Early GIS systems

A shared name

On 26 June 1989 a national conference opened in Wellington under the title Getting it all together with GIS. The published proceedings described it as the First National Multi-Disciplinary GIS Conference and ran to 116 pages. The title had a useful honesty: by then there was already plenty to get together. New Zealand organisations had spent more than a decade digitising maps, processing satellite imagery, storing land-resource polygons, calculating terrain and building computer-assisted mapping systems. What had changed during the 1980s was the growing use of one term for work that had previously belonged to separate technical communities.

Conference proceedings showed GIS appearing in several fields. P. R. Stephens, R. G. Gibb and J. R. Dymond wrote on integrating remote sensing and GIS for natural-resource management. P. G. Luckman, R. G. Gibb and M. R. Jessen wrote on combining expert systems and GIS for land classification. Clive Elliott wrote on copyright and intellectual property. Science, image processing, computational classification and information law were appearing at the same meetings under one emerging field name.

Digital geography was already being developed in New Zealand when the conference took place. MAPPAK had been used for cadastral compilation by 1979. DSIR researchers had been processing Landsat data since the mid-1970s. The NZLRI had been digitised from 1977 to 1980 and processed through LADEDA. The change visible in the 1980s was that these kinds of systems were no longer isolated examples that later historians needed to group together retrospectively. Contemporary practitioners were increasingly using the GIS label themselves.

That shift can be seen clearly in 1985. P. R. van Berkel and R. D. Williams published “Introduction to LADEDA: a Geographic Information System” in the New Zealand Geographer. The title did not need a later historian to promote LADEDA into GIS after the event. The people documenting the system were already using the term. In the same year the University of Auckland Geography Department was teaching a senior undergraduate GIS course, while DSIR’s Science Mapping Unit was beginning to use an Intergraph GIS. The term was starting to describe a field rather than one particular machine.

North Taranaki, 1986

Mary McEwen’s 1990 Department of Conservation report preserves contemporary evidence that the Biological Resources Centre was considering digitisation of ecological-region and ecological-district boundaries in late 1985 while DSIR’s Science Mapping Unit was beginning to use an Intergraph GIS. McEwen thought the installation was probably one of the first commercially available GIS systems in New Zealand apart from the specialised NZLRI environment, but she did not claim an absolute national first. A 1986 mock-up for a proposed Biological Resources Information System covering North Taranaki combined graphical and textual material from datasets held in different projections. Commercial GIS was already being used as the GIS label spread across institutions; Chapter 9 describes the scientific application in detail.

McEwen’s 1989 papers also warned that GIS was expensive, complicated and dependent on clear objectives, trained staff, documented data and continuing maintenance. She recommended defining the organisational problem, piloting the system and assessing costs before wider commitment. Practitioners were already considering how GIS would fit into an organisation’s work.

Commercial workstations

Intergraph represented one route into the emerging commercial GIS market. The company had strong international roots in computer graphics, CAD and engineering systems, and its New Zealand presence was not limited to GIS. An August 1987 University of Canterbury Chronicle item independently names Intergraph Corporation (NZ) Ltd in a computer-integrated manufacturing context.

A purpose-built local system could be effective but required local software development and continuing support. A commercial environment offered a wider set of standard functions, vendor support and a path for skills to move between organisations. The cost remained high enough that these were institutional systems rather than ordinary office tools.

The machines also concentrated GIS physically. Workstations, graphics displays, digitising equipment, plotters and specialist software tended to sit in particular units or laboratories. Users came to the GIS rather than carrying it around the organisation. Data conversion could occupy substantial staff time, and much of the work happened before a map appeared on a screen or plotter.

Organisations adopted commercial products alongside earlier New Zealand systems. They entered organisations that already understood coordinate data, digitising, raster imagery, topology, attributes and map output. In some cases the commercial product replaced locally written software; in others it provided functions that had not previously been available in one environment. The transition was therefore more often a migration than a beginning.

Overseas software and New Zealand data

Most of the software used by New Zealand GIS practitioners came from overseas. Esri developed ARC/INFO in the United States and released it in 1982. MapInfo originated in Troy, New York, where its company was established in 1986. New Zealand organisations created the datasets describing this country and put the software to work on them. Surveying, field observation, digitising, coding and checking supplied the information those systems needed.

The software developer, local distributor, data producer and application developer often performed different parts of the work. A New Zealand team could create a dataset and build an application around an imported GIS or database platform. Local software development also continued, including the earlier LADEDA system described in Chapter 5. Each account in this history identifies the people and organisations responsible for the particular contribution, from developing the platform to maintaining the records used in it.

