Government GIS

By the middle of the 1980s, geographic computing inside the New Zealand state was no longer confined to one experimental programme or one specialist scientific group. The Ministry of Works and Development had been maintaining and analysing land-resource information through the NZLRI and LADEDA lineage. Lands and Survey had experience with computer-assisted cadastral mapping, photogrammetry and automated cartographic work. DSIR units were processing satellite imagery and beginning to use commercial GIS. These systems had different purposes, data models and technical histories, but they had created a pool of staff who understood that maps could be maintained as digital information rather than produced only as finished sheets.

Departmental use created a different set of requirements. A system had to survive beyond the project that created it, accept new data, support corrections, produce dependable output and fit within government computing arrangements. It needed operators as well as designers, maintenance budgets as well as purchase budgets, and rules for what counted as an acceptable record. By the end of the decade, some government spatial systems were becoming national production environments with continuing obligations. The work remained specialised and expensive, but it was becoming routine departmental production.

Lands and Survey sat near the centre of that transition because it was responsible for national cadastral and topographic mapping and held large analogue record collections that required conversion and integration alongside any software purchase. In 1987 the Department of Lands and Survey became the Department of Survey and Land Information, or DOSLI. The change of name coincided with a period in which digital land information was becoming a larger part of the department's production work. New GIS systems were introduced alongside earlier mapping systems. DOSLI was operating several digital production lines at once, each developed around a particular body of records and a particular government requirement.

The Digital Cadastral Database required sustained national production and maintenance. Computer-assisted cadastral mapping had already been explored through MAPPAK, but the DCDB put conversion of the cadastral record-map system onto a national programme. A 1991 United Nations country report records conversion beginning in 1986. By December 1990, 1.066 million parcel polygons had been captured, about 42 per cent of an estimated 2.5 million parcels. Chapter 13 describes the DCDB’s positional limitations and its later relationship with Landonline.

The work was much more than tracing parcel lines. Record maps had to be registered to their coordinate framework, geometry captured and edited, descriptive information linked, and inconsistencies checked. Current cadastral changes continued while the historical backlog was being converted, so some areas needed routine digital maintenance while other districts were still being digitised. National conversion also required common coding, topology and quality-control procedures so records produced in different Land Districts could later operate as one maintained system. Digitising created a new production line rather than a finite data-entry job.

Later technical documentation describes the production environment using Sun Microsystems hardware, Unix, VISION* software supplied by SHL Systemhouse and an Oracle relational database. VISION* supported digitising, survey entry, editing, maintenance, enquiry and output. Such a system required specialist operators, survey and mapping knowledge, database and system support, backup procedures and continuing funding. The department was therefore acquiring an organisational capability as well as a set of digital parcel files.

The conversion also demonstrates why institutional GIS cannot be dated simply from a software purchase. The database began acquiring users and maintenance obligations before national capture was complete, while hard-copy survey and title records remained part of the working environment around it. Digital geometry carried forward the strengths and limitations of the source records, so a clean screen did not make every boundary survey accurate. Those cadastral details are developed later; the relevant point here is that, by the end of the 1980s, DOSLI was operating a large national spatial database as continuing government infrastructure.

The 1991 United Nations regional cartographic material records several DOSLI digital production systems at the end of this period. Alongside the DCDB, it records a 100-metre digital terrain model covering about 10 per cent of New Zealand. Coverage was concentrated along major communication routes in the North Island and the east coast of the South Island. The terrain data supported perspective simulation, slope analysis and orthophoto production. This was a separate production capability with its own source data and outputs, not another layer inside the cadastral system.

DOSLI was also computerising its aerial-photograph index. An analogue photo collection is only useful if staff can locate the right photography for a place, date and project. Computerising the index turned retrieval information into a searchable departmental resource even when the photographs themselves remained physical objects. Its immediate function was administrative and production-oriented: staff could manage and find a large national imagery archive more systematically.

Photogrammetry was becoming more computational as well. The UN report records DOSLI purchasing software in 1989–90 for photogrammetric restitution from stereoscopic SPOT satellite imagery. Restitution converts the geometry visible in overlapping images into mapped positions and features, a task that had long depended on specialist photogrammetric instruments and trained operators. Digital processing did not remove the need for photogrammetric judgement. It added software, image data, coordinate transformations and computer storage to a production craft already built around precise measurement.

Aeronautical mapping formed another distinct line. The 1991 country material describes a digital aeronautical information bank used for totally automated aeronautical chart production and records digital maps being generated and updated for radar-control centres. Aeronautical charts have little tolerance for casual updating because airspace, navigation information and chart content have operational consequences. A maintained digital information bank allowed repeated chart production from controlled information rather than rebuilding every edition as a largely manual drawing. Printed charts still remained part of the operational output, so digital production and hard copy coexisted rather than one immediately replacing the other.

By 1991 the 100-metre DTM covered only part of the country. Staff were working with an expanding national resource rather than a completed elevation surface. Its documented uses included slope analysis, perspective views and orthophoto production. Coverage was concentrated along major communication routes, reflecting a staged acquisition and processing programme rather than immediate national coverage.

