Walking to scale: an index to assess walkability at the residential scale

Abstract

Background: For the most part geographic information science has relied upon aggregate data and areal analysis. This approach, however, is plagued by two systematic shortcomings. The use of areal neighbourhoods introduces unavoidable statistical errors, while the results are only applicable to the specific neighbourhood unit used in the analysis. Taken together this severely limits the utility of such science. However, modern methodology and computing means that it is now possible to overcome these shortcomings through the use of disaggregate analysis. Using Auckland, New Zealand, as the backdrop, this thesis seeks to apply modern methodological and computational approaches to develop of a residential scale walkability index which would avoid these inherent shortcomings. Methods: Using a geographic information system, a residential scale walkability index was developed at both the five and ten minute walking scales. For each individual property parcel (n = 88,348), normalised scores were calculated for population reach, network connectivity and land use mixing. These three normalised metrics were then brought together to create a total index score between 0-1, with larger scores suggesting a more walkable environment. Results: Normalised metric scores, and a total index score, were assigned to each parcel. When mapped out the results demonstrate that the central city and neighbouring inner-west suburbs were the most walkable areas. Other highly walkable areas were found around the north-south colonial-era thoroughfares through the centre of the isthmus, and in the metropolitan centres of Onehunga and Otahuhu. On the other hand, large swathes of the eastern isthmus and south-west suburbs were characterised as low walkable areas. By implementing the index at the residential scale, the statistical errors associated with using neighbourhoods was avoided entirely. Conclusion: Modern computing and methodologies can now enable the implementation of residential scale walkability indices. These indices are free from the statistical shortcomings that affect neighbourhood scale indices, and provide a more detailed, and realistic, representation of the urban environment. As such, these disaggregate approaches are thus able to produce better science. Key words: Auckland, Geography, Geographic Information Science, Scale, Walkability