AUSTRALIA COMES TO IT’S CENSUS

The Public Sector Mapping Agencies and the 1996 Australian Census of Population and Housing

John D. Mobbs
Public Sector Mapping Agencies
c/o Land Information Centre
Bathurst NSW 2795
Ph: +61 02 6332 8317
Fax: +61 02 6332 8266
E-mail: jdmpsma@netinfo.com.au

Australia regards itself as a member of the developed world. The actual degree of development and need for planning to support future development is gauged, in part, by the results of the five-yearly census undertaken by the Australian Bureau of Statistics (ABS). The Census of Population and Housing conducted on the 6th August 1996 was planned, mapped and executed with the benefit of a new national digital topographic and cadastral map database. This is the most detailed, accurate and comprehensive multi-resolution spatial database available in Australia.

The Public Sector Mapping Agencies (PSMA) consortium consists of State, Territory and Commonwealth mapping agencies. It was contracted by the ABS to capture, integrate and deliver the digital map data in conjunction with a private sector consortium which provided cartographic enhancement, system definition and customised software and hardware.

A further objective of providing Australia with a national topographic dataset useable for other purposes has been achieved and the sense of using a cooperative approach has been realised. A version of the PSMA dataset has since been simplified and reduced in volume to support the display of 1996 Census statistics on a CD-ROM product released by the ABS in 1997. Other redevelopments of the database have also taken place.

This paper reviews some historical approaches to census taking in Australia and provides a general outline of the PSMA project. Advantages, mapping challenges and utility of the database are identified. Political, economic and jurisdictional obstacles are discussed and the importance of this project in accelerating progress towards an Australian Spatial Data Infrastructure is explained.

The transportation of convicted felons from England to a distant penal colony in Australia was a political, social and scientific experiment designed to ascertain if a more hospitable climate and fertile soils might provide for their sustenance and redemption. It was mooted following the 1776 American Declaration of Independence and made urgent by the accumulation of convicts in England that could no longer be transported to that colony (Knibbs, 1914).

Colonisation of Australia showcased the crucial role of the surveyor, the mapmaker and the statistician from the time the First Fleet of 11 ships arrived with some 1100 souls (some 700 convicts), at the recommended site of Botany Bay on 20th January 1788. Within days, Captain Arthur Phillip surveyed and selected Port Jackson, several kilometres to the north, as a better site for the first permanent settlement of the colony of New South Wales (NSW), which at that time extended east of the 135th meridian, i.e. about half of the continent. Had Captain James Cook been less diligent in his charting of the Eastern coast of Australia eighteen years earlier and had he not specifically noted and named Port Jackson, the colony might have had an entirely different beginning.

As in any scientific experiment, raw materials and parameters had to be controlled and monitored and the economics of the model observed. Prior to embarkation Phillip was provided with instructions requiring him to:

".....use every proper degree of economy, and be careful that the Commissary so transmit an account of the issues from time to time to the Commissioners of our Treasury, to enable them to judge of the propriety or expediency of granting further supplies. The clothing of the convicts and the provisions issued to them and the civil and military establishments, must be accounted for in the same manner"

"....an account of the numbers inhabiting the neighbourhood of the intended settlement" and in relation to grants of land to emancipated convicts, to "cause copies of such grants as may be passed to be preserved and make regular return of the said grants" (Historical Records of Australia, cited in Forster & Hazlehurst, 1988).

Phillip was industrious, since by 9 July 1788, in his fourth dispatch, he included an account of population numbers, tables of livestock, a general return on the marines plus a return on the sick and dead since the landing. In July 1790 a detailed and categorised return of the colony’s population was sent. Phillip established as directed, within a month of founding Sydney, a convict settlement on Norfolk Island, some 900 miles north-east (Knibbs, 1914). Details of 87 land grants in New South Wales and on Norfolk Island were returned in November 1791, with the owner’s status, marital situation and date of settlement recorded, plus the size, area and location of grant (HRA in Forster & Hazlehurst, 1988). Those land grants, within 34 months of settlement must surely have been surveyed by the Surveyor of Lands (later Surveyor-General) Augustus Alt who arrived with the First Fleet (Lines, 1992).

