Visualizing Accessibility with GIS
Marc Schlossberg, Ph.D.
University of Oregon
Planning, Public Policy, and Management
As the environmental, social, and health costs of sprawling, automobile dependent development patterns become well understood, accessibility, or walkability, becomes a significant goal of planners, policy makers, and citizens. Our current planning environment is one of auto-mobility, which has the goal of reducing the cost per mile of travel within a metropolitan area. An auto-mobility approach may find success for a 15 minute commute that travels fifteen miles at speeds of sixty plus miles per hour – the cost per mile is relatively low in terms of time and delay. Similarly, an auto-mobility approach to regional travel would be considered a failure when congestion inhibits automobile travel from traveling at maximum speed limits; the cost per mile becomes quite high on account of time delays in traffic.
In contrast, an accessibility focus of development seeks to help people gain access to their destinations at a low cost per trip. In an accessibility-centered approach, popular places to visit cause increased numbers of people on sidewalks and in street intersections. These increases in turn tend to slow down the speeds of automobiles in the area. There is a tradeoff of mobility that favors the pedestrian rather than the automobile.
Developers and planners are increasingly incorporating such tradeoffs involving pedestrian accessibility into their visions and plans. They tend to base their decisions on a variety of principles, including increased quality of life, more active community interaction, environmental benefits of reduced automobile dependence, and congestion reduction. These principles are often characterized, in part at least, under a variety of terms: "New Urbanism," "Neotraditional Planning," "Pedestrian Pockets," "Transit Oriented Development," or "Nodal Development." The claimed or potential benefits of these schemes is beyond the scope of the current discussion. The focus here is on visualizing accessibility principles: to visualize is to clarify.
What are the various ways
that one can visualize accessibility using Geographic Information Systems
(GIS)? This presentation uses the centralized area of Eugene, Oregon
(USA) as the case study. Eugene has a centralized downtown with a gridded
street network, has several old, established neighborhoods, and has some
newer developments as well. Most of Eugene’s topography is flat, except
for portions of South Eugene, which ascends up some foothills. Eugene has
clearly identified neighborhoods that are recognized by the City and are
represented by elected neighborhood association presidents. Measuring accessibility
at a neighborhood scale can be facilitated by these pre-existing boundaries
of the neighborhoods.
Accessibility through Buffering
An easy way to visualize accessibility to a specific place is to use "buffer." Buffers target areas all of which are within a given distance of a point, line, or area. Thus, Figure 1 shows four buffers around the Library location. Buffering is a common GIS technique and can be used to quickly identify a geographic area that is considered accessible or walkable to a given location. Planners often consider a ¼ mile distance from a location as being the maximum distance that people are willing to walk to get to the destination they desire. Thus, Figure 1 shows ¼ mile rings of accessibility to a new downtown library that is being constructed in Eugene. The buffer rings (in Figure 1) are “as the crow flies”, and do not take into account the actual paths that people may need to take to access the library. Thus, Figure 2 shows ¼ mile rings around the library based on the actual walking path of the street network (assuming that all streets have sidewalks and that there are no other walking-only paths). The diamond shaped buffer rings reflect the gridded street pattern of this part of Eugene.
When Figures 1 and 2 are combined as Figure 3, the new Figure shows the overlap between the two different accessibility measurements. In this so-called “Ped Shed” of Figure 3, the ¼ mile buffer area of each technique can be compared by dividing the area of one by the area of the other to calculate a Ped Shed ratio [Rood, n.d.]. Different ratios imply areas that are more or less walkable.
Additional aspects of urban life may also be identified within the walkable buffers. For example, planners at the library may wish to provide sensitive services to people with special needs for social services located near the library. Figure 4, plots the location of social services with the buffer rings to give the library a sense of the type of potential demand it may receive from any of a number of specialized populations.
While the image of Figure 4 is fairly intuitive and easy to read, additional visualization manipulations are possible to increase the clarity of the information being presented. Since the data underlying the image is spatial, data within each buffer can be individually selected and color coded based on its location. Figure 5 illustrates this approach by altering the color of variables (buffer, streets, and social services) based on geographic location. Thus, the visual representation of accessibility is enhanced and the capacity to distinguish or visually segregate the data based on geographical location is improved.
Figure 5: Color Coding Data by Distance
Figure 6: 3D Tiers of Accessibility
Figure 7: 3D Tiers of Accessibility with Social service Program Locations
Accessibility through Intersection Density
The images above visualized accessibility in terms of the distance to a specific place. One might, instead, look across a landscape to ascertain which sub-areas are characterized by potentially more accessible movement patterns. Some areas within a region may have street networks (and therefore sidewalk networks) that are more conducive to walkability. Thus, accessibility may be visualized by investigating different patterns of street networks. Within the development schemes mentioned at the outset (New Urbanism, Nodal Development, and so forth), one idea is that street patterns that are based on a grid are more accessible than non-grid patterns. Within a gridded street network, there are redundant paths that walkers can use to access the same destination. This increase in path choice can be represented by areas with numerous street intersections and thus relatively great accessibility. One way to view this idea is to consider the difference in numbers of intersections and accessibility between a downtown street network grid suburban development with many cul-de-sacs. Regions with higher concentrations of intersections are regions with higher potentials for accessibility. The following series of images visualizes this characterization of accessibility.
