Running every suburb in Melbourne

A while ago I wrote that my Canberra street-running project was defunct and that a Melbourne version was coming. Finally the Melbourne version is here!

I don’t currently have the goal to run every street in Melbourne, but instead I want to run at least 1km in every suburb of Melbourne at least once.

Progress so far

Progress data is not available in this build. Run scripts/update-melbourne-runs.sh to add data/melbourne_progress.json.

Suburb definitions

For now I’m using the Victorian government’s gazetted locality boundaries, from Vicmap Admin’s LOCALITY_POLYGON layer, trimmed to a bounding box around metropolitan Melbourne. This isn’t the only way to define things, but it’s what I ended up on.

How it works

Runs come off Strava as GPX files. Each trackpoint gets tested against the suburb polygons, which is a single spatial join rather than anything clever, and the result is the table and map above.

I originally tried to do this by matching runs onto a graph of Melbourne’s streets, which is what you’d want for the “minimise retraced streets” part of the project. It turns out to be the wrong tool for this question. Suburbs are big simple polygons, so the GPS noise that makes graph matching fiddly — a point landing on the wrong side of a road, a tunnel, a bridge over a river — doesn’t matter at all when the thing you’re asking is “which 2-kilometre-wide shape is this point in”. Point-in-polygon is both simpler and more robust here. The street graph is still there; I may use it for some route planning later.

What’s next

The interesting part is the planning problem. With 457 suburbs to go and a fixed starting point, I’d like to be able to ask for a route of a given distance that picks up as many new suburbs as possible without retracing steps, and I may include some sort of public transport routing as well.

The code is all at github.com/swbuchanan/melbourne-run-project.