Governor Arthur Phillip clearly lodged the first Australian records of a cadastral nature, and performed the role of first Australian statistician. Charts prepared by Captain James Cook were the key to successful location and settlement of the colony and must have been scrutinised on many occasions in London as Phillip’s first colonial returns were analysed. Colonial Governors not only had the duty of reporting on the state of the colony; they had to administer a large and growing gaol in a wilderness, develop a system of free settlement and manage and maintain a convict labour force. Principal statistics required related to Population, the Commissariat, Births and Deaths and Agriculture (Forster & Hazlehurst, 1988).

It can be seen that the professions of surveying, mapping (charting) and statistics were linked at the earliest stages of the developing world in the Australian region. Much later a more demonstrable link was forged when in 1906, George Handley Knibbs, formerly a surveyor and lecturer in the engineering school at Sydney University, President of the Institution of Surveyors 1892-93 and 1900-01, honorary secretary of the Royal Society of NSW for nine years and president in 1898-99, was appointed as the first Commonwealth Statistician. Knibbs served in that capacity until 1921, creating considerable change, innovation, jealousy and controversy as he tightened the Commonwealth’s grip on its aim of becoming the principal statistical collection agency. Notably, Knibbs initiated the first Commonwealth Year Book (8,000 copies) in 1908, from States’ data and which attracted wide acclaim (Forster & Hazlehurst, 1988).

Between 1788 and 1813 the concentration of settlers on the coastal plain around Sydney enabled the population to be counted by means of "musters" or assemblies, proclaimed on certain dates at certain settlements. A "carrot and stick" approach seems to have found early favour, since the incentive of victualling plus the threat of incarceration or punishment were both used. Musters were initially conducted over several days, to allow travel time for the Commissary. However, people of sly persuasion also put the travel time to good use, forcing Governor Hunter in 1796 to declare simultaneous musters, since the previous system:

"...gave good time for imposters and other villains to practise their tricks and ingenuity by answering the first call at Sydney, where they have receiv’d provisions and slops as one resident in that district; on the day of call at Parramatta they have again appear’d there, have been enter’d in the muster list of that place, and have been again victual’d and sometimes cloathed; the attempt has sometimes been made (and not always successfully) at the third muster" (HRA in Forster & Hazlehurst, 1988).

Australia did, in a sense, come to its census at this time. However, once the route across the Blue Mountains was established and inland settlement commenced, the muster became more difficult and it soon became evident that the census would have to go to Australia. In 1809 the muster consumed a fortnight, with different classes of people attending on different days but by 1825 there were twenty muster stations (Coghlan, 1894 in Forster & Hazlehurst, 1988).

The gradual dispersal of settlers across NSW necessitated changes to census procedures. By 1828 the muster had been abandoned and people were thereafter counted at their usual place of residence. Some aspects and challenges of the method are echoed in present procedures.

Responsible persons chosen from the police, magisterial or teaching professions were enlisted and paid to distribute printed forms to householders. Where the householder could not read or write, the collector completed the form (Govt Notice 1828 in Forster & Hazlehurst, 1988). This practice may be regarded as the precursor to similar modern census collection practices used in many developed countries, including Australia. Field Managers are often recruited from the professions and collectors with special cultural affinities are appointed to assist ethnic, indigenous, and homeless persons.

At the beginning of the 1850s and leading up to the granting of self government to NSW, Victoria, Tasmania and South Australia in 1855-56, the colonies had a total population of around 400,000, with a quinquennial census established as the usual means of monitoring it. With the granting of self government the obligation to submit Imperial statistics for these colonies ceased. However, a succession of British-trained statisticians, appointed as Registrars-General, continued to oversee the census. The British census date of 7 April 1861 was also observed in Australia when Victoria, NSW, Tasmania and South Australia each conducted censuses (Forster & Hazlehurst, 1988b). There are precursors to modern day census methods to be observed in the conduct of the 1861 Censuses. It is instructive to examine some aspects in more detail.