The aeronautical system had a different rhythm. Its digital information bank supported chart production and the updating of maps for radar-control centres. Here the database existed because the output had to be produced again and kept current. A change to aeronautical information could be incorporated into the maintained digital source and carried into later output, rather than requiring the whole cartographic product to be reconstructed. By the end of the decade, this work was part of DOSLI’s mapping responsibilities.

The computerised aerial-photo index addressed a daily information-management problem. Government aerial photography represented a large accumulation of flights, frames, dates and coverage. An index had to connect a request for a location with the photographs that actually existed. Computerisation made the catalogue more tractable even though a user might still finish the search by retrieving film, prints or other physical imagery from storage. Digital production continued to depend on large analogue collections.

In 1990, DOSLI used several systems across these functions. The cadastral database, terrain models, aerial-photo index, photogrammetric processing, topographic conversion and aeronautical production each arose from different records and production requirements. They shared departmental infrastructure, specialist computing and a growing dependence on maintained digital information. Staff had to keep each production system operating between publication cycles.

Digital topographic production was moving in the same institutional direction. Conversion of national topographic information was under way by the later 1980s, and by the 1991 AURISA conference Geoff Howard, Geoff O'Malley and Robin Pickering could present a paper titled "New Zealand's Topographic Database - Where To Next?" The paper records a change in the organisation’s work: DOSLI was no longer dealing only with printed map editions. It was maintaining digital topographic information that could support later products and supply users outside the traditional cartographic production line.

That changed the role of the national mapping agency. Roads, rivers, contours, place names and other features could be maintained separately from the particular sheet on which they were published, while cartographers still had to decide what belonged on each map and how it should be shown. Chris Hoogsteden and Derry Gordon's 1991 discussion of the national surveying and mapping agency in a GIS world reflects that broader shift from map producer towards custodian and supplier of spatial information. The national topographic database, later digital series and Topo50 are treated in detail in Chapter 15; here it is one more example of digital spatial production becoming ordinary departmental responsibility.

Other departments also faced these requirements. The Ministry of Works and Development's Water and Soil organisation had already built a mature spatial-information capability around the New Zealand Land Resource Inventory. LADEDA had been developed internally for handling land-dependent data and was explicitly described as a GIS by 1985. Near-contemporary Department of Conservation material records that the NZLRI operation changed from its internally developed software to commercial GIS software in 1988.

Chapters 5 and 7 cover the NZLRI in detail. By the late 1980s, spatial information capability was already distributed among organisations with different statutory and operational responsibilities. Water and Soil used GIS for land-resource analysis, Lands and Survey and DOSLI used digital systems for cadastral and national mapping production, and DSIR units used spatial systems for scientific and environmental work. Capability was distributed rather than concentrated in one national GIS unit.

Data exchange between those systems could be awkward. Systems had been built for different purposes, sometimes on different computing platforms and in proprietary formats. Coordinate systems, feature classifications and update cycles did not automatically align. Moving data between organisations could require conversion as well as agreement about what the data meant. The increasing professional interest in land-information systems during the late 1980s and early 1990s grew partly from these practical problems.

Earlier NZLRI data had been held on an IBM 3033 at the Vogel Computer Centre, a central computing environment in Wellington. By the later 1980s commercial GIS and specialist workstations were allowing more spatial processing to sit closer to the teams that used the data, although central services and shared computing did not disappear. Departments had to decide where specialist staff, machines and master databases would sit and how regional users would obtain output. Those arrangements affected response times, support and access to the systems.

The new systems also increased the number of formats a department had to care about. A printed map could be stored, catalogued and read with no software dependency. A digital dataset depended on its file structure, coordinate system, attribute definitions and often a particular software environment. Proprietary systems could make exchange between agencies awkward, while magnetic media and limited storage encouraged disciplined decisions about what was retained online and how copies were managed. The continuing demand for plots and printed sheets was therefore practical rather than evidence that digitisation had failed.

An advisory body known in historical technical material as "old LINZ" was established in 1987 to advise the LIS/GIS community on standards. It was separate from DOSLI and entirely separate from the statutory Land Information New Zealand created in 1996. The shared abbreviation is an easy source of chronological error because late-1980s documents can appear to refer to an agency that did not yet exist. In this period, land-information coordination and standards were already government concerns, but the later institutional structure had not yet been formed.

Mary McEwen's 1989–90 work for the new Department of Conservation captures the management problem from another angle. DOC had inherited biological, protected-area and other spatial records from predecessor organisations, and McEwen examined whether GIS could help organise them. She drew on the Biological Resources Centre’s earlier work with the DSIR Science Mapping Unit and Intergraph. She was considering what it would take for a department to keep such a capability useful.

Her recommendations concentrated on organisational questions. Objectives had to be defined before a system was selected. A pilot could test whether the proposed uses justified the investment. Costs included more than software, because trained staff, data preparation, documentation, equipment and continuing maintenance all had to be provided. A database that was not kept current could quickly become a more efficient way of distributing old information.