From LADEDA to ARC/INFO

Water and Soil moved NZLRI from the purpose-built LADEDA environment to ARC/INFO in 1988. The migration carried the national land-resource inventory into a commercial GIS platform for ongoing production and maintenance.

The staff making the transition already understood polygon attributes, geographic retrieval and plotting. They had been doing those things through LADEDA. ARC/INFO offered a commercial data model and toolset that was being used internationally and would increasingly be taught in universities and supported by local suppliers. The change gave an established national dataset a route into a more standard professional environment.

The NZLRI database and LADEDA preceded the purchase of ARC/INFO. Commercial GIS later brought that earlier work into a more standardised software environment.

ARC/INFO also brought terminology that became familiar to a generation of practitioners. Its vector data structure used arcs, nodes and polygons linked through topology and attribute tables. Later users would know the term “coverage” as a specific ARC/INFO data structure rather than the ordinary meaning of the word. None of those concepts was entirely new to New Zealand spatial computing, but the commercial package supplied a common implementation and vocabulary.

The migration occurred as other organisations were considering the same general question: continue building and maintaining specialised software, or adopt a supported commercial environment. Organisations answered it differently. Scientific image processing retained specialist packages. CAD remained strong in engineering and utilities. In-house programs continued where they fitted established work. GIS developed alongside these tools rather than replacing them all.

A local market

A software package was only part of a working GIS. Organisations also needed hardware, installation, training, data conversion, troubleshooting and someone to explain why a file from another system did not quite behave as expected. During the 1980s a small New Zealand market began forming around those services.

Later practitioner and company histories place an Esri and ERDAS distribution relationship in New Zealand during the mid-1980s. By the later 1980s, Water and Soil and university teaching programmes were using ARC/INFO.

Explorer Graphics was founded by David and Beryl Pimblott in 1987 and became an early specialist GIS and computer-mapping consultancy. Its later history spans workstation GIS, local-government work and eventually public web mapping. By the late 1980s, organisations could purchase GIS services from specialist firms.

Intergraph followed a different commercial lineage, with strong links to graphics, engineering and mapping. Esri’s ARC/INFO became increasingly visible in vector GIS. ERDAS connected GIS teaching and remote-sensing analysis. The market was already plural before the large expansion of GIS during the 1990s.

This variety is easy to flatten when later market share is projected backwards. New Zealand did not move from no GIS to one dominant platform. Organisations used bespoke software, Intergraph environments, ARC/INFO, remote-sensing packages, CAD systems and other specialist tools in overlapping combinations. The shared term GIS grew across a mixed technical landscape.

Teaching GIS

The appearance of GIS in university courses provides another dated change. University of Auckland Geography was teaching a senior undergraduate GIS course from 1985. University of Canterbury introduced a graduate GIS course in 1987. By 1989 Canterbury Geography had workstation ARC/INFO and ERDAS capability.

A university course differed from learning one organisation’s internal production system. Students could be taught general ideas about spatial databases, raster and vector data, analysis and cartographic output using tools that were also appearing in agencies and consultancies. The resulting skills could travel with graduates. An employee no longer needed to begin entirely with the peculiarities of one locally developed system.

The equipment still shaped how the subject was taught. Workstation software and specialist licences meant GIS was concentrated in laboratories rather than spread across ordinary student computers. Data had to be prepared and storage managed carefully. Output required access to suitable graphics hardware and printers or plotters. The laboratory was both classroom and computing facility.

The pairing of ARC/INFO and ERDAS at Canterbury also reflected the two technical traditions described in the preceding chapters. Vector GIS and raster image processing were now being taught within one academic environment. A student could work with polygon databases and remotely sensed imagery without treating them as wholly separate professions. The international software market had begun packaging together skills that New Zealand organisations had previously developed through different institutional lineages.

Otago, Auckland, Canterbury and other institutions would build GIScience, surveying, remote-sensing and spatial-information teaching and research over the following decades. For the 1980s, the course dates establish a simpler point: GIS had acquired a syllabus.

A field described from New Zealand

Keith Carr’s 1988 paper “The Implementation of GIS in New Zealand”, presented to an ESCAP regional workshop, described systems already in use before the Wellington conference. His examples included the national resource database, digital cadastral work, remote sensing and the growing use of commercial systems. Carr brought these activities together under the term GIS.

