A national land inventory
In 1973 the began the first edition of the New Zealand Land Resource Inventory, usually shortened to NZLRI. The programme brought together several kinds of physical land information and mapped them across the country as land-resource units. Each mapped unit represented a particular combination of characteristics rather than a legal parcel or administrative area. The inventory recorded geology, soil, slope, erosion and vegetation, and those descriptions supported the associated land-use-capability classification used for planning and land-management work.
The source material was analogue. Staff worked from aerial photographs, existing geological and soil mapping, topographic information and field observations, then drew boundaries around areas judged to have sufficiently consistent combinations of physical characteristics. A boundary on an NZLRI sheet therefore represented interpretation. It did not claim the legal certainty of a cadastral boundary or the measurement precision of a survey control point. Two neighbouring polygons might differ because the soil, slope, erosion pattern or vegetation changed enough to justify a different inventory code.
National first-edition mapping was completed in 1979. By then the programme had produced a large collection of sheets and coded land-resource descriptions covering New Zealand. The paper map remained the normal way most users encountered the inventory, but the production team had already begun converting the mapped units into computer-readable form. Manaaki Whenua’s later institutional history dates digitisation of the first-edition NZLRI database to 1977–1980.
That placed the NZLRI alongside the other digital mapping lineages developing during the same period. Lands and Survey was working on MAPPAK and computer-assisted cadastral compilation. DSIR’s Physics and Engineering Laboratory was processing Landsat data as digital rasters. The NZLRI team was converting thematic land-resource units into vector polygons with descriptive codes attached to them. The systems were built for different organisations and different jobs, but all were shifting geographic work away from a final paper product as the only durable form of the information.
From coloured units to polygons
The first-edition NZLRI contained roughly 85,000 mapped units. In the paper series, the unit was defined by a boundary and a compact code representing the recorded physical characteristics. Digitisation separated those two parts into computer-manageable geometry and descriptive data. The boundary could be stored as coordinates while the coded information about the land unit could be retained as attributes associated with that polygon.
This was different from simply making a digital picture of the sheet. A scanned map would preserve the appearance of the page but would not automatically give the computer a set of separate land units that could be selected or compared. In the NZLRI database, each map unit could be treated as a distinct geographic record. Software could retrieve polygons meeting specified criteria and produce new tables or plots from the same stored data.
The polygon remained dependent on the original mapping judgement. If field interpretation or aerial-photo analysis placed a boundary in a particular location, digitising the line did not improve the original evidence. The database preserved the mapped interpretation and made it easier to retrieve, combine and reproduce. It also made inconsistent coding or geometry easier to propagate across many outputs if errors were not found during checking.
The scale of the conversion created its own production work. Tens of thousands of polygons had to be captured, coded and related correctly to their descriptions. Staff needed consistent identifiers and coding conventions so that a polygon retrieved from the database still referred to the same land characteristics shown on the source mapping. The national database therefore depended on field science, cartographic compilation, digitising, coding and data checking rather than on software alone.
LADEDA
The ’s Water and Soil Division developed its own computer system to process the growing inventory. The system was called LADEDA, an abbreviation recorded in project research as LAnd DEpendent DAta. It was built for the NZLRI and later generalised to handle other geographic data. By the early 1980s the database and software formed a working environment for retrieving, processing and plotting land-resource information.
and developed LADEDA. Williams produced its Plotting Handbook in May 1982 as Hydrology Centre Publication No. 1. Van Berkel and Williams presented LADEDA in their 1985 New Zealand Geographer paper, “Introduction to LADEDA: a Geographic Information System”.
Water and Soil staff built LADEDA for their own computing environment and the inventory’s analysis tasks. A 1990 report explains that the mapped NZLRI dataset had been stored in a GIS because manual analysis was difficult. It identifies LADEDA as the initial software, confirms that the team developed it and records the move to commercial GIS in 1988.
Mapping had begun in 1973, digital capture ran from 1977 to 1980, national first-edition coverage was finished in 1979, and database documentation appeared at the beginning of the 1980s. A plotting handbook existed by 1982 and a refereed paper explicitly described LADEDA as GIS in 1985. By the beginning of the 1980s, staff were using a purpose-built vector GIS.
The mainframe database
By the early 1980s NZLRI data were held on an IBM 3033 at the Vogel Computer Centre in Wellington. The stored boundaries used New Zealand Map Grid coordinates, allowing the land units to be managed as national spatial data rather than as unrelated local drawings. The associated resource descriptions could be extracted into tables, while selected geographic information could be sent to plotting routines. LADEDA provided the retrieval and processing layer between the stored records and the maps or reports produced from them.