Figure 8 shows the street
pattern within the central Eugene Neighborhoods. The downtown core is located
at about the center of the map. From only this simple map of one layer,
it is visually possible to get a sense of which areas in Eugene are more
Figure 8: Eugene Street Network
Although one can get a general sense of accessible places by simply viewing the street layer, it is possible to perform a series of calculations based on the location and density of intersections (or cul-de-sacs). By viewing the concentration of intersections, one can get a better grasp of the connectivity of the street network across space. Figure 9 and Figure 10 visualize the street network based on the location of intersections and cul-de-sacs (or dead-ends).
Figure 9: Intersection and Dead End Points
Figure 10: Close-up of Intersection and Dead End Points
Visualizing concentrations of intersections is helpful, but it may be that one would want to characterize the different neighborhoods in Eugene based on the density of intersections within the neighborhoods. Neighborhoods with higher intersection density (intersections per square mile) might be considered as more accessible than those neighborhoods with lower intersection densities. Figure 11 visualizes the aggregation of intersections within each neighborhood divided by the total area of each neighborhood to calculate a relative intersection density figure. Figure 12 visualizes a similar calculation, but is based on the concentration of cul-de-sacs – areas that can be classified as having low accessibility.
In Figure 11 there is a clear
pattern of higher accessibility in the centralized area of Eugene, the
location with the tightest grid pattern of development. This is the oldest
developed portion of Eugene and was developed before the predominance of
automobiles. The lighter colored neighborhoods out to the west are areas
where more industrial development has occurred and the street network,
and thus the density of intersections, follows a much less dense pattern.
In Figure 12, the areas that have a more characteristic suburban style
of development are clearly visualized. The southern portion is hilly and
the street network tends to transect the mountains in long straight swaths
with few intersecting streets. The dark area to the north in Figure 12
is an area more recently developed and follows a street pattern much more
characteristic of the post-war suburban approach.
Figure 11: Street Intersection Density by Neighborhood
Figure 12: Dead End Density by Neighborhood
The figures above aggregate intersections to specific Eugene neighborhoods, which allows one to visualize accessibility on a neighborhood by neighborhood basis. Aggregating intersections to these pre-defined boundaries, however, is a bit artificial in nature. Alternatively, as shown in Figure 13, intersection density can be calculated by exact location in space. The intersection density of each spatial location can be calculated and then visualized based on the number of intersections that surround it. By transforming the vector data above to raster data (cells), a computation of the intersections within a ¼ mile of each cell can be calculated and displayed. Individual cells that are centrally located in relation to many intersections will appear in darker colors. Thus, regions of high intersection density can be visualized independent of the arbitrary borders of neighborhoods (or city boundaries, census tracts, and so forth). The neighborhood boundaries in Figure 13 are displayed, however, to give reference to the intersection density visualization.
Figure 13: Intersection Density by Point Location
The same type of calculation
and visualization can be conducted on the density of dead-end streets or
cul-de-sacs as shown in Figure 14.
Figure 14: Dead End Density by Point Location
Figures 13 and 14 suggest locations where development has occurred in a way that is highly walkable and highly unwalkable.
Figure 15: Elevation by Intersection
The three dimensional approach
can be further augmented by overlaying the street network on top of the
intersection topography to help visualize the concept of accessibility.
Figure 16 illustrates this combination with streets within ¼ mile
of the library highlighted in pink. That Figure also shows that the location
of the new library is on the most accessible land of downtown Eugene. While
not shown explicitly, the center of the pink streets (the location of the
library) is just to the right of the tallest mountain peak (the location
of the highest intersection density).
Figure 16: Intersection Elevation and Streets
Finally, aerial photographs
can be draped on top of this new intersection topology to allow one to
visualize the actual development of an area in relation to intersection
density. Figure 17 visualizes accessibility using color aerial photos and
the intersection-based topography. Some areas on the image below do not
have aerial photos displayed in order to reveal the underlying connectivity
as illustrated in Figures 15 and 16.
Figure 17: Intersection Elevation with Aerial Photos
In Figure 17, then, one can
visualize the landscape of a city in a new way based on accessibility.
Areas of high accessibility can be represented as mountain peaks (or alternatively
as flat spaces) and the photographs of actual development can be viewed
with this new underlying elevation. A policy connection, as well as a visual
connection, might then be made between development patterns and accessibility.