Sunday, 9 February 2014

Does sprawl lead to shorter commutes?

Examining the relationship between urban area density and median commuting distance yields a surprising result. Canada's less dense urban areas tend to have shorter commutes, the linear trendline has an R2 of 0.2688. This excludes the Abbotsford, Barrie, Oshawa and Hamilton urban areas since these have a relatively high percentage of workers commuting long distances to other metropolitan areas, although I've included them in the graph for illustrative purposes.

Toronto despite having the densest urban area and one of the most heavily used transit systems has some of the longest distance commutes, only surpased by Oshawa's metro area. Unlike most cities, the Oshawa CMA has a bimodal commute distance distribution with 33% commuting over 25km mostly to Toronto and 57% commuting less than 14km mostly within the Oshawa CMA. Barrie is even more bimodal, and Abbotsford slightly less so.

Anyhow, the correlation between dense urban areas and long commutes exists, but it's not very strong. Could something else be going on?

Looking at the relationship between density and population, there is a moderate correlation, especially among Canada's largest cities with an R2 of 0.6027. Not too surprising, most people associate density with big cities.
So are these higher densities enough to keep the bigger cities from covering much larger areas? Not really, the densities are actually not that much higher in big cities so there is a strong (R2 0.8944) correlation between population and land area.

So is it possible that low density cities have smaller commutes because they generally have much smaller populations that allow them to cover small land areas (despite lower densities)?
When excluding Oshawa, Barrie, Hamilton and Abbotsford, the correlation is quite strong at R2 = 0.6865 using a linear trend (although it might be more logarithmic (R2 = 0.7073) as eventually commuters start to stick to their corner of the metro area). 

Moderately correlation*moderate correlation*strong correlation = weak correlation between low densities and short commutes. 

The shortest commutes for major metropolitan areas are 4.6 km for Regina and Victoria, both relatively small cities that are dominant within their greater region. The census agglomerations with the shortest commutes are mostly small towns.

Hawkesbury: 2.0km
Swift Current: 1.9km
Estevan: 1.8km
Wetaskiwin: 1.8km
Prince Rupert: 1.7km
Dawson Creek: 1.7km
Elliot Lake: 1.7km
Thompson: 1.6km

Ironically, Thompson is one of a only a few cities that is a 100% post-WWII mining town with autocentric malls as its town centre. 

And just for fun

R2 = 0.4199 for density vs area

 R2 = 0.5743 for pop vs distance 

Note on methodology:

The median commuting distance data from StatsCan was only available for 2006 and for either municipalities or metropolitan areas, but not urban areas. However, density values at the metropolitan areas are essentially meaningless and municipalities are far from closed systems for these purposes with high amounts of commuting between the various municipalities of a metropolitan areas. Therefore, 2006 urban areas (population centres) were used. In some cases, a relatively high proportion of the metro area residents lived outside the main urban area, so smaller urban areas within the metropolitan area (ie exurbs and satellite towns) were added until the urban population reached 90% of the metropolitan total or included all urban areas within that metropolitan area*. The population, land area and density values were obtained using these combined urban areas.

*Some metropolitan areas had a surprisingly high share of residents living outside urban areas. Over 25% for Kingston and Peterborough and Saint John, Moncton, Saguenay/Chicoutimi, Sherbrooke and Halifax all had over 19% of metropolitan residents living outside urban areas, making these metropolitan areas more rural than Canada as whole.

No comments:

Post a Comment