The working environment was far removed from later desktop GIS. Users did not have a national dataset continuously visible in a modern graphical interface with instant pan and zoom. Mainframe processing, file preparation and plotting imposed more structured workflows. The existence of the 1982 plotting handbook indicates that producing mapped output had enough commands, conventions and operational detail to require dedicated user documentation.
The separation between database and output was already clear. A paper NZLRI sheet showed one cartographic presentation of the inventory. The digital system could extract a subset from the stored polygons and descriptions and produce a different plot for a particular land-resource question. A user interested in erosion, slope or another coded characteristic did not need a new national field survey or a fresh round of manual tracing before every map could be produced.
That capability depended on standardised coding. The compact NZLRI descriptions had to be consistent enough for a computer to retrieve comparable records from different parts of the country. A spelling variation or informal local notation that a human reader might understand could become a retrieval problem in a database. The work of designing and maintaining the codes therefore sat alongside the geometry and plotting software.
Querying the inventory
The inventory lent itself to computer selection because each land unit contained several coded characteristics. A map that had been compiled to show all of them could also be queried for only one part of the description. The database could retrieve land units matching chosen criteria and send those units to tables or mapped output. The user was no longer restricted to the combinations of information chosen when the standard sheet was printed.
This was a practical analytical change rather than a cartographic one. On paper, comparing several resource factors across many sheets required repeated visual interpretation and manual compilation. In LADEDA the same coded records could be selected repeatedly under different conditions. The stored polygon provided the geographic extent, while the attributes supplied the characteristics used for retrieval.
The system was also generalised beyond the original NZLRI application. Van Berkel and Williams’ 1985 paper describes LADEDA as having been developed to process NZLRI information and later extended to other geographic data. The software provided reusable spatial data and processing functions. The data model and retrieval functions could be applied to geographic records that shared the same basic structure of location plus descriptive information.
The distinction between map and database was visible in routine work. Updating a code in a maintained database could affect later tables and plots without altering every previously printed sheet. New outputs could be produced from the same national dataset without redrawing the entire inventory. The value of the system depended on keeping the database and its classifications maintained after the first conversion work had been completed.
Inventory boundaries
A land-resource polygon had a different meaning from the cadastral geometry discussed in Chapter 3. A property boundary represents a legal relationship established through survey and title processes. An NZLRI boundary divided areas according to mapped combinations of geology, soil, slope, erosion and vegetation. The boundary followed visible terrain or interpreted resource conditions; the transition on the ground could extend across an area.
Digital storage preserved the distinction between interpreted land units and surveyed boundaries. Once displayed on a computer, both cadastral and land-resource boundaries could appear as clean vector lines, and both could be enlarged far beyond the scale at which they were originally compiled. The visual similarity could hide differences in origin and intended use. The NZLRI data remained thematic resource mapping even after it had become a national GIS database.
The original scale also affected geometry. First-edition mapping was produced at 1:63,360, the one-inch-to-one-mile scale inherited from the pre-metric mapping environment. A line compiled at that scale was suitable for regional land-resource interpretation but was not a substitute for large-scale site survey. Later digital conversion allowed the geometry to be revised and transferred into a new map framework, but it did not turn the underlying classification into property-level data.
The difference became especially visible during the national conversion programme of the late 1980s. Changing the database to match the newer 1:50,000 national mapping framework required staff to revisit where the polygon lines sat in relation to updated topographic and photographic information. The job involved much more than multiplying coordinates by a scale factor.
Converting the national series
Between 1987 and 1990 all NZLRI mapping was converted from 1:63,360 to 1:50,000 using GIS. Manaaki Whenua’s historical account records a checking process against aerial photographs and topographic maps. Staff corrected the position of map-unit polygons in relation to coastlines, lakes, trig points, latitude and longitude positioning, and other geographic control. The conversion therefore combined data transformation with systematic review of the mapped geometry.
A simple reproduction of the old sheet at a different scale would have preserved every mismatch between the earlier resource mapping and the newer topographic framework. Instead, the conversion staff compared the polygon boundaries with more current reference information. Some lines could be accepted, while others needed adjustment to fit coastlines, water bodies or positional control more consistently. The revised work produced a 1:50,000 digital dataset aligned more closely with the national mapping base used by other organisations.
The work also exposed the limits of automation. GIS could transform coordinates, manage polygons and support editing, but people still had to judge whether the transformed geometry was plausible. A coastline could be checked against aerial photography and topographic mapping, but a thematic boundary through hill country might still depend on the original land-resource interpretation. The conversion team therefore worked with both geometric control and subject knowledge.