By 1861 the difficulty of locating the population of NSW in an area which was "more than four times the area of England, Wales, and Scotland united" (Rolleston, 1861) was apparent. Reporting the results of the 1861 Census to Parliament on 30 June 1862, Mr Chris Rolleston, the NSW Registrar General, noted that:

"The District Registrars were, with one exception, appointed Enumerators for their respective districts, and upon them devolved the duty of selecting the Collectors employed on the enumeration - a duty which, considering the difficulties attending the work in a widely scattered and sparsely populated country, with small choice of qualified persons, was very satisfactorily discharged." and;

"Seven days were allowed for the distribution of the house-holders schedules previous to the 7th April, and seven days for their collection afterwards. In many of the widely scattered districts of the interior, however, it was found necessary to extend the period of collection beyond the time specified; and it should be noted that the flooded state of the country at the time is estimated to have added fully one-third to the cost of the service." and;

"There were employed on the work, 80 Enumerators and 464 Collectors - together, 544 persons." and "The total cost of collection was - ..........£ 10,087 18 6 " (Rolleston, 1861).

Rolleston also noted that "The work has been arranged according to the several divisions of the Colony and is classified into five series of districts" (Rolleston, 1861). The results had, in effect, been spatially cross-referenced.

NSW was divided into 20 counties and 13 pastoral districts, some of which extended into parts of what is now Queensland. Boundaries had been determined for these, plus 78 Census Districts (corresponding to Registry Districts), 62 Police Districts (identical to or multiples of Registry Districts), 57 Electorates, and the 194 Municipalities, Parishes, Towns and Villages of the County of Cumberland.

Maps must have been a crucial component of the census. Lines (1992) indicates that detailed maps of this era were mainly those resulting from the trigonometric surveys commenced around 1827 by Major Thomas Mitchell (Surveyor General 1828-55). These included topographic maps at the scale of 8.6 miles to the inch (approx. 1:540000). Other maps at scales as large as 1:21120 plus plans of subdivisions, were produced by private land companies using their own Surveyors when government mapping was non-existent. The indexes of land grants would also have served as rough maps to indicate settlement locations.

The importance of preparing demographics especially for urban populations also seems to have been emphasised, since separate tables were prepared for Cumberland County in the 1861 Census results. 44% of the Colony’s total population of 350,860 lived in Cumberland County, within about 150 miles of Sydney, in 24 Municipalities (further categorised as Urban, Suburban or Rural), 57 parishes and 113 towns or villages. Population density varied enormously, from 86 persons per square mile in Sydney, to 1 person per four square miles in pastoral districts. Census results were categorised by Sex, Age, Conjugal Condition, Occupation, Native Country, Religion and Education (Rolleston, 1861).

The Victorian Registrar General, Mr W. H. Archer, in his Letter of Instructions to Enumerators for the 1861 Census, left little to chance with regard to the importance of maps and boundaries. He stated that;

"It is desirable, when it can be done consistently with apportioning a fair quantity of work to each sub-enumerator, that the limits of the sub-districts should coincide with boundaries already known, such as those of a parish, township, electoral district, & c., and should not be made up of parts of two or more.

It is essential to the accuracy of the enumeration, that the sub-districts be well defined and clearly described, so that no doubt may exist among the sub-enumerators to which district any one house may belong, and that there may be no chance of the inmates being, through any such doubt, either omitted or numbered twice over; I would therefore impress upon you that this can only be done effectually by adopting natural boundaries, such as streets, fenced in or well- beaten roads, well-known creeks, the water flow of gullies, &c. Imaginary lines drawn from point to point on map are, except in rare cases inadmissible."(Archer, 1861).

We find this emphasis echoed in the 1996 instructions to census design staff which state, inter-alia:

"... boundaries should, if possible, be readily identifiable on the ground, be described in terms of permanent features, follow the centreline of a road or river if these features are used and should delimit Census Districts which conform to existing or proposed land uses..."

In 1861, maps were evidently obtained by the Registrar beforehand and sent to Enumerators:

"Having arranged the formation of the sub-districts as speedily as you can, in conformity with the following instructions, you must before the (date inserted) return to this office one of the two maps sent, with the plan of subdivision on it, reserving the other for your own use..."

Land Surveyors were ranked fourth in the occupation hierarchy, below "Government Service"; "Learned Professions"; and "Teachers". They were categorised within "Other Educated Professions (British Class IV, sub class 1,2,3)" and were thus counted with Artists, Authors, Botanists, Editors, Reporters and Sculptors, to name but a few of their peers in this grouping, which comprised only 0.15% of the population (Archer, 1861). If problems arose with census boundary definition the Surveyor was consulted:

"If any serious doubts should occur to you as to the best mode of dividing your district, it is hereby recommended that you should communicate without delay to the district surveyor, warden, Chinese protector, or any other of the officers of the government, who may be possessed of special information calculated to aid you, and who will doubtless give you any advice"(Archer, 1861).

