The 2006 Australian national census will be among the many beneficiaries of a new government database that helps translate addresses into geocoded locations.
The Australian Bureau of Statistics (ABS) is the first to license the Geocoded National Address File (G-NAF), a database of the exact locations of 9.5 million addresses across Australia.
Geocoded addresses are an index of locality, street number and co-ordinates of longitude and latitude. Owner/resident names are not used for privacy reasons.
The project was officially completed last week after eight years of development and over A$2 million (US$1.52 million) in expenditure.
Led by government-owned company Public Sector Mapping Agencies (PSMA), the G-NAF project cross-referenced 32 million addresses from 15 government databases. This eliminated inaccurate records, resulting in the 9.5 million geocoded addresses.
The database will be available to both government and (via resellers) commercial organizations, but will first be used by G-NAF contributors such as ABS, the Australian Electoral Commission and Australia Post.
For the ABS, being able to relate addresses to statistical geography means new types of statistics, according to director of geography Frank Blanchfield.
Previously the department could only monitor areas where it was certain of the addresses.
“We’ll be able to expand the range of areas we can output statistics for,” said Blanchfield.
“Also there’ll be new sources of statistics, it’ll give us more flexible geographical outputs.”
The ABS will create one or many Web services by July next year to help government agencies code their own administrative (address) data into geographical locations, according to Blanchfield.
“They can stream addresses to the Web service, which will probably be via XML, and get back the number of the mesh block they’re in,” he said.
The service would likely be only for small government agencies.
“The Centrelinks of this world would have G-NAF incorporated into their own infrastructure,” said Blanchfield.
The ‘mesh block’, made possible by G-NAF, will be a new geographical unit for the next Census.
The current smallest unit for ABS statistics is the Census Collection District (CD), but a mesh block “is about a fifth of the size of a CD unit,” said Blanchfield.
“People have been looking for statistics for catchment areas, electorates…but mesh blocks will give us new statistical locations. We’ll be coding all the data from the Census to mesh blocks,” he said.
The CEO of PSMA Dan Paull said G-NAF was “a milestone” as an independent mechanism to verify addresses.
“(For example) Australia Post only knows addresses through what people write on letters.
“We can provide (via G-NAF) a view of both the address and location.”
G-NAF could also be used in other ways with existing mapping datasets, he said.
“You can use PSMA’s other datasets such as the road network which could be overlaid on G-NAF and query the shortest route between two points.
“So you’ll see the textual address and a link to latitude and longitude. They appear as dots, but it’s true distance.”
This has attracted interest from state emergency services, he said. However, there were still many G-NAF maintenance procedures to setup, said Paull.
The first build of the 10G bytes database took nine months, and as it’s distributed on DVD, will only be updated once or twice this year.
“At the moment PSMA relies on other land registries, Australia Post to inform us of changes.
“We want to move to real-time eventually,” said Paull.
This will be done by linking all the databases of all contributing government departments.
“Currently G-NAF’s in an Oracle database, but we’re aiming to move it to a data warehouse by the end of 2005. We’ll use the same software, but we’re working to establish protocols to transact over the Net,” he said.