The digitised national inventory contained roughly 85,000 map units by the late 1980s. The programme revisited the geometry across the full series, requiring teams to digitise, check, correct and replot polygons. It was a national data-maintenance exercise rather than a one-off demonstration.
Working with a maintained database
The NZLRI database also changed the sequence of work after the first map had been compiled. In a paper-only system, a revised interpretation could require changes to individual sheets, legends and derived maps. In the digital system, the maintained polygon and its coded description could be edited once and then used in later retrievals and plots. That made data maintenance part of ordinary production rather than a separate task performed only when a new printed edition was prepared.
The same structure allowed one national dataset to support several outputs. A standard NZLRI sheet displayed a broad set of land-resource information, but the database could also be used to select only units meeting chosen criteria. A table could summarise the selected records while a plot showed where they occurred. The database did not need a separate geometry file for every subject because the same map units could be reused with different selections of their recorded attributes.
This placed more weight on database discipline. A polygon had to remain linked to the correct resource description after editing, conversion or replotting. Codes had to retain the same meaning across regions and over time. New or revised information had to be inserted without breaking the relationships already used by other outputs. Work that had once been hidden inside cartographic compilation became explicit database maintenance.
The 1982 LADEDA Plotting Handbook explained how to produce hard-copy maps from stored geographic information. Users could generate different plots from the same database, separating the maintained inventory from its printed presentation. Later GIS continued that approach.
The organisation of the NZLRI also made revision possible without abandoning the original national framework. New field observations or improved mapping could be incorporated into the maintained data while the broader coding system continued to provide national consistency. That did not make updates cheap. Every alteration still required scientific judgement, geometry checking and database control, and a change in one part of the inventory could affect later plots and analyses derived from it.
The database therefore accumulated institutional memory. A map sheet recorded the result of a particular compilation, while the database held a structure that could continue to be queried and revised. Keeping that structure usable depended on documentation as well as software. Plotting manuals, data dictionaries, coding conventions and staff knowledge became part of the system even though they were less visible than the maps produced from it.
Custom GIS development
The Water and Soil Division had a large national dataset and analytical requirements before suitable commercial GIS packages were common in New Zealand. Staff wrote LADEDA around that work, developing retrieval, vector processing and plotting for NZLRI data.
The approach had advantages. The system could be shaped around established inventory codes, national-grid coordinates and the outputs required by the land-resource programme. It did not need to support every data model or every type of user. Staff who understood both the inventory and the computing environment could adjust the software as requirements changed and later generalise it to other geographic data.
The same arrangement carried a maintenance burden. Locally written software required local knowledge to keep it running, modify routines and support users. Hardware changes, operating environments and new requirements could all create work for the team maintaining the code. By the later 1980s commercial GIS products were offering functions that previously had to be developed internally, while also creating their own licence, hardware and training costs.
The 1985 van Berkel and Williams paper captures LADEDA before that transition was complete. Its title uses the term geographic information system without qualification and describes software developed by the for NZLRI processing and later generalised to other geographic data. By then the system had moved beyond an experimental digitising project. It had a national database, user documentation, plotting procedures and a published technical identity.
The 1990 account provides the later institutional view. It describes NZLRI as a large mapped dataset placed in a GIS because manual analysis was difficult, identifies LADEDA as the first software used, and records that staff developed it themselves. The report then notes the change to commercial GIS software in 1988. The 1982 handbook, 1985 paper and 1990 account describe successive stages of the system’s development and use.
From in-house to commercial GIS
The NZLRI team changed from its locally developed GIS environment to commercial GIS software in 1988. Later historical material identifies ARC/INFO as the commercial system adopted in the Water and Soil lineage. The change occurred after years of digital capture, database management, selection and plotting had already been established around NZLRI and LADEDA. Staff transferred an existing geographic database and working methods into a commercial platform.
Commercial software altered the development burden. A purpose-built system required local staff to maintain code for functions that GIS vendors were increasingly supplying as standard products, including vector editing, polygon processing, database links and cartographic output. The organisation could redirect effort from maintaining every part of the software environment towards managing the data and using a wider set of tools. The migration also made it easier to exchange methods and skills with other organisations adopting the same commercial platforms.
Staff transferred the data and working methods into the new platform. LADEDA had been built for a specific national dataset at a time when suitable commercial GIS products were not yet common in New Zealand. By the late 1980s the market had changed. Intergraph systems were appearing in scientific mapping, Esri software had a growing local presence, and universities were beginning to teach GIS explicitly. A system developed internally in the late 1970s now