Carr’s account also helps separate the national history from the marketing histories of individual vendors. The systems he could point to did not share one technical genealogy. NZLRI had grown from land-resource mapping and locally written software. Cadastral computerisation came from survey and mapping requirements. Remote sensing had developed through scientific image processing. Commercial GIS was arriving across that already populated technical landscape.

The phrase “implementation of GIS” therefore meant several things at once. It could refer to converting an existing national dataset into a GIS environment, installing a commercial workstation, extending a specialised database, or developing new applications around existing digital information. Those activities had different costs, staff and institutional histories. A purchase date alone could not describe them.

This was particularly clear where an organisation already owned years of spatial data. Replacing software did not replace the database, field methods or accumulated classification work. An NZLRI polygon remained the result of resource mapping even when stored in ARC/INFO rather than LADEDA. A cadastral line retained the limitations of its source map regardless of the workstation used to display it.

Data conversion as ordinary work

The growing GIS market created a large amount of conversion work that rarely appeared in short histories of software adoption. Existing maps had to be digitised. Coordinates had to be transformed between systems. Attribute codes needed to be checked and sometimes redesigned. Files from one package could require translation before another package could read them correctly.

Map projection differences were an everyday example. The North Taranaki biological-resources mock-up used datasets compiled in different projections. Bringing them into one geographic display required more than copying files onto the same disk. Staff needed to know the source coordinate system, transform the geometry and check whether the resulting overlay made sense.

The same applied to scale. A national ecological boundary and a detailed local map could both be digital while still representing geography at very different levels of precision. A computer could overlay them exactly in mathematical terms without making their source accuracies equivalent. GIS made comparison easier, but it also made it easier to place unlike datasets on top of one another and forget how they had been compiled.

Attribute conversion could be just as laborious as geometry. Different organisations had their own codes, identifiers, field structures and naming conventions. A commercial GIS could store those fields, but the software did not decide whether two agencies meant the same thing by similar terms. Much of the practical work involved translating institutional knowledge into data structures that another system could use.

Staff performed this work under several job titles. Cartographers prepared source material. Digitising operators captured geometry. Programmers wrote conversion routines. Scientists checked classifications. Database staff maintained codes and identifiers. Survey and mapping specialists dealt with coordinate systems and positional problems. The developing GIS profession was built partly from these existing occupations rather than from a new workforce arriving fully formed.

Specialist hardware

The physical setting of GIS also distinguished the 1980s from later desktop use. Workstations and graphics hardware were expensive enough to be shared resources. Digitising tables occupied real floor space, large monitors and plotters required specialist support, and storage remained limited by later standards. A GIS installation could therefore be a room, a laboratory or a dedicated unit as much as a software licence.

This influenced access inside organisations. A planner or scientist might request work from a specialist GIS or mapping unit rather than operate the system directly. Analysts who used the software regularly accumulated technical knowledge about command syntax, data structures and plotting that ordinary information users did not need. The system could support many organisational functions while remaining physically controlled by a small team.

The arrangement affected turnaround times. A map request might require data preparation, a batch or interactive processing step, plotting and manual checking before the result reached the requester. The computer reduced some kinds of repetitive work but introduced queues around scarce equipment and skilled operators. GIS was faster than older methods for many tasks without being instantaneous.

Training therefore covered both geographic ideas and machine-specific practice. Users needed to understand the software commands, storage conventions and ways of recovering from failed jobs. Commercial support became valuable because a local organisation no longer had to solve every system problem from first principles. The support relationship could include installation, troubleshooting, training, data conversion and advice on workflows.

Software support and dependence

Moving from in-house software to a commercial package changed who carried some of that technical responsibility. A local development team could alter its own program if the source code and expertise were still available. A commercial user depended more on the vendor or distributor for fixes, upgrades and documentation. That reduced some development work while introducing licence and supplier dependencies.

The trade-off was acceptable to many organisations because commercial GIS offered broader functionality and a larger user base. Staff could learn techniques that applied outside one department. Manuals and training materials were shared across customers. Consultants could build experience on one platform and reuse that knowledge with several organisations.

Standardisation was never complete. Different vendors used different file structures and command languages, while CAD and remote-sensing systems continued beside GIS. Data exchange could still involve tapes, custom conversion programs and manual checking. The benefit was that a growing number of practitioners now understood the general problem even when their preferred software differed.

This also made technical documentation more valuable. A locally developed system could rely heavily on the knowledge of the people who built it. A commercial environment required organisations to document how their own datasets fitted the package. Field definitions, coding rules, coordinate systems and quality limits had to be recorded if data were to survive staff turnover or move between systems.

