August 2016 G-NAF. Data Release Report

Transcription

1 Product: Prepared: G-NAF Data Report

2 Revision History Date Version Change Coordinator 1.0 Initial Version Anthony Hesling Disclaimer PSMA Australia believes this publication to be correct at the time of printing and does not accept responsibility for any consequences arising from the use of information herein. Readers should rely on their own skill and judgement to apply information to particular issues. This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may be reproduced by any process without prior written permission of PSMA Australia Limited. Page i

3 Contents Introduction... 1 Document Purpose... 1 Audience... 1 Support... 1 Overview... 2 Updates and Issues... 3 Updates... 3 Issues... 9 Summary Statistics Future Considerations Annex A Detailed Statistics by Jurisdiction Page ii

4 Introduction G-NAF (Geocoded National Address File) is the authoritative physical address index for Australia. It contains the state, suburb, street, number and coordinate reference or geocode for street addresses in Australia. G-NAF uses existing and recognised address sources (referred to as contributors) including the state and territory government land records, as well as address data from the commonwealth government and other national sources. For more information please refer to the G-NAF Product Description. Document Purpose This document provides a description of the key issues encountered and production changes applied during the assembly of this update. In addition, overarching release statistics and general change volumes are also included. This document should be read in conjunction with the accompanying G-NAF Product Description. Audience This document is made available to PSMA staff, data contributors, direct clients and end users. It is intended to be used by business managers, data managers and analysts as a guide and reference to understanding the content and variations within a data release. Support PSMA Australia Support PSMA Australia Limited Page 1

5 Overview Nationally, this update of G-NAF shows an increase of 66,684 addresses overall which equates to an increase of 0.48%. The total number of addresses in G-NAF now stands at 14,015,754 of which 13,495,574 or 96.29% are principal. Confidence levels nationally show a slight improvement in confidence level 2 addresses which have increased by 44,576 and now represent 64.94% of principal addresses. Confidence 1 addresses remain unchanged since the May 2016 release at 14.90% while confidence 0 addresses have decreased by 0.2% to 20.16% of addresses. There was an increase of 77,573 address level, a drop of 13,272 in street locality addresses and an increase of 118 in locality level addresses. The reliability of principal addresses in G- NAF is now made up of 95.88% address level, 3.84% street locality addresses and 0.28% locality addresses. Updates An update to LEGAL_PARCEL_ID processing has resulted in a significant overall increase in population of this attribute across all states and territories. Please refer to the Updates and Issues section of this document for a more detailed explanation including examples of LEGAL_PARCEL_ID processing. In conjunction with the 2016 Census, the Australian Bureau of Statistics (ABS) has released a revision of the ASGS Volume 1 Main Structure and Greater Capital City Statistical Areas. The 2016 ABS data has been included in the G-NAF product in this release. The G-NAF data model has been updated to include two additional tables called: o ADDRESS_MESH_BLOCK_2016 and o MB_2016. Please refer to the Updates and Issues section of this document for a more detailed explanation of the data model changes. Issues There have been no significant production issues in the release of PSMA G-NAF Data. Page 2

6 Updates and Issues Updates Link between G-NAF address and Cadastre Background The LEGAL_PARCEL_ID field in the ADDRESS_DETAIL table is the link between the address and the cadastre. Since May 2012 this field had been populated using a point in polygon process on the Cadlite dataset from the previous release. (Prior to this, LEGAL_PARCEL_ID was only populated where it had been provided by the contributors as part of their address information). Whilst this revised process at the time significantly increased the number of addresses with a link to the cadastre, there remained many where addresses did not have a LEGAL_PARCEL_ID. For example, no secondary addresses were allocated a legal parcel value as these could not be derived through the point in polygon process. In the May 2016 release of G-NAF there were 9.3 million principal addresses with a LEGAL_PARCEL_ID and some 4.2 million addresses without. 86% of the addresses without LEGAL_PARCEL_IDs were secondary addresses, 13% were street-locality level geocoded addresses and 1% locality-only level geocoded addresses. New process In this release of G-NAF, a new process has been implemented which directly links addresses contributed by jurisdictions with an identifier to corresponding cadastral parcel information. The generation of this linkage is part of the PSMA G-NAF Live dataset processing and the results are now being transferred to G-NAF in conjunction with the quarterly maintenance process. As a result, the majority of secondary addresses now have a LEGAL_PARCEL_ID. In this release there are 12.5 million principal addresses with a LEGAL_PARCEL_ID, an increase of some 3.2 million addresses. However, 985,000 addresses remain in the release which do not have a LEGAL_PARCEL_ID. Whilst 58% of these addresses are street-locality and locality-only addresses, there remain some 484,000 parcel level addresses without a LEGAL_PARCEL_ID as the new process is dependent on addresses contributed by jurisdictions. As a result of this dependence, addresses provided by other contributors may not be allocated a LEGAL_PARCEL_ID. Future updates may further improve this process. Changes in Legal parcel ids between May 2016 and releases As well as allocating LEGAL_PARCEL_IDs to addresses that previously had no cadastral identifier, the new process has also resulted in changing the existing LEGAL_PARCEL_ID to some 680,000 addresses. Of these addresses, approximately 180,000 now have no LEGAL_PARCEL_ID whereas they were previously linked to a cadastral identifier. 62% of these are gap geocoded addresses and as such, cannot be confidently linked to a specific parcel apart from a point in polygon process where there is no certainly as to whether it lies within the correct parcel. The remaining addresses that have not retained a LEGAL_PARCEL_ID in the revised process are primarily addresses that are either, not contributed by a jurisdiction, contain a geocode that is co-located with multiple jurisdictional addresses or the LEGAL_PARCEL_ID cannot be allocated with any certainty. Further investigations will be undertaken to endeavour to resolve the LEGAL_PARCEL_IDs for these addresses in future releases. The remaining 500,000 addresses have been allocated updated LEGAL_PARCEL_IDs primarily as a result of one of the following reasons (The first two accounting for the majority): The previous identifier was that of the parent parcel and as such was incorrect The address relates to a multi parcel property and the LEGAL_PARCEL_ID now relates to another one of the parcels comprising the property. Page 3

