Geocoding with ESRI Technology
ArcGIS has an in-built geocoding capability that can
be customised to suit address structures and
reference databases from any country. In brief, a
user needs a reference database of addresses to
match against and a set of geocoder definitions or
rules that define how an input address is parsed and
then subsequently matched against the reference
databases.
Reference Data
NAVIGATE uses the G-NAF (Geographic National Address
File) product from PSMA Australia as the source for
the reference databases. The G-NAF contains approx
13 million addresses and geocodes and is delivered
to NAVIGATE as a highly normalised data model that
includes a number of data tables and a series of
associated look-up tables.
This structure is not suitable for immediate use
with the ArcGIS Geocoder as reference data. NAVIGATE
has taken the complex structure of the G-NAF and
generated single, flat tables that contain all the
necessary details for geocoding and is in a suitable
format for use as reference data with the ArcGIS
geocoder. This table is derived by querying the
numerous G-NAF tables and adding a point geometry
from the longitude and latitude supplied on the G-NAF
record.
NAVIGATE Rule Base Customisations
NAVIGATE has reworked the out of the box rule base
definitions (designed for North American addressing
conventions and syntax) using the ESRI Geocoder
Development Toolkit. We now publish a series of
Australian rules and definitions (called locators)
for the ESRI ArcGIS Geocoder which refer to the G-NAF
reference data.
Our Australian address pattern and locator files
result is up to 15% - 20% improvement in address
parsing and subsequent matching to the reference
data. Examples of improvements using GDK Customised
to Australian address syntax include:
·
2/85 George St (2 George St with standard address
parser)
·
10-12 Clarke St (1012 Clarke St with standard
address parser)
·
St Kilda Rd (Saint Kilda Rd with standard address
parser)
·
10 West End Pde (10 W End Parade with standard
address parser)
·
Support for "The Corso," "The Avenue," Etc..
Geocoding Product
Structure
The NAVIGATE Geocoding kit for feature based
geodatabases or SDE consists of
a)
a set of three feature datasets, each holding one
feature class of reference data, indexes and
internal geocoding rules base (termed a locator).
a.
Feature class 1 is a set of geocoded street address
and suburbs
b.
Feature class 2 is a set of geocoded street
addresses with an alternative suburb name
c.
Feature class 3 is a set of geocoded street segment
centrepoints (within a suburb)
b)
an external file or locator that calls these feature
datasets in sequence.
ArcGIS reads the input addresses, parses the
structure as per the rules bases and then attempts
to find a matching address in the first feature
dataset. If the address cannot be found there the
process cascades to the second reference set and
then the third.
Geocoding in ArcGIS with the NAVIGATE G-NAF
Geocoding Kit
Schematic Diagram
