ParkServe® includes a comprehensive standardized database of local parks in the 13,913 cities, towns and communities. In phase one, census designated urban areas were used to define where to collect and create local data for cities, towns and communities. For each city, town and community, geographic boundaries were obtained from the US Census 2010 Places geospatial dataset and associated population estimates are derived from ESRI’s 2017 Demographic Forecasts. The ParkServe® team attempted to contact each city, town and community with a request for their parks data. If no GIS data was provided, the ParkServe® team created GIS data for the place based on available resources, such as park information from municipal websites, GIS data available from counties and states, and satellite imagery.

Cities, towns and communities were then emailed a link to view the park data compiled in their area to verify the boundaries and attributes of the parks in the database through our custom web-based ParkReviewer™ application.

ParkServe® Data Inclusion

Property eligibility criteria for ParkServe®:

Examples of property types not included in ParkServe®:

The ParkServe® local parks database will be freely available for download later in 2018. Additionally, the data will be integrated into the USGS Protected Areas Database of the US (PAD-US) and the combined database will be available for download later in 2018.

Service Areas

For each park, the ParkServe® team created a 10-minute walkable service area using a nationwide walkable road network dataset provided by Esri. We use the Esri Network Analyst extension to create the 10-minute walk service area. The analysis identifies physical barriers such as highways, train tracks, and rivers without bridges and chooses routes without barriers.

In order to create service areas, we created access points for each park using an auto-generation model. The model generates access points at any location where a walkable road is within a predefined distance of the park boundary. ParkServe® does incorporate the verified access points for the 100 cities in ParkScore® (see “How is ParkServe® similar to or different from ParkScore®?” below).

Using the 10-minute walk service areas, overall access statistics are generated for each park, place, and urban area included in the database. Access is then disaggregated by several demographic variables – race/ethnicity, age, and income. In communities with an exceptionally small number of block groups, 10-minute walk demographic calculations are not available.

All calculated population statistics are based on 2017 US Census Block Group estimates provided by Esri.

Park Need

All populated areas in a city that fall outside of a 10-minute walk service area are assigned a level of park need, based on a weighted calculation of three demographic variables from the 2017 Forecast Census Block Groups demographic data provided by Esri:

To ensure a large enough sample size of demographic data for each analysis, park need weightings for smaller cities, towns and communities included data from nearby jurisdictions.

Optimized Points

For each city, town and community, the ParkServe® team created up to 5 optimized locations where new parks could make a biggest impact on the 10-minute walk in that neighborhood.  Initial points are generated in areas with the highest total census block population and highest level of park need. A 10-minute walk service area is created for each optimized location. The optimized locations are then ranked by the number of potential new residents served.


How is ParkServe® similar to or different from ParkScore®?

Because ParkServe® is a project that covers almost 14,000 cities, towns and communities, we needed to take a different approach to our process and methodology for analyzing the 10-minute walk than we took for ParkScore® which covers the 100 largest cities in the United States. Some of those differences are outlined below.

  1. In ParkScore®, all publicly accessible parks and open space are collected from each city and cross checked with the database collected by the Center for City Park Excellence in their yearly City Park Facts survey. In ParkServe®, of the 13,913 cities, town and communities, only 1,622 (12%) were able to provide GIS data for their park systems. Thus, much of the ParkServe® database was manually created without local knowledge. Approximately 636 agencies provided edits to the ParkServe® database through the ParkReviewer™.
  2. In ParkScore®, the city boundary was provided by each city directly. In ParkServe®, city, town, and community boundaries were obtained from the US Census 2010 Places geospatial dataset.
  3. In ParkScore®, the 10-minute walk service areas are created using a local walkable street network for each city. In ParkServe®, a nationwide routing network, ESRI’s StreetMap Premium is used.
  4. Access points are automatically generated in ParkServe® based on the proximity of streets immediately surrounding each park. These access points were not reviewed for quality assurance. In ParkScore®, every access point for every park is quality checked to make sure access is permitted into the park at that location. This quality control is done for all 100 cities in ParkScore®. The auto-generated access points using ParkServe® methodology were replaced with those used in ParkScore® for the 100 cities.
  5. In ParkServe®, regional (urban area) median household income is used to define low income when calculating Park Need. In ParkScore®, each city’s median household income is used to define low income.
  6. In ParkScore® 2017, the demographic statistics are based on 2016 Forecast Census Block Group data provided by Esri.This release of ParkServe® uses 2017 Forecast Block Group data provided by Esri. ParkScore® 2018 (released in late May 2018) will match ParkServe® and also use 2017 Forecast Block Group data provided by Esri.