7 The geocode for the address was outside of the parcel to which it related There are two occurrences of the same lot number in a street and the incorrect LEGAL_PARCEL_ID had been selected In Victoria, some 17,000 parent addresses (ie. common property) have been incorrectly linked to a lot number at the same site. These will be corrected in the November 2016 release. In NSW some 20,000 addresses have been incorrect allocated a LEGAL_PARCEL_ID of //. These will also be corrected in the November 2016 release. Examples of several of these updates are provided below. Changes in LEGAL_PARCEL_ID - ly the parent id had been assigned In this case the point in polygon process had previously assigned a LEGAL_PARCEL_ID of the parent parcel. These comprise 40% of the instances where the id has changed. Three examples are provided below. Page 4

8 Changes in LEGAL_PARCEL_ID - Multi parcel properties Addresses in many instances relate to properties that are comprised of multiple parcels. As such, these addresses are linked to multiple LEGAL_PARCEL_IDs. As the LEGAL_PARCEL_ID field can only accommodate one occurrence, the new process has randomly selected one. In many cases this is different to the identifier in the May 2016 release. In that release and those previously, no actual selection was necessary as the point in polygon process identified which parcel the address was located in. Two examples are provided below. Page 5

9 Changes in LEGAL_PARCEL_ID - Geocode outside the parcel to which it relates In a relatively small number of cases, the geocode supplied by the jurisdiction falls in the road casement. In these instances the identifier associated with the road was identified by the point in polygon process as the LEGAL_PARCEL_ID. This has been corrected in this release as shown by the example below. Changes in LEGAL_PARCEL_ID - Duplicated lot numbers in same Street Changes in LEGAL_PARCEL_ID - Common property now lot number Some 17,000 Victorian parent parcel addresses associated with strata developments (ie. common properties) have been allocated the incorrect LEGAL_PARCEL_ID. In the May 2016 release these addresses had been linked to the common property cadastral identifier (eg. CM/PS310693), however, in this release, these addresses have been linked to one of the unit numbers in the development. This will be corrected in the next (November 2016) release. Page 6

10 Statistics relating to LEGAL_PARCEL_ID The tables below compare the number of principal addresses with a LEGAL_PARCEL_ID between the May 2016 and releases. Page 7