The conditions of the period made those requirements substantial. Specialist workstations, graphics displays, digitisers, plotters and storage were expensive enough to make GIS a departmental investment. Staff needed training, and organisations still had to decide whether computing would be concentrated centrally or supported through regional production arrangements. Large analogue collections had to be converted before many of the promised benefits appeared. Printed maps remained necessary for field, operational and public use, so departments often had to maintain digital databases and conventional outputs at the same time.

By 1989 and 1990, DOC was assessing GIS in terms of objectives, staffing, costs, pilots, maintenance and organisational responsibility. The department was considering how inherited biological, protected-area and other spatial records could be maintained within a continuing information system. The surviving reports describe GIS as an organisational investment requiring defined uses, trained staff and ongoing data management. Mature conservation GIS followed later.

Professional discussion widened during the same period. The first National Multi-Disciplinary GIS Conference in Wellington in June 1989 brought together people working with land resources, remote sensing, databases, software, law and mapping. Government-related discussion included intellectual property, land information and resource management as well as technical systems. These subjects crossed departmental and professional boundaries.

The 1991 AURISA conference in Wellington makes the policy shift more explicit. Brian McLay's paper, "LIS/GIS/LINZ The View From A Different Perspective", addressed the land-information debate while the older LINZ advisory usage was still current. Stephen Walsh and Russel McQuoid examined the resource-management process and its implications for land-information systems. Other papers addressed the institutional position of the national surveying and mapping agency and the future of the national topographic database. By then, GIS and land-information systems were being discussed as parts of public administration as well as mapping technology.

The systems described here were already developing before the Resource Management Act 1991 and before the later surge of desktop GIS. Around 1989–91, professional and policy material increasingly addressed standards, custody, data access, interdepartmental use and the responsibilities of the national mapping agency alongside the older production questions of digitising and plotting.

Government agencies were also beginning to encounter a distinction between producing digital information and making it widely reusable. A national database could exist without being cheap or easy for outside users to obtain. Historical DCDB documentation records licensed distribution and per-feature charging arrangements. Data exchange therefore involved contracts, formats and costs as well as technical transfer.

DCDB also became a data source for organisations outside DOSLI. A department could create a national digital framework for its own statutory work while other users saw it as an input to their systems. The technical documentation describes links and indexing that made the database useful beyond simple map production, including relationships to survey plans and other land records and the inclusion of statistical meshblock and derived boundary information. Once that framework existed, councils, utilities, government agencies and private GIS users had reasons to obtain extracts even though they did not share DOSLI's cadastral responsibilities.

Supplying those users created work of its own. Data had to be extracted in known formats, licensed, documented and delivered, and users had to understand the limitations of the source. A line that looked exact on a computer screen might still derive from a record map compiled at a much smaller scale. Distribution therefore carried a quality problem alongside the commercial and technical one. The supplier had to explain enough about provenance and accuracy for the data to be used sensibly outside the production environment that created it.

Maintaining large national spatial datasets was expensive, and government information policy was moving toward user charging. Authoritative digital geography was becoming more useful to councils, utilities, government agencies and private GIS users while access remained licensed, priced and technically demanding. The later shift from priced Crown datasets to open distribution is treated in Chapter 33.

A printed map could be bought and read without the production system that made it. Digital data required compatible software, storage, coordinate knowledge and enough documentation to interpret the files. Suppliers therefore had to provide more than geometry: users also needed format, projection, provenance and quality information. Standards and documentation became part of digital distribution.

By 1991, departments had established digital mapping within several separate systems. DOSLI was converting cadastral records at national scale, maintaining growing digital databases, developing terrain data, indexing aerial photography, supporting digital photogrammetry and automating specialised chart production. Water and Soil had years of operational GIS experience. DSIR and DOC were confronting the management of digital environmental information. Professional forums were debating standards, land-information policy and the changing role of national mapping organisations.

The work remained labour intensive. Paper maps and photographs had to be interpreted and captured. Digital features needed codes, attributes and topology. Databases needed backup, maintenance and correction. Plotters and printers still produced much of what users carried into offices and the field. The transition to GIS did not remove the production workforce; it changed the material they worked on and added database administration, programming and digital quality control to established surveying, photogrammetric and cartographic skills.

By 1991, adoption remained uneven, but several systems had become continuing departmental assets. Some remained specialised production environments, while others were beginning to supply digital information to a wider LIS/GIS community. Record conversion was expensive, maintenance continued after conversion, and standards became more necessary as data moved between systems. In several parts of central government, spatial databases and digital mapping systems now required permanent staffing, funding, maintenance and governance.

During the 1990s, desktop software, cheaper computing and expanding digital datasets brought GIS into more policy, regulatory and operational functions across central government. The wider adoption of GIS across central government is treated in Chapter 24. By then, the departmental databases and production systems established during the 1980s had already created an institutional base for later use.