In 1861, Enumerators or sub-enumerators divulging the contents of census forms could be fined 20 pounds. Even today, the confidentiality of census forms is stringently guarded, with any collector divulging the contents of a form being liable to penalties of a $5,000 fine, a gaol sentence of up to 2 years or both!

All British colonies were required to compile and return statistics. 1822 saw the introduction of the so-called Blue Book to Australia for this purpose. This book, despatched empty from London was to be filled in, signed by the Governor, duplicated by hand and returned annually. The 1822 version contained 154 pages with standard data categories of "Abstract of the Nett Revenue and Expenditure"; "Schedule of Taxes, Duties, etc"; "Military Expenditure"; "Establishment"; and "Schedule of the Fees, etc." Added to these in 1823 were the additional categories of "Population"; "Export and Imports"; and "Currency" (Colonial Office in Forster & Hazlehurst, 1988). - key parameters indeed, for an experiment so far from the main laboratory!

Most of the analysis of Imperial censuses seems to have been conducted in London, since the Blue Books had been standardised across the colonies to facilitate consolidated reports.

Rolleston and Archer both provided their own analysis with the 1861 Census results. They seem to have anticipated that their audiences would have difficulty grappling with the statistics. Rolleston recorded a total NSW population of 350,860. Amongst 22 pages of summary and analysis, he attempted to analyse population density using a comparison with the Mother Country. He calculated that if the density of the County of Cumberland excluding Sydney, of 47 persons/square mile, was extrapolated across the colony then;

"New South Wales would admit of a population exceeding fifteen millions of souls, or nearly forty-four times the number of its present inhabitants; and were the Colony peopled as densely as England, it would contain one hundred and three millions five hundred thousand inhabitants" (Rolleston, 1861).

Archer is more entertaining in his 1861 Census analysis for Victoria, dated 25 April 1862:

"28. The total population of Victoria on the night of the Census was 540,322. Scanty as such a population appears, when contrasted with the populations which in older countries occupy such a territory as ours, it is not by any means easy to form at once an adequate conception even of this number. If these 540,322 individuals could be collected together so as to allow one square yard for each individual, it would require an allottment of upwards of 111 acres to give them all standing room. Could they be arranged in a line, four abreast, still allowing a yard for each, such a line would extend nearly 77 miles, or further than from Melbourne to Castlemaine........."

"29. Another way of appreciating such a number will be to see how long it would take to count it. At the rate of one unit per second, the task of counting 540,322, would occupy eighteen days of eight hours each, and six hours and five minutes over"(Archer, 1861).

These examples of dissemination and analysis contrast strongly with the modern census where it is the user of statistics, public or private, who often decides how the results are to be analysed and which tables or profiles from the census are needed. Users are now consulted if any questions are proposed to be added to or deleted from the census form.

In the modern Australian census, defined boundaries are as important to the collector as they were in 1861, whilst the terrain and climate still influences boundary mapping and location, just as they did in 1861. The need to locate and maintain boundaries ensures that the interlocking roles of Surveyor, Mapmaker and Statistician continue to be recognised.

Even 130 years after the 1861 Census, the maps displaying statistical boundaries for use in the national census, were the results of efforts by Surveyors, Draftsmen and Cartographers. The fundamental collection techniques of 1861 survive, although the scale of the undertaking is now massive. Throughout the revolution in Surveying and Mapping, from plane table to satellite navigation and spaceborne imaging, it is the Surveyor and Cartographer who has sensed many of the opportunities that lie in new technologies and has facilitated their use by other professions.

Nowhere is this more evident than in the 1996 Census, where a Geographic Information System (GIS) was used in Australia for the first time to assist with census mapping, collection and dissemination.

