Australia

What the record shows

Australia's largest contribution to this atlas is a system that mostly runs somewhere else. SCATS was established by the New South Wales roads department in 1975, one of the first commercially viable adaptive signal systems, and a Transport for NSW team still runs it — now across more than 63,000 intersections in 216 cities and 32 countries, of which roughly 11,000 are in Australia. Half a century on it is a national export with an inconvenient footnote: the widely quoted gains of 28 per cent shorter travel times, 25 per cent fewer stops and 12 per cent less fuel are the operator's own averages across all those deployments, not an independent measurement in any one city. Half the world's adaptive-signal claims trace back to that sentence.

The Australian habit that deserves more copying is unglamorous: instruments that are still working a decade later. Melbourne buried sensors under roughly 5,300 parking bays starting in 2011 and more than 4,600 still feed the open data portal and third-party apps — a first-generation municipal sensor rollout that outlived nearly all of its international contemporaries, which were dug up as maintenance costs outran the benefit. The same city has counted footfall on infrared sensors since 2009, now at 99 locations, and publishes the counts hourly. That feed has been used to track the city's pandemic recovery, test pedestrian-safety correlations and nowcast state retail sales.

It also produced the single most transferable methodological finding in the country's file, and it is a correction to itself. Melbourne's 2025 total of 312.9 million pedestrian movements looked like 7.7 per cent growth on the previous year. Comparing only the sensors present in both years, the increase was 1.2 per cent. The rest was the network getting bigger. Any city publishing year-on-year totals from an expanding sensor estate is quietly reporting its own procurement as public behaviour, and this is the entry to hand them.

A second self-inflicted caveat sits in the open data portal, and it is worth stating because it inverts the usual licence complaint. Melbourne licenses 226 of its 239 datasets under CC BY 4.0 — 94.6 per cent, at the tidy end of this atlas's licence ladder. The thirteen with no licence recorded include the pedestrian counting feed, which is the city's most reused dataset and underpins at least three peer-reviewed studies. The most legally uncertain file in the catalogue is the one academics have already built on.

Australia is also the cleanest place in the atlas to see that automatic train operation and driverless operation are different purchases, because it has both. Sydney Metro's Northwest line opened on 26 May 2019 as genuinely unattended rapid transit — no driver, platform screen doors at every station, supervision from a control centre, services as often as every four minutes at peak — and extended into the CBD in August 2024. Melbourne's Metro Tunnel, opened on 30 November 2025 with Australia's first brownfield moving-block signalling, is marketed as high-capacity signalling and keeps a driver in the cab for doors, supervision and emergencies. Both are correct engineering; only one of them removes the person. Melbourne's capacity ceiling of 24 trains an hour is a signalling figure that arrives when every train carries the equipment, against about 18 at opening.

Ownership is the other recurring question, and Adelaide answers it plainly. The council named a telecoms provider its official network partner in December 2017 and subsidised a fibre build that reached 82 kilometres and 1,000 buildings by October 2020, with businesses paying nothing for the in-building connection. The provider built, owns and operates the network and sells over it. A city can pay for the pipes and still not own them — the opposite of the municipal dark-fibre model, and a decision that cannot be revisited once the build is done. No uptake figure has been published since completion.

Water is where Australian ambition stays deliberately provisional. Perth's utility has been running 16,000 smart meters across ten districts since 2023 and calls it a trial, not a rollout. Sydney Water installed 7,500 across four suburbs, reported 313 leak alerts and an average bill reduction of $70, and plans to replace 1.6 million mechanical meters — but as of early 2025 was still running a market sounding to decide how to contract the work. Both are honest about the gap between a pilot that works and a programme that exists.

Two entries show the limits arriving from directions nobody specified. Melbourne replaced CBD litter bins with solar-powered compacting units — 397 of them displacing 772 old bins in 2018 — and cut public-bin collections from about 93,000 a month to 14,500, most of them moved overnight. Then the shaded canyons between towers stopped keeping the batteries charged, disabling compaction and fill alerts, and the bins had to be moved to sunnier spots. Solar power is not location-agnostic. And the shared e-scooter trial produced opposite verdicts a few kilometres apart: after a peer-reviewed trauma study counted 256 hospital presentations in the first year at just under AU$2 million in direct medical cost, the City of Melbourne voted 6–4 in August 2024 to end its contracts, while neighbouring councils kept theirs and the state legalised the vehicles permanently months later.

The rest of the record is patient and modest about itself. Microclimate poles have published fifteen-minute readings since November 2019 from a rotating estate where only about six of sixteen sites are live at any time — six years of usable data from a network that never stopped moving. The heat-risk pilot's own evaluation lists what it could not do: vulnerability data resolving only to statistical areas, a pool operator with no system access, staff without time to use the tool. More than seventy thousand street trees carry individual IDs and email addresses. And Sydney's air has been monitored since 1951, which makes it the oldest measurement in this atlas — and when the reading is rated poor, the documented response is an alert and health advice. The longest measurement series in this atlas, and the consequence at the end of it is still a notification.

Sources beyond the atlas

Every factual claim above is carried by one of the linked atlas entries; no external source was needed for this section.