Client: Bill Pink, Road Safety and Road User Management Directorate, Roads and Traffic Authority, New South Wales
The NSW Roads and Traffic Authority maintains an extensive database of motor vehicle accidents with approximately 50,000 accidents being recorded each year since 1992. This database represents the hub of the Authority's business operations for safety research and reporting. This database has been geo-coded against an ageing road centreline network.Spatial Vision was contracted to translate the accident location database of 450,000 points from their existing road centre line network based on Australian Geodetic Datum 1966 (AGD66), to a new version of road centre lines compliant with the Geocentric Datum of Australia 1994 (GDA94).
Spatial Vision developed a technique of relating the known intersections of the AGD66 and GDA94 road centreline datasets through the underlying road name attributes. A framework of related intersections was used as a spatially co-located linkage table to ‘rubber sheet’ the old points to the GDA94 road linework. The process to create logical consistencies at intersections involved a mixture of text strings of road names and spatial proximity tests.
The Roads and Traffic Authority now can use an up to date road centre line for any new accident location. Analysis of Black Spot sites can be made using historic information to understand trends in accident locations.
"Having made a substantial investment geocoding traffic accidents over a number of years, the necessity of moving the the new datum presented us with a dilemma. On the one hand, we could ensure near-100% accuracy by re-geocoding - however, this would be like starting from scratch and would duplicate the cost of the investment. At the other extreme we could simply use a script to transform to the datum & then snap to the nearest road - however, our analysis indicated that this would result in an unacceptably high error rate.
The solution offered by Spatial Vision has resolved the dilemma - acceptable accuracy at a fraction of the cost of re-geocoding. The technique of relating known intersections enabled us to segment the work and provide the most attention to major roads and more severe accident types.
We have been using the transformed accident set for several months now and it has met all our expectations."
– Bill Pink, Road & Traffic Authority
- Data modelling
Phill King designed the rule system architecture.
Amanda Tyrer developed the Editing system and supervised the transformation process.
The existing accident points are attached to the old road centre line network (grey lines).
The old road network is ‘rubber sheeted’ to conform with the new road network
The accident points are snapped to the new road base using a series of validation rules some examples being;
- Points set between the lanes of a new divided road
- Set point on known intersection.
- Flag for further investigation where a point occurs on a road that is not in the new road dataset.
- Microsoft Excel