"A beautiful set of numbers. The five yearly Census is one of the best public investments Australia makes. It pays for itself many times over in the information it provides for planners in both the private and public sectors. At the everyday human level its findings are engrossing, especially when tracked over time" (Adelaide Advertiser, 17 July 1997)

The Census of Population and Housing is a major logistical task for the Australian Bureau of Statistics (ABS) to measure the number and key characteristics of people in Australia on census night (Castles: 1991). A major national publicity campaign precedes the event. In 1996, over 40,000 temporary staff covered 7.7 million square kilometres, delivering 10 million forms to over 7 million households in the 6 states and 2 territories of the mainland. In addition, they covered the offshore territories of Cocos and Chistmas Islands some 3,600 km to the north-west of Western Australia and the Australian Antarctic Territory which is 22 degrees of latitude further south than Tasmania. Forms on the mainland were distributed in the 10 days before Census night and retrieved within 8 days. Greater lead times were necessary for the islands and for Antarctica where use was made of the supply ship that left Australia in January 1996. 52 questions were asked on a diverse range of topics including religion, ethnicity, education, labour force, transport, income, housing and demographic characteristics (ABS, 1997).

It is hardly surprising that the ABS looked to GIS to facilitate the planning of this level of productivity, although it is not the only national census agency to have done so. David Rhind (1991) provides an excellent overview of recent British and American efforts in this regard. The challenging Australian collection environment and the distribution of population within it, has always biased Australian mapping programmes. It continues to do so and in the case of the 1996 Census Mapping Project it has influenced a modern GIS solution that is unique.

Some 85% of the Australian population huddles in coastal conurbations but there are also thousands of rural and remote settlements in the hinterland and remote interior that must be included in the census and therefore in any supporting mapping program. Census collectors face conditions in the coastal cities that contrast with the outback. One collector may only walk a block of flats, whilst the other may have to cover several thousand square kilometres by air or four-wheel drive. One collector may find little change in established urban areas whilst another may encounter large new subdivisions on urban fringes or discover that a major freeway now crosses the area, making urban access more difficult.

All collectors ideally require a map showing inhabited sites, however a national large scale map series is difficult to justify for the interior and therefore, it has never been created. The national small scale map series that has been created is not adequate for coastal conurbations. Furthermore, no single Australian mapping agency ever had the role of making both small and large scale maps across the country. These roles have always been separated between the Commonwealth and State agencies. A compromise of roles, traditional cartographic principles and map series mentality was needed to achieve a solution for 1996.

Readers requiring a detailed technical description of this project should refer to Mobbs (1995). The following is an outline of the project, for which, map data production was managed by the author - a former Topographic Surveyor and Cartographer.

Following the 1986 and 1991 censuses, growing dissatisfaction with the manual system of maintaining census boundary overlays on traditional paper maps (particularly since the maps were maintained by a separate agency), plus criticism from map users and increased awareness of digital mapping technology led the ABS to re-evaluate its procedures (ABS, 1991)

In March 1992 the ABS advertised for Expressions of Interest in the provision of digital mapping services and facilities (ABS, 1992a) for the 1996 Census, namely; a single efficient system for production of higher quality collection maps and dissemination products using colour if appropriate, more efficient Collection District (CD) design procedures, integration of census data with other related textual data; and preference for a single contractor who would use digital base map data from the public sector.

The principles that the ABS wished to see incorporated in maps from the new system included the following:

Most importantly, the ABS wished to acquire a national digital base map from which to create the maps. This would solve the problems of the range of map scales required for collectors, the future updating process and the distribution of data for CD design, which was previously done manually in each of the nine ABS offices around Australia.

A consortium of Australian public sector mapping agencies agreed to jointly establish the feasibility of providing the ABS with the data required (Grant, 1992). This consortium became known as the PSMA, after the nature of its members’ roles and activities. It comprised the 9 mapping agencies of the Commonwealth, the States and the Territories. The PSMA had not previously contemplated pooling their data in a single database.

The Land Information Centre (LIC) of NSW, headed by the NSW Surveyor General, was selected by the PSMA as the lead agency and on 1 May 1992 the LIC lodged an Expression of Interest for data supply. In the feasibility study, the PSMA reviewed the status of mapping programmes, establishing that by 1995 most digital capture programmes would be complete but some would need rescheduling for the 1996 ABS requirement, It was also realised that some feature capture lists would need to be altered to comply with the ABS requirements. Most significantly, it was realised that a combination of small, medium and large scale source data would satisfy the ABS requirement.

The ABS project therefore represented a catalyst for creation of a valuable national geospatial framework and an imperative to accelerate jurisdictional mapping programmes. The PSMA also foresaw other applications for an integrated national digital map database.