11 ABS 2016 Mesh Block 2016 Update In conjunction with the 2016 Census, the Australian Bureau of Statistics (ABS) has released a revision of the ASGS Volume 1 Main Structure and Greater Capital City Statistical Areas. The 2016 ABS data has been included in the G-NAF product in this release. The G-NAF data model has been updated to include two additional tables called: ADDRESS_MESH_BLOCK_2016, and MB_2016. See the relevant section of the new model below: dm G-NAF Data Model - page 2 G-NAF Data Model - Page 2 Legend Textual Attribute Table Spatial Attribute Table Authority Code Table MB_MATCH_CODE_AUT *PK CODE: VARCHAR2(15) * NAME: VARCHAR2(100) DESCRIPTION: VARCHAR2(250) 1 1 Note: Spatial Attribute Tables contain a GEOMETRY field that can be derived from the LATITUDE/LONGITUDE, but is not included in the product. MB_2011 *PK MB_2011_PID: VARCHAR2(15) * DATE_CREATED: DATE DATE_RETIRED: DATE * MB_2011_CODE: VARCHAR2(15) 1 MB_2016 *PK MB_2016_PID: VARCHAR2(15) * DATE_CREATED: DATE DATE_RETIRED: DATE * MB_2016_CODE: VARCHAR2(15) 1 0..* ADDRESS_MESH_BLOCK_2011 *PK ADDRESS_MESH_BLOCK_2011_PID: VARCHAR2(15) * DATE_CREATED: DATE 0..* DATE_RETIRED: DATE *FK ADDRESS_DETAIL_PID: VARCHAR2(15) *FK MB_2011_PID: VARCHAR2(15) FK MB_MATCH_CODE: VARCHAR2(15) 0..* 0..* ADDRESS_MESH_BLOCK_2016 *PK ADDRESS_MESH_BLOCK_2016_PID: VARCHAR2(15) 0..* * DATE_CREATED: DATE DATE_RETIRED: DATE * ADDRESS_DETAIL_PID: VARCHAR2(15) *FK MB_2016_PID: VARCHAR2(15) FK MB_MATCH_CODE: VARCHAR2(15) 0..* ADDRESS_DETAIL ADDRESS_DETAIL 1 *PK ADDRESS_DETAIL_PID: see page 1 VARCHAR2(15) 1 A * 0..* PRIMARY_SECONDARY *PK PRIMARY_SECONDARY_PID: VARCHAR2(15) * DATE_CREATED: DATE DATE_RETIRED: DATE *FK PRIMARY_PID: VARCHAR2(15) *FK SECONDARY_PID: VARCHAR2(15) *FK PS_JOIN_TYPE_CODE: NUMBER(2) PS_JOIN_COMMENT: VARCHAR2(500) PS_JOIN_TYPE_AUT *PK CODE: NUMBER(2) 0..* 1 * NAME: VARCHAR2(50) DESCRIPTION: VARCHAR2(500) Page 8

12 Issues There were no significant issues in this release. Page 9

13 Summary Statistics Note: The statistics below do not include retired addresses (i.e. Confidence =-1). Geocode Reliability of Principal Addresses Only Reliability No of Addresses Percentage Change since May NO GEOCODE % SURVEYING STANDARD % WITHIN ADDRESS SITE BOUNDARY OR ACCESS POINT 12,782, % 76, NEAR (OR POSSIBLY WITHIN) ADDRESS SITE BOUNDARY 156, % UNIQUE ROAD FEATURE 517, % -13, UNIQUE LOCALITY OR NEIGHBOURHOOD 38, % 118 TOTAL 13,495, % 64,419 Address Confidence Levels of Principal Addresses Only Confidence level Count Percentage of Total Change Since May 2016 Addresses with a Confidence of 0 2,720, % 9,230 Addresses with a Confidence of 1 2,011, % 10,613 Addresses with a Confidence of 2 8,763, % 44,576 TOTAL 13,495, % 64,419 Page 10

14 Geocode Reliability of Addresses (Principals and Aliases) Reliability No of Addresses Percentage Change since May NO GEOCODE % SURVEYING STANDARD % WITHIN ADDRESS SITE BOUNDARY OR ACCESS POINT 13,297, % 78, NEAR (OR POSSIBLY WITHIN) ADDRESS SITE BOUNDARY 158, % 1, UNIQUE ROAD FEATURE 521, % -13, UNIQUE LOCALITY OR NEIGHBOURHOOD 38, % 115 TOTAL 14,015, % 66,684 Address Confidence Levels of Addresses (Principals and Aliases) Confidence level Count Percentage of Total Change Since May 2016 Addresses with a Confidence of 0 3,078, % 10,225 Addresses with a Confidence of 1 2,169, % 11,645 Addresses with a Confidence of 2 8,768, % 44,814 TOTAL 14,015, % 66,684 Page 11

15 Future Considerations As identified earlier in this report, further investigations will be undertaken to endeavour to resolve LEGAL_PARCEL_IDs for addresses in the November For example: In Victoria, some 17,000 parent addresses (i.e. common property) have been incorrectly linked to a lot number at the same site. In NSW some 20,000 addresses have been incorrectly allocated a LEGAL_PARCEL_ID of //. Page 12

