The potential for different transport modes in Melbourne (part 1)
In this previous post we spoke about how to understand the different types of spatial averages, and asked how we'd understand their stability. This blog looks at beginning to use potential maps to better understand populations.
We will be looking at public transport in Melbourne. We want to understand the demand for services (and ultimately predicting demand potential). The first step is to map where users of services live. We want to map this distribution so we can ask ourselves questions like:
- Do people who use the public transport network do so because they live near a lot of transit options and it is convenient?
- If not, can we find pockets of patrons who are specifically travelling to use a service?
- Knowing the demographics of the current patrons, can we infer what the unrealised preference looks like? (This requires a bit more modelling so we will leave it for another post).
About the methods
To answer the question above we used spatial distribution potentials, and not point estimates like the arithmetic average, which we showed in previous work do not necessarily represent a helpful value.
We have based this work off the VISTA (Victorian Integrated Survey of Travel and Activities dataset). We sourced all population counts and spatial geography from the 2016 Australian census.
In the contour maps below we wanted to show the current trip generation potential throughout Melbourne for each mode of transport. From the travel data we know how many trips are generated from the households in a given area (we used SA2s). We then generated a map of how those trips varies across the whole region, by looking at how similar each region is to each other region.
The easiest way we have at the moment to visualise usage potentials is through a contour map, highlighting the iso-lines of common potential.
These contours describe the shape of usage patterns, including local focal points, and the main balance point of activity (shown with a pin). In these contour maps, the colour of the lines respresents the value of potential - yellow to green is lower potential, and dark purple is high potential.
Where the lines are fairly even, there is little variation. Where they are closer together there's a more rapid change from one region to the next.
As we are using SA2 aggregations, you can get artefacts of little hilltops around the centroid of the SA2s themselves. These can be resolved by moving to finer resolution, but that's not the aim here.
In all these maps, the contours show where people live who use there transport networks.
The balance point is in Box Hill North. It's not surprising that this is a focal point - there are a large number of bus routes servicing this area (as the PTV map on the right shows).
In general, there is a ridge extending south from this activity centre; connecting the Box Hill railway station and Westfield Hub. There's also a ridge running west along the northern edge of the city.
Not surprisingly the tram network supports mainly those who live near it. There is a broad flat plateau running from Brunswick, through East Melbourne and out to Hawthorn.
There is the suggestion of a slight tail to this plateau, heading out towards Caulfield. Even though there is a tram service here, where is the trip generation?
And here they are. These people tend to use the rail network more. We have a big triangular peak, covering the triumvirate rail connections of East Malvern, north to Camberwell Junction, and west to Prahran.
The natural question remaining is the private travel and cars. What does this look like?
Strangely enough it is a much more normal population contour. It balances near Glen Iris/ Camberwell, with an east west ridge running from Essendon to Doncaster, and a southern ridge running to Glen Waverley.
That's right, it looks like the population of the Greater City of Melbourne, because car trip generation is not a distinguishing factor. We all do it.
Want to know more?
This post is an example of some spatial methods we have been developing, to help understand the relationship between transport, people and place. In future posts we will look at more nuanced metrics and ways to predict unrealised transport potential.
We are developing these tools and methods all the time and would love to hear your comments, suggestions or applications. Drop us a line in the comments or get in touch.