Victorian intersections where two different roads meet at less than 75°, measured on Vicmap Transport road centrelines. Coloured by the land tenure around each site, with recorded injury crashes from 2020 to 2024.
How this was made
- Network: Vicmap Transport TR_ROAD (DTP / DEECA, CC BY 4.0), operational segments of road, bridge, tunnel, ford, roundabout and connector types; road classes 0–5, plus class 6 where the road is named. Trails, paths, 4WD and seasonal tracks are excluded.
- Intersection: a network node with at least three legs and at least two different road names. Unnamed legs count as separate roads.
- Angle: each leg's direction is taken from the node to a point 20 m along it. Legs are sorted by direction and the gap between each neighbouring pair of different roads is measured; the site's angle is the smallest gap. A 40 m reading is shown as a check. Sites within 25 m of each other (for example both carriageways of a divided road) are merged, keeping the smallest angle.
- Land tenure: Victorian Land Use Information System 2021/22 release 3 (DEECA, CC BY 4.0). Within 30 m of each site, the parcel area is split into Crown tenure (codes 2xx: conservation reserve, public forest, other Crown purposes) and Freehold (code 100). A site is "Crown / public" when Crown land is at least half of that area. Road reserve itself has no tenure in VLUIS and is left out.
- Crashes: Victoria Road Crash Data (DTP, CC BY 4.0), police-reported injury crashes from 1 January 2020 to 31 December 2024, grouped by the crash's severity (Fatal, Serious Injury, Other Injury). Each crash is spatially matched to the nearest intersection node within 30 m; crashes near closely spaced nodes may be matched to a neighbouring one. Victoria records injury crashes only, so a site with no recorded crash may still have had non-injury collisions.
Counts are not risk. Crash counts are not adjusted for traffic volume, so busier intersections will show more crashes. This page does not compare skewed and right-angle intersections. In our own check, a pooled comparison changed direction once it was split by road class and number of legs, so it was confounded and does not support any conclusion about angle and crashes.
Read with care. This is a desktop screening layer from a 1:25,000 centreline network, not a surveyed geometry or engineering assessment. "Freehold" does not mean privately owned: councils and agencies also hold freehold land. Tenure describes the land around the site, not who manages the road.
Data extracted 2 October 2026. Analysis and map by Sabour Khosravi, GoldNet Group.