16 Annex A Detailed Statistics by Jurisdiction Victoria Victorian addresses represent 24.91% of the national total. Victoria has had an increase of 25,845 addresses for this release. Address reliability shows improvement in Victoria however confidence 2 address matching has fallen away slightly with both confidence 1 and confidence 0 addresses increasing slightly. Confidence - Vic No of Addresses % Addresses with a Confidence of 0 546, % 7, % Addresses with a Confidence of 1 513, % 6, % Addresses with a Confidence of 2 2,300, % 11, % Level Geocoded - Vic No of Addresses % Addresses geocoded to Property level 3,299, % 26, % Addresses geocoded to Street level 57, % % Addresses geocoded to Locality level 4, % % New South Wales New South Wales addresses represent 30.84% of the national total. There was an increase of 16,561 principal addresses for NSW. Address confidence and address reliability both show improvement for the August release in NSW. Confidence - NSW No of Addresses % Addresses with a Confidence of 0 827, % 1, % Addresses with a Confidence of 1 664, % 3, % Addresses with a Confidence of 2 2,670, % 12, % Level Geocoded - NSW No of Addresses % Addresses geocoded to Property level 3,949, % 24, % Addresses geocoded to Street level 197, % -8, % Addresses geocoded to Locality level 15, % % Page 13

17 South Australia South Australian addresses represent 7.85% of the national total. South Australia has had an increase of some 4,908 principal addresses. Address confidence for South Australia shows a slight improvement while address reliability shows a slight swing from property to street addresses. Confidence - SA No of Addresses % Addresses with a Confidence of 0 264, % % Addresses with a Confidence of 1 140, % % Addresses with a Confidence of 2 653, % 5, % Level Geocoded - SA No of Addresses % Addresses geocoded to Property level 1,009, % 4, % Addresses geocoded to Street level 46, % % Addresses geocoded to Locality level 3, % % Tasmania Tasmanian addresses represent 2.40% of the national total. There was a small increase of 1,011 principal addresses for Tasmania which continues to show improving trends for both reliability and address confidence for the August release. Confidence - TAS No of Addresses % Addresses with a Confidence of 0 59, % % Addresses with a Confidence of 1 48, % % Addresses with a Confidence of 2 215, % % Level Geocoded - TAS No of Addresses % Addresses geocoded to Property level 304, % 1, % Addresses geocoded to Street level 18, % % Addresses geocoded to Locality level % % Page 14

18 Western Australia Western Australian addresses represent 10.42% of the national total. Western Australia has had an increase of some 8,401 principal addresses. Both confidence and address reliability show a slight improvement for the release. Confidence - WA No of Addresses % Addresses with a Confidence of 0 250, % 1, % Addresses with a Confidence of 1 179, % 1, % Addresses with a Confidence of 2 976, % 6, % Level Geocoded - WA No of Addresses % Addresses geocoded to Property level 1,354, % 8, % Addresses geocoded to Street level 48, % % Addresses geocoded to Locality level 4, % % Australian Capital Territory ACT addresses represent 1.55% of the national total. There was an increase of 1,896 principal addresses in ACT. While confidence 2 address numbers have increased by 702, an increase of 1,341 confidence 0 addresses has caused a drop in the percentage of confidence 2 addresses. The reliability of ACT addresses shows an improving trend over the May release. Confidence - ACT No of Addresses % Addresses with a Confidence of 0 30, % 1, % Addresses with a Confidence of 1 16, % % Addresses with a Confidence of 2 162, % % Level Geocoded - ACT No of Addresses % Addresses geocoded to Property level 207, % 2, % Addresses geocoded to Street level 2, % % Addresses geocoded to Locality level % % Page 15

19 Queensland Queensland addresses represent 21.23% of the national total. There was an increase of some 5,380 addresses in Queensland. Both address and confidence match rates show a slight improvement for Queensland for the release. Confidence - QLD No of Addresses % Addresses with a Confidence of 0 711, % -2, % Addresses with a Confidence of 1 424, % % Addresses with a Confidence of 2 1,729, % 6, % Level Geocoded - QLD No of Addresses % Addresses geocoded to Property level 2,715, % 10, % Addresses geocoded to Street level 140, % -4, % Addresses geocoded to Locality level 8, % % Northern Territory Northern Territory addresses represent 0.77% of the national total. Northern Territory had a small increase of 419 addresses. Address confidence continues to show improvement while address reliability has fallen slightly in the August release. Confidence - NT No of Addresses % Addresses with a Confidence of 0 26, % % Addresses with a Confidence of 1 23, % % Addresses with a Confidence of 2 54, % % Level Geocoded - NT No of Addresses % Addresses geocoded to Property level 97, % % Addresses geocoded to Street level 6, % % Addresses geocoded to Locality level % % Page 16

20 Other Territories Other territories addresses represent 0.02% of the national total. Overall there was a drop of 2 addresses for OT with a slight improvement in both the confidence and reliability for OT in the August release. Confidence - OT No of Addresses % Addresses with a Confidence of 0 2, % % Addresses with a Confidence of % % Addresses with a Confidence of % % Level Geocoded - OT No of Addresses % Addresses geocoded to Property level 2, % % Addresses geocoded to Street level % % Addresses geocoded to Locality level % % Page 17