Friday, October 14, 2011

Measuring Connectivity, part II

In the previous post, I mentioned the concept of metric reach which is being developed by John Peponis and his students (including commenter Patrick!) in Atlanta.  What I didn't say was why I thought this particular measurement so useful for measuring connectivity.  Again, consider the reinforcing virtues of density, mixed uses and accessibility:
"Well functioning cities can ... be thought of as 'movement economies'. By this it is meant that the reciprocal effects of space and movement on each other ... and the multiplier effects on both that arise from patterns of land use and building densities, which are themselves influenced by the space-movement relation, that give cities their characteristic structures, and give rise to the sense that everything is working together to create the special kinds of well-being and excitement that we associate with cities at their best." Bill Hillier, Space is the Machine, p. 113-14.
Or, as put more bluntly by the early 20th century economist Paul Nystrom:
"The greater the number of people, other things being equal, who live near, who come to, or who pass by a certain location, the more valuable that location is." Paul Nystrom, The Economics of Retailing, p. 189.
As Hillier alludes to, the overall effectiveness and efficiency of urbanism can be expressed as a multiple of density, land use and movement.*  For a person either living or visiting a city, is there a higher purpose of its streets than providing access to the greatest possible number of attractions over the shortest possible distance?  This is essentially the basis behind the Walkscore website, and it is what metric reach simply and elegantly tries to show us: a city's overall potential for providing the greatest range of nearby options from each point in the urban street network. 

One study examining this measure in the context of walking to transit stations in Chicago, Dallas and Atlanta has shown its usefulness in predicting pedestrian behavior:

"When multivariate regressions are run ... street connectivity [based on metric reach] is found to be a rather significant predictor of ridership levels in all three catchment areas when controlling for population density and transit station measures. However, the best results are obtained for the 0.5 mile range. This supports the findings of various studies which suggest that within short distances people will walk to transit regardless of local street connectivity ... . In other words, people residing within 0.25 mile distance from a station are inclined to use transit irrespective of the street connectivity levels of the station area. Higher correlation coefficients within the 0.5 mile buffer suggest that the decision to walk a slightly longer but still very manageable distance is strongly affected by the density of street connections."  The Effects of Street Configuration on Transit Ridership, Ozbil, Peponis and Bafna (2009).
Since the area between .25 and .5 miles from a given point contains three times the total area as the space within .25 miles, this is an important insight for TOD at the very least, and for thinking about walkability in general.

Even reach, as great a tool as it is, does not capture the whole picture.  Steve Stofka on his blog draws attention to what he has termed "traversal amenities," which represents the ease and comfort in walking a given distance.  This is not a subjective point: a line drawn down the midpoint of a street cannot tell us its width, or its traffic conditions, or any of the other factors which may affect as simple a decision as crossing from one side to another.  Disincentives to crossing a street, however, may reduce a pedestrian's range of movement options, effectively limiting metric reach (Walkscore, which Steve notes does not take into account these factors, has acknowledged this limitation). 

Ultimately, measuring pedestrian connectivity, and putting a number to it, is a difficult task given the  fine-grained and spontaneous nature of pedestrian accessibility and movement, and the ease with which it can be disrupted by vehicular traffic or other obstacles.  As it is, the most intricate feats of "pedestrian engineering" currently take place in indoor, self-contained or controlled environments, such as malls, stadiums and cruise ships rather than in urban planning, but Peponis and his students, through their efforts, are developing one way of thinking that may help change that.

Next and last in the series: a quick look at navigability, or, finding your way around.

*See also Modeling street connectivity, pedestrian movement and land-use according to standard GIS street network representations: A Comparative Study, p. 1.