Fig 1.
Step-by-step personalized pedestrian network analysis.
(1) An annotated pedestrian network is generated from a geospatial dataset. In this case, a network of sidewalks and street crossing locations annotated with path length and incline (grade or slope) information was interpreted as a graph where edges are sidewalks and crossings, and nodes are where these elements meet end-to-end. (2) One or more pedestrian mobility profiles (PMPs) are enumerated, representing an individual pedestrian who may or may not represent a larger population. Individual pedestrian preferences are translated into a quantitative pedestrian mobility profile, which parameterizes a cost function that returns a numerical weight (infinity or a real number) based on metadata stored in a network edge (e.g., a sidewalk). (3) This PMP-parameterized cost function is evaluated over every edge in the network. This panel shows 1/15th of all weights to avoid overplotting. (4) Any appropriate or exploratory spatial network-based analysis can now be applied to this individually-weighted network. This panel shows the set of reachable paths for a stereotyped manual wheelchair user starting at one point in the Fremont neighborhood of Seattle, WA, where reachability is defined as a filled-in shortest-path tree.
Fig 2.
Alternative technical formulations of 400-meter pedestrian service areas.
(Left panel) A circular buffer, calculated “as the crow flies” from a point of interest, is overlaid on the transportation network without concern for physical barriers. (Center panel) A street-based walkshed that uses a monolithic pedestrian model more realistically models traversal of the street network but overlooks pedestrian-specific infrastructure and navigational concerns. (Right panel) A PPNA-based walkshed reveals areas inaccessible to a stereotyped manual wheelchair user. An area in the Northeast section of this map was considered reachable by the street walkshed approach but not by the PPNA approach due to steep sidewalks between the Northwest “Y” intersection and the starting point.
Fig 3.
Distinct PMPs generate distinct walksheds.
Each panel shows a PPNA-based walkshed originating at the same location but modeled with a different PMP. (Left) A normative walking PMP walkshed, reaching a wide, roughly diamond-shaped area in downtown Seattle. (Right) The walkshed of a more-constrained PMP that stereotypes a powered wheelchair user with moderate incline constraints and a requirement for lowered curbs when crossing the street. It is approximately ⅔ the size of the normative PMP walkshed, missing a large segment on the Southwest area due to steep hills and the lack of lowered curbs in the downtown neighborhood.
Fig 4.
Normalized sidewalk reach for normative and manual wheelchair PMPs for every street in Seattle, WA.
Normalized sidewalk reach was evaluated at the center point of every street in Seattle using either a normative (left) or stereotyped manual wheelchair (right) PMP. These normalized sidewalk reach values reveal granular spatial variation in access as well as a visual means by which to compare city-scale pedestrian accessibility between two pedestrians. Namely, the normative PMP generated NSR values exceed 0.75 for three large, contiguous regions representing the North Seattle, Central Seattle, and West Seattle/Delridge areas, whereas the wheelchair PMP generated NSR values rarely exceed 0.75, with many small islands of relative accessibility divided by large regions of poor (less than 0.25) NSR values. The figure contains information from OpenStreetMap and OpenStreetMap Foundation, which is made available under the Open Database License.
Fig 5.
Sidewalk reach quotients for manual and powered wheelchair pedestrian mobility profiles for every street in Seattle, WA.
Sidewalk reach quotients provide a (relative) quantitative basis of comparison for access to public sidewalk infrastructure between two pedestrians profiles. They were evaluated at the center point of every street in Seattle using either a stereotyped manual wheelchair (left) or stereotyped powered wheelchair (right) pedestrian mobility profile (PMP) for the numerator and a normative walking PMP for the denominator. The stereotyped powered wheelchair PMP is less constrained than the stereotyped manual wheelchair PMP since powered wheelchair users tend to report fewer concerns about steep inclines. While the maps produced by both PMPs exhibit a “splotchy” pattern, indicating wide spatial variation in this equity metric, the stereotyped manual wheelchair PMP frequently produces lower SRQ values. For example, the downtown region (near central, Western coast) has noticeably higher frequencies of high-SRQ values for the stereotyped powered wheelchair profile than for the stereotyped manual wheelchair one. This figure contains information from OpenStreetMap and OpenStreetMap Foundation, which is made available under the Open Database License.
Fig 6.
Per-sidewalk amenity access for normative and manual wheelchair PMPs for each street in the Seattle, WA region.
Access to two amenity categories (parks and schools, vertical axis) were evaluated by 400-meter walksheds using two pedestrian mobility profiles (stereotyped normative walking and stereotyped manual wheelchair, horizontal axis). For both amenity categories, the manual wheelchair PMP is constrained by inclines and the use of lowered curbs, leading to a distinctly smaller set of service areas. In both cases, the overall number of streets that can reach at least one example of the amenity (a school or public park) is dramatically smaller for the stereotyped manual wheelchair profile. In addition, the quantity reachable is noticeably higher (darker coloration) for the normative profile for both amenities. This figure contains information from OpenStreetMap and OpenStreetMap Foundation, which is made available under the Open Database License.
Fig 7.
Disconnected subgraphs in the pedestrian network.
A map shows sidewalks that have been uniquely colored based on sharing a unique subgraph. Subgraphs result from disconnections in the street crossing network as interpreted by a stereotyped manual wheelchair PMP; a stereotyped manual wheelchair user would have difficulty entering or leaving these subgraphs as a pedestrian due to the lack of lowered curbs. Over 100 disconnected subgraphs are shown, with one neighborhood (Madison Park) thoroughly and multiply disconnected from the larger pedestrian network.
Fig 8.
400-meter betweenness centrality for normative walking and manual wheelchair PMPs in Seattle, WA.
Betweenness centrality evaluated for a central, rectangular region of Seattle, WA. Values are normalized and unitless but share a color map, with darker colors corresponding to more central (higher betweenness) network segments. The upper panel shows betweenness centrality for sidewalk network segments evaluated by a stereotyped normative walking pedestrian mobility profile while the lower panel shows the same for a stereotyped manual wheelchair PMP. While some contiguous sets of pedestrian network elements have high betweenness within both profiles, indicating consensus on betweenness between the profiles, several disagreements become apparent. The downtown region (Western/left section of the map) has conspicuously divergent paths of high-betweenness values. Its normative profile produced more evenly distributed betweenness and a single major path in contrast to the manual wheelchair profile, which produced a small and different set of high-betweenness paths with very low betweenness values surrounding them, reflecting the steepness of downtown Seattle in the Southwest/Northeast directions. This figure contains information from OpenStreetMap and OpenStreetMap Foundation, which is made available under the Open Database License.