The ABS was committed to the census date of 6 August 1996 and therefore sought guarantees that the map data could be provided in time. The PSMA also needed to know that they were on solid ground and to this end, sought endorsement for the project at the highest level via communications to the Inter-governmental Committee on Surveying and Mapping (ICSM), the Australia New Zealand Land Information Council (ANZLIC), Lands Ministers, and Heads of Government in the States and Territories. Endorsements subsequently received for the project convinced the ABS that it should proceed.

A draft Request for Tender was released for comment in September 1992 (ABS, 1992b). Sample PSMA datasets and prototype maps were provided. Despite the ABS preference for PSMA data, tenderers were at liberty to propose alternative datasets, however, the alternatives proposed did not fully meet the ABS requirements. In December 1992, after tender evaluations, the ABS invited the PSMA to contract for the supply of base map data. A private sector consortium contracted for the hardware and system development.

Contract negotiations commenced in January 1993 and were concluded in June 1993 with the signing of a Memorandum of Understanding (MOU) worth $3.4 million over 5 years. Professional legal advice was necessary to guide both parties to a suitable compromise and to facilitate the production of draft agreements, the final MOU and bilateral agreements that were later struck, between the agencies and the LIC, to mirror the rights and obligations of the MOU. These documents bind the ABS and the PSMA in this project for two full census cycles extending to 1 January 2008. Following the highly successful 1996 Census, the ABS is seeking extension of the MOU to 2012.

The ABS had high expectations of the PSMA. Consequently, in early 1993 during the feasibility study, when the full picture emerged regarding incongruities of jurisdictional datasets, there were expressions of dismay and frustration. Not every State and Territory held its map data in the same system, format, or specification! Moreover, not every jurisdiction held equivalent map coverage! This situation has its historical roots in the jurisdictions’ manner of settlement, their status, size, political systems over time, degrees of autonomy, levels of funding, rates of take-up of technology and different terrain. These differences affected both topographic and cadastral mapping databases.

In the absence of a national standard for either type of data, a single topographic data specification was devised for the proposed Census Mapping Project by the LIC and key aspects of a specification established by the Commonwealth for a 1:250 000 scale digital mapping product were acknowledged as the ideal. Data description tables were compiled for each jurisdiction to characterise their data, highlight differences between datasets, estimate compliance with ABS requirements, and provide the ABS with a common data specification.

Significantly, population density and dispersion were recognised by the adoption of three distinct data zones; Urban, Rural and Remote, as illustrated by Figure 2. These zones not only reflect extremes of density and dispersion as shown in Figure 1 but they reflect how the demands of the population have influenced the allocation of resources for mapping; from the Urban capital cities and major population centres, well served by detailed, expensive and large scale topographic and cadastral mapping databases, to the Remote interior where population is extremely low and predominantly topographic mapping is concentrated in the smaller scales. Thus the integrated PSMA dataset was always going to be a unique "multi-resolution" dataset, incorporating data from 1: 500 to 1:250 000 source scales.

For each data category, data quality attribute tables were prepared to indicate Completeness, Positional Accuracy, Logical Consistency, Label Accuracy, Source, etc.

The first processing performed on data from some jurisdictions was translation to a common CAD format - MicroStation design files. This provided scope for a great deal of error and frustration until procedures were standardised. Standard symbology was then applied and data for each jurisdiction were conformed to the same feature levels. 2047 tiles of cadastral linework data, registered to 1861 tiles of topographic data were created; their dimensions dictated by the sequential order and location of the data deliveries to the ABS and by the need to observe maximum file sizes. The dataset therefore consisted of a patchwork assembly of edge matched topographic and cadastral tiles occupying 4 gigabytes of computer storage space. Features that crossed tile edges were joined precisely so as to create a "seamless" dataset and to allow the statistical boundaries to do likewise for the formation of polygons.

The PSMA features listed in the table at Figure 3 were integrated and assigned to 50 of the 64 available data levels in the MicroStation environment. Aerial photographs, where available, plus previous collector comments on 1991 census maps were used to check completeness. The road centreline network, again classified differently in each jurisdiction, was reclassified to a hierarchy representing the best possible national compromise. Name attribution was implemented for highways, main roads and major drainage, with the remaining features to be attributed after 1996 as part of upgrade and maintenance. Features not attributed were named using text labels.

