Home
About

CARA Access Metrics

This page provide a brief technical summary of how CARA calculates national distance & time access metrics.

Context

The time and distance it takes to travel to the services is one of the core characteristics of the built environment. Many of the key decisions people make, such as the purchase of a home, revolve around access to the education, health and government services they rely on.

Detailed and accurate measures of access to services are important to understanding of the supply and demand for services and central to the development of policy to fairly distribute the availability of services across a population

The Centre for Australian Research into Access (CARA) has applied modern data and computing processing techniques to develop infrastructure that rapidly calculates access to services metrics for every Australian private dwelling. Further, CARA has developed spatial micro-simulation models to simulates people, families and households within these dwellings, which allows access to different services to be evaluated within different groups of the Australian population.

The ability to calculate metrics directly from every dwelling in Australia to the building within which a service is located, allows CARA to produce metrics that are independent of any arbitrary spatial unit and therefore can be aggregated to any type of area. This allows these metrics to be free from the statistical biases and constraints imposed by arbitrarily aggregation such as the Modifiable Areal Unit Problem (MAUP) and the Ecological Fallacy (Openshaw, 1984).

Calculation Process

The inputs to the CARA time/distance metrics are;

  • A collection of starting points called origins from which to measures are made.
  • A network of lines composed of nodes and segments that model the topology (connectiveness), restrictions (one/two-way, overpasses etc.) and traffic speeds associated with the Australian road network.
  • A collection of end points or destinations (services) to which measure are made.
Origins

In calculating Service Access Metrics CARA uses the centroids of the building footprints of residential dwellings as origins. As Australia is lacking a Register of residential dwellings, CARA has generated a data processing methodology to model their location across Australia. The output of this modelling is called the CARA Residential Dwelling Reference Frame (RDRF).  The RDRF is an estimate of the location of residential dwelling in Australia based on the best available data on building locations and their use. After buildings associated with known non reidential services (e.g. petrol stations) are removed, a pool of possible residential building is created, based on its spatial  association with GNAF address points, Indigenous location points and the size of the foot print. This pool is used to distributed the Census of Population & Housing’s mesh block count of private dwellings to building footprints based on a range of criteria tailored to the nature of the mesh block.

CARA Residential Dwelling Reference Frame 2021
CARA Residential Dwelling Reference Frame 2021
Residential Dwelling Reference Frame in Darwin
Residential Dwelling Reference Frame in Darwin
Destinations

The destinations (service locations) are sourced from either national registers, such as ACARA or from commercially curated data sets of points of interest, when no official sources are available. Table X provides a listing of the registers used by CARA.

Road Network

The network topologically models the roads, junctions, and traffic directions of the physical road system. Each road segment is associated with an impedance (distance, speed). Data on road speeds is derived through aggregate measurement of active vehicles by TomTom. Time is derived from the minimum speeds during either the morning or evening peak periods.

Time/Distance Calculations

Shortest time/distance network times and distances are calculated by traversing the network using the Dijkstra’s algorithm which is enhanced by applying a Contraction Hierarchy Heuristic. This algorithm is implemented in C++ and enabled as a Python library called Pandana. Pandana it was developed by UrbanSim at the University of Berkley (Foti & Waddell, 2012)
The Pandana library operates by linking the point location of origins (and destinations) to their nearest node on the network. The distance of this straight line is derived, and the associated time is calculated based on a mean speed value across the whole of the network (20km/h). These times and distances are added to each of the shortest routes calculated by Pandana. For the majority of origins and destinations this distance is small in both absolute and relative terms because of the density of the road network in urban areas where the majority of origins, destinations and their associated routes exist.

References

Foti, F., Waddell, P. & Luxen, D., (2012) A Generalized Computational Framework for Accessibility: From the Pedestrian to the Metropolitan Scale., Transportation Research Board Annual Conference.
Openshaw, S. (1984). Ecological Fallacies and the Analysis of Areal Census Data. Environment and Planning A: Economy and Space, 16(1), 17-31. https://doi.org/10.1068/a160017
Openshaw, S. (1984), The Modifiable Areal Unit Problem, Concepts and Techniques in Modern Geography, Geobooks.