The consulting strand

Explorer Graphics’ establishment in 1987 belongs in this broader service environment. David and Beryl Pimblott created a company whose later work covered GIS implementation, consulting and local-government applications. The firm supplied specialist GIS services before the end of the decade.

Consulting reduced the need for every organisation to assemble all skills internally before starting. A customer could purchase help with software selection, database design, conversion, application development or training. This was particularly relevant for organisations that had useful geographic records but little experience turning them into a structured digital system.

The model also encouraged movement of methods between sectors. A consultant who had worked with one council, science agency or mapping organisation could carry technical approaches into another project. Vendor distributors performed a similar role through training and support. Universities added another route by sending graduates into organisations already using or considering GIS.

By the end of the decade, a small ecosystem therefore existed around the systems themselves. Government agencies produced and maintained national data. Universities taught GIS and remote sensing. Suppliers provided commercial software. Consultancies assisted implementation. Practitioners had begun meeting across institutional boundaries.

Early GIS practitioners

By 1989 GIS was still far from an ordinary desktop tool. Workstations, licences, graphics equipment and data conversion placed it beyond casual adoption by small teams with no technical support. National datasets were incomplete or expensive to obtain, and moving data between systems could require considerable effort. Many future GIS users in councils and utilities had not yet begun the large conversion programmes that would dominate the first half of the 1990s.

Yet the field no longer depended on one organisation sustaining one experimental system. If a specialist left a department, there were other places where GIS skills could be learnt or bought. University courses existed. Commercial suppliers and consultants existed. Several large public organisations had working spatial databases. Conference papers provided a national forum for exchanging methods.

New Zealand had individual geographic information systems before it had a broad GIS profession. During the 1980s the profession began to form around shared software concepts, data practices, courses, suppliers and meetings. The technology became portable between institutions even though the data and organisational problems remained local.

That portability set up the next stage of the history. Once an organisation no longer had to invent its own GIS software, and once staff could be trained in general GIS methods, adoption required less technical invention. It still required money, data conversion, management support and a reason to use the system. Those conditions would come together unevenly across councils, government agencies, utilities, science organisations and businesses during the 1990s.

Defining GIS

Practitioners continued to use different definitions of GIS. A system could draw maps from coordinates without supporting spatial analysis. A remote-sensing package could perform sophisticated raster classification without maintaining vector land records. A CAD system could manage highly accurate engineering geometry while lacking the database structures associated with GIS. Organisations continued to use several of these technologies side by side.

Contemporary use of the term depended partly on context. LADEDA was explicitly called a geographic information system in 1985. McEwen called the DSIR Intergraph installation GIS. University courses used GIS as a subject name. By 1989 the national conference used GIS as a term covering several products and datasets.

A practical common core was emerging. GIS combined geographic coordinates or other spatial reference, structured digital features or cells, descriptive information, storage, retrieval, analysis and mapped output. Not every system provided every function equally well, and specialist tools remained stronger for some jobs. The field was recognisable because those capabilities were increasingly being expected within the same working environment.

The change also affected occupations. Earlier practitioners might have described themselves primarily as cartographers, surveyors, photogrammetrists, programmers, soil scientists, geologists or remote-sensing specialists. GIS created another professional description that crossed those older boundaries. The older occupations did not vanish, but some of their digital work could now be grouped within a common technical field.

This was also the period when organisations began buying capability rather than isolated hardware. A GIS purchase implied software support, data conversion, staff training and ongoing maintenance. A university course implied a body of knowledge that could be taught. A national conference implied enough practitioners to exchange methods across organisations. These developments were occurring at the same time rather than in a neat sequence.

By the end of the 1980s

By 1989 New Zealand had locally developed GIS, commercial GIS installations, national digital spatial datasets, university teaching and specialist consulting. It also had practitioners using the term GIS in refereed papers, government reports and a national conference. The technology remained expensive and concentrated in specialist environments, and much of the national data infrastructure was still being built. The large organisational expansion into councils, utilities and routine desktop work would come mainly in the following decade.

GIS developed through several overlapping programmes. Different components appeared earlier: electronic spatial calculation, automated cartography, raster image processing, vector databases and analytical land-resource systems. The 1980s added a shared professional vocabulary and commercial environment around them.

The 1989 conference is therefore a useful endpoint rather than an origin. By the time people assembled in Wellington to discuss “getting it all together”, there was already plenty to assemble. The next phase would take GIS from specialist teams and research units into the everyday work of organisations across New Zealand.