About me
Do green space accessibility measures vary by which data source you use? - Urban parks offer benefits for physical activity, public health, and overall well-being. However, measuring accessibility to these spaces relies on spatial data sources that vary in coverage, accuracy, and digitization standards—potentially introducing bias into accessibility analyses.
This study investigates how data source selection influences park accessibility measures by (1) validating the accuracy of common data sources and then (2) assessing their impact on accessibility measures. We focus on large public urban green spaces (LPUGS), defined as green spaces of at least 2 acres that are accessible to the public at least some of the time. We compare four widely used data sources: OpenStreetMap (OSM), the Trust for Public Land (TPL), Esri, and SafeGraph.
We validate these LPUGS from different data sources against real-world park locations using satellite images and Google Street View. This manual validation process uses three U.S. cities of different sizes and demographic characteristics (Portland, Buffalo, and Vicksburg) as case study sites. Then we measure each data source’s accuracy through precision, recall, and F1 scores. Next we perform a network-based isochrone analysis at 15- and 30-minute walking thresholds to quantify park access through (1) the cumulative count of parks within isochrones and (2) the cumulative park area within isochrones.
Our findings reveal that park accessibility measures do vary significantly depending on the data source used, which can impact empirical findings and policy decisions. In particular, OSM and TPL capture more real-world parks but also have more false positives (duplicates), whereas Esri has higher precision but lower recall, potentially underestimating access. These differences lead to different estimates of the number and total area of accessible parks, affecting how spatial equity is understood by researchers. We argue that relying on a single data source may introduce bias, underscoring the need for careful selection in accessibility research.
