacency to vacant parcels (Planning dataset, 2013); and adjacency to open air parking lots (MTA parking census dataset, 2013) – within a 150ft buffer of the street segment being considered adjacent. This variable is scored as yes if the count of any of these amenities is greater than 0. 

Overlay subset variables – These variables are not mapped but can be used to prioritize segments with low PEQI approximation scores.

Street TCFs – This dataset consists of traffic calming features and was provided by the MTA (2013). It has been noted by SFMTA staff that this dataset may be incomplete as it only records the last 5 years of traffic calming installations. Traffic calming features include:  speed cushions, speed humps, traffic islands, chicanes, and speed radar displays.

Sidewalk Impediments – This dataset was provide by DPW and consists of all complaints of sidewalk conditions (e.g. cracked and lifted sidewalks) from summer 2012 to summer 2014. This is a complaint driven system and may ignore streets with issues where residents do not make a complaint to DPW or 311.

Illegal Graffiti – This variable is complaints of graffiti on public property, private property, or sidewalks. This is a complaint driven system and may ignore streets with issues where residents do not make a complaint to DPW or 311.

Litter – This proxy variable consists of the sum of the following DPW actions: packer trucks for illegal dumping, abandoned waste, bulky items; steam cleanings of any time (e.g. urine, feces, general); street cleaning and general litter pickup; and litter from overflowing cans and can tip overs from summer 2012 to summer 2014. This is a complaint driven system and may ignore streets with issues where residents do not make a complaint to DPW or 311.

Map data is presented at the level of the street segment.

Table data is presented by anlaysis neighborhood.

Detailed information regarding census data, geographic units of analysis, their definitions, and their boundaries can be found at the following links:

Interactive boundaries map

http://sfindicatorproject.org/resources/data_map_methods

  1. PolicyLink, Prevention Institute, the Convergence Partnership. 2009. Healthy, Equitable Transportation Policy, Recommendations and Research. Available at:  http://www.convergencepartnership.org/atf/cf/%7B245a9b44-6ded-4abd-a392-ae583809e350%7D/HEALTHTRANS_FULLBOOK_FINAL.PDF