Cadastral linework and road centrelines derived from DCDB were supplied over all populated places and everywhere else where available. This created a large workload of integration, alignment and checking against other source data since integration was the key to the success of the project. The function of Cadastral linework was to facilitate a casement style of road representation in the Urban zones and townships, to facilitate alignment of CDboundaries where no topographic feature was present and to aid in the depiction of housing density and dispersion via selective printing over topography on the census collection maps.

Figure 4 illustrates the outline data processing schema for the project. From the LIC the digital data went to the ABS’ contractor for checking, cartographic enhancement, digital alignment of the 1991 census boundaries plus building and linking of the other textual components of the system.

The Contractor transferred the data to the ABS Central Office for checking prior to transmission across a wide area computer network to the regional offices for census design for 1996. All ABS offices were equipped with the same computer hardware and software, facilitating interactive census design via a customised Computer Assisted Design (CAD) package. Where necessary, due to population change or changes to Local Government boundaries, new boundaries for 1996 were drawn and linked to the map base and to the census district record, with the objective of estimating and equalising collector workload.

Quality Assurance of the design work was conducted in ABS Central Office, and the new CD boundaries and records were returned to the contractor for incorporation into the mapping system. A computer plot file was created for each collection district and multi-colour high quality field maps were printed (Figure 5). Other maps for group leaders and field managers were produced at smaller scales.

All maps were sent to ABS Central Office for dispatch to field staff. In all, some 37,500 different maps were produced from the one map database (Mobbs, 1996). After all maps were produced, the whole mapping system was transferred from the contractor to the ABS for maintenance, update and use in the planning of future censuses and for other purposes..

The ABS wished to achieve conformity between collection maps and dissemination products. Once the collection task was completed for the 1996 Census, dissemination took first priority. To meet the output need, the PSMA dataset in its cartographically enhanced format was reduced in size from some 10 gigabytes to around 600 megabytes to fit on a standard CD-ROM for use in a personal computer (PC) desktop mapping package.

Two levels of detail were produced to suit basic and advanced analysis by the user. Aggregated results of the 1996 Census plus the digital statistical boundaries were integrated with the map data and MapInfo software. The ABS product known as CDATA96 was made available for sale to the public in October 1997. Sales figures up to December 1996 show that 757 units of CDATA96 Basic and 157 units of CDATA96 Detailed have been sold. An impressive range of hardcopy dissemination products and reports for the public has also commenced.

Since the 1996 Census, the PSMA has received a steady stream of inquiries from private sector companies wishing to incorporate the dataset within other desktop mapping packages or wishing to value-add it for other commercial purposes. Contracts have already been entered into with two such companies and several more are in train.

On the government side, applications for the dataset are either under investigation or implementation in the areas of Electoral Roll maintenance, street address geocoding for automated mailing systems, an automated interactive Electoral Redistribution system, geodemographic support and monitoring of "green sinks" for carbon dioxide absorption.

It is also likely that the dataset will form the interim spatial layer of an Australian Spatial Data Infrastructure currently being mooted by ANZLIC. If this happens then the ultimate use for the dataset will have been realised - that of it being available to every Australian who has a need for consistent and accurate land information based on a national spatial foundation.

Australia achieved an enviable under-enumeration rate of 1.6% in a total population count of 17,892,423 in 1996, verified by post-census sampling. This is the best result for Australia since sampling commenced, 30 years ago and it engenders a high level of confidence for users of the statistics. The government will look to improve this result with a view to better decision-making based upon even more accurate information.

With a refusal rate reported at 0.1% for the 1996 Census, Australians appear to be cooperative and very aware of the census and its importance, not only for Commonwealth grants to the States and Territories and hence, to Local governments, but also for Parliamentary representation at Federal level. Politicians and the public alike need to be assured that electoral redistribution is conducted according to the quota rules. This can be facilitated through accurate referencing of census results to the landscape.

On average, the 1996 Census cost each Australian $7.60. This was 2.3% cheaper than 1991 and 3.9% cheaper than 1986, adjusting for inflation and changing government policy in the interim periods. The government will be looking to reduce this cost even further, whilst maintaining the efficacy of the census.

The Internet has been suggested as a future tool to conduct censuses. However, the ABS surveys have shown that a mere 300,000 Australians currently use the Internet at home and therefore, for the time being, the use of this means of individual census collection appears to be a pipe dream. Even if this form of census collection was realised, geospatial analysis of the census data will continue to require considerable mapping support to facilitate understanding of the population’s density, dispersion, condition, position in the developed world and hence, the allocation of resources, political representation and government assistance.

The experiences of the PSMA in implementing a national digital base map for the ABS has lead to detailed reviews of the topographic and cadastral data models in each jurisdiction. A national working group is now addressing a common data model for use after the 2001 Census that will permit both types of data to be more easily integrated within a virtual data infrastructure. This augurs well for Australia’s future in the Information Age.

The value of this project; to the ABS for conduct of the census and to the nation for other purposes, is apparent. Map data for the entire country have been placed in the hands of those who most need it to carry out their functions in the conduct of an efficient national census. The value of the project for creating Edition 1 of the national geospatial framework is also evident judging from investigation underway with other government departments. Commercial organisations have also recognised that the PSMA dataset, with the socio-economic data from the ABS aligned to it is, de-facto, the authoritative national dataset.

The database not only represents the first edition of a national topographic database but could also be the basis of a national cadastral database and a national digital road network. Australia, through the ABS project, has completed a cooperative national digital geospatial framework - four years ahead of the United States where it required a Presidential Order to commence the work. Not only has the census come to Australia but Australia has come to its senses!

HYDROGRAPHY	INFRASTRUCTURE		RELIEF	ADMINISTRATION
FRAMEWORK	ROADS	WATER	POINTS	BOUNDARIES
Coastline	Highway	Bridge,	Mountain	State / Territory
Island	Main Road	Crossing, Ferry	Hill (Prom.)	Local Govt Bdy
	Sealed Road	Jetty		Localities
DRAINAGE	Unsealed Road		AREAS	Suburbs
Major Drainage	Vehicular Track	AIR	Desert	National Parks
Other Drainage	Other Road	Airport	Dunes	Cadastral Linework
Lake	Lane or Pathway	Airstrip	Sand Ridge
Reservoir
Major Dam	RAIL	CULTURAL
Canal	Railway	Cultural Facility
Pipeline	Railway Station	Minor Area
Swamp		Defence Estab.
Subj. Inundation		Parks/Reserves
Intermittent Lake		Powerlines
Dry Lake		Mine/Quarry
Tidal Flat		Major Fence

ABS, 1991, An Evaluation of the 1991 Census of Population and Housing, Canberra, ABS.

ABS, 1992a Expressions of Interest to Supply Mapping Services and Facilities for the Australian Bureau of Statistics Census of Population and Housing, Canberra, ABS.

ABS, 1992b, Draft Request for Tender (ABS 5213/01) to Supply Mapping Services and Facilities for the Census of Population and Housing, Canberra, ABS.

Archer, W. H., 1861, Appendix C to Report of the Census of Victoria, pp. xix-xi, Melbourne, John Ferres, Government printer.

Forster, C. & Hazlehurst, C., 1988, Australian Statisticians and the Development of Official Statistics, special feature in Year Book Australia 1988, pp. 1-59, Canberra, ABS.

Grant, D., 1992, August, Personal Communication to Director General of the Department of Conservation and Land Management, Bathurst, Land Information Centre.

Knibbs, G.H. (Ed), 1914, Federal Handbook on Australia, pp. 5, Melbourne, Commonwealth of Australia.

Lines, J. D., 1992, Australia on Paper, pp. 19 - 32, Box Hill, Fortune Publications.

Mobbs, J.D., 1995, The Public Sector Mapping Agencies and the 1996 Census - A Mapping Revolution, pp. 1 - 16, Bathurst, Land Information Centre.

Mobbs. J.D., 1996, Report on ABS Census Mapping Project, in 1995-96 Annual Report of the Land Information Centre, Bathurst, Land Information Centre.

Rhind, D.W., 1991, Counting the People: The Role of GIS, in Geographic Information Systems Vol 2; Applications, pp. 127-137.

Rolleston, C., 1861, Census - 1861, Report to the Honorable Charles Cowper, Colonial Secretary, pp. 1-605, Sydney, General Registry Office.

AUSTRALIA COMES TO ITS CENSUS