Measuring Park Access in
Oregon
Author: Josh Roll
Unit: ODOT Research Unit
Email: josh.f.roll@odot.oregon.gov
Github Handle: JoshRoll
Date: October 2025
Unit: ODOT Research Unit
Email: josh.f.roll@odot.oregon.gov
Github Handle: JoshRoll
Date: October 2025
1 Introduction and Motivation
In 2022 the Oregon Legislature passed HB 4077 which directed the Environmental Justice Council (EJC) to build the Oregon Environmental Justice Mapping Tool (EJMT) and for agencies to consider its results when developing rules, policies, and programs. The tool’s core purpose is to pinpoint areas disproportionately affected by pollution, climate risks, health burdens, and socioeconomic barriers, so the state can prioritize outreach, services, and investments where need is highest.
Access to parks has been demonstrated to provide both physical and mental health benefits(Derose, Han, et al. 2021). However, access to parks is not uniform and past research has demonstrated that across the U.S. only 39% of people have access to parks with states like West Virginia only having access to parks for 9% of the population. Additionally, disadvantaged populations have been shown to have less access to parks than more affluent groups (Ussery et al. 2016).
The work documented in this report summarizes the technical process for developing a measure of access to parks for the EJMT. This access to parks measures will be combined with other measures gathered through the public outreach process and with direction from the EJC to become a tool for helping agencies guide equitable government action. Agencies can use the tool to target community engagement, inform program design, and consider resource allocation to communities historically under-represented in public processes and harmed by environmental and health hazards.
The data and methods for the park measures are described below. For questions about any of these technical processes feel free to email Josh Roll at Oregon Department of Transportation using the contact info above. All code is available at the repository here
2 Data and Methods
The goal of this analysis is to determine how access to parks by Oregon’s population varies across the state at the tract level. To meet this objective an inventory of park locations is needed as well as routable street network. An up-to-date and constantly maintained database of park locations in Oregon is not a dataset maintained by a state agency. And given much of the uncertainty with federal government resources and technical assistance it was desirable to use data that does not require the federal government for maintenance and access. After reviewing data from USGS called Protected Area Dataset(PAD) and data from Open Streets Map (OSM) it was decided that OSM data best suited the needs for this project.
2.1 Park Location from OSM
Parks data from OSM was collected in September of 2025 using the osmdata (Mark Padgham et al. 2017) R package that harnesses an application programming interface (API) to streamline retrieving data. The API calls were made for each county and then combined with census block data retrieved using the tigris (Walker 2025) package using the R open source statistical computing language. Once the parks locations are retrieved they are spatially joined with Census blocks so that all blocks have a measure of the number of meters2 of park area. Other green spaces exist that are not classified as city or urban parks but those are not included in these data. The map below allows readers to explore select parks in the Marion county.
2.2 Street Network Data from OSM
Measures of access to parks have mostly used Euclidean distance (Derose, Han, et al. 2021) but this approach can over estimate access since barriers to walking such as an interstate or rivers are not accounted for so a network-based approach was preferable. In order to use a network-based approach a street network is needed that is topologically accurate and able to be converted to a graph. Many applications and research efforts have shown that the OSM data, though not without its limitations, can work well for this kind of analysis (Graser, n.d.). Some of the limitations include the fact that these network data are mostly crowd sourced and transportation authorities do not use these data as the official data of record. However, these data are the best available and also work well with an available routing algorithm described below.
In the map below the OSM network data is presented showing all the network links within 800 meters of the Mt. Scott Park in Portland, Oregon. Thi spark in Portland will be used below to highlight the data and methods used to calculate park access. The limited data (buffer of 800 meters) is shown instead of all the network data since it allows the map render quickly but network data for entire state was used to develop statewide measures of access to parks.
2.3 Measuring Access by Routing Walk Trips
As mentioned above, many analysis have used simple measures of proximity to determine if people have access to parks but this approach can oversimplify conditions on the ground where barriers to these amenities from controlled access interstates or bodies of water are not taken into consideration. To account for barriers to walking and more accurately measure access to urban parks a routing algorithm is used in conjunction with the OSM street network data. This work utilized the Rapid Realistic Routing on Real-world and Reimagined networks (R5) (Pereira et al. 2021) R-based routing package which informs the back-end of the Conveyal accessibility scenario planning tool. Previous studies have used the R5 routing engine to measure access to amenities such as parks, groceries and libraries Goetz and Zipf (2016) with success. The R5 routing engine uses the OSM network data to route trips from origins to destinations and can show routes used for those trips while also calculating accessibility to amenities such as parks.
For this work, origin and destination points were developed by using the centroids of Census blocks. This approach helps to make the routing analysis more computationally efficient than if a more disaggregate set of origins and destinations were used such as household unit locations. The map below shows an example of how the R5 routing engine works by showing the walk trip routes to the Mt. Scott park in Portland, Oregon. Many routes are used by more than one origin so the road network segments are visualized to display the number of walk trips from each origin point to Mt. Scott park (destination) with lighter color indicating segments that were used more than others and darker colors reflecting lower count of walk trips. This example shows the Census blocks able to access the Mt. Scott park within 10 minute walk and which routes the R5 routing engine assumed people walking would take to get to the park. These routes assume a shortest distance path and do not account for perceptions of traffic or crime safety that might alter a person’s revealed trajectory.
Access to urban parks is calculated for all Census blocks in all of Oregon’s 36 counties using the R5 routing engine, the OSM street network, and the inventory of urban parks available from OSM. The calculated park area accessibility measures are shown for a sample of 1,433 blocks (blocks within 2,500 meters of Mt. Scott Park) to show how the access measure appears for the example area around Mt. Scott park though the map below also shows block level measures of access to other parks within the selected buffer.
2.4 Creating Tract Level Measures
The Census block measures are created using the methods above but for the EJC mapping tool, tract level measures are required as that is the level of geography where EJC mapping measures are being developed for the state. This work uses population weighted aggregation, a common approach to aggregating spatial units from different geometric levels (Spoer et al. 2023). This process is accomplished by aggregating block level measures to the tract level by weighting each block-based access measure by the proportion of the Census tract’s population represented by the block and then averaging all the block’s access to parks within that tract. Population data used in the weighting comes from the decennial (2020) Census product since this is the only vintage of data available for data at the block level. Block level population estimates were aggregated to the tract level to get the tract level measure of population for the weighting calculation. Tract level measures are presented in the final map below for Multnomah County.
3 Overall Workflow
The overall workflow is presented in the figure below. This graphic summarizes how the two R scripts, titled get_osm_data.r and calc_access_to_amenities.r retrieve data from OSM and U.S. Census, perform necessary spatial operations and prepare those data for routing. The figure then shows the routing and population-based weighting process that develops the final tract level measures.
Figure 3.1: Process Work Flow
4 Download Data
To download the spatial data via geodatabase click on the link below to start retreive the statewide data.
⬇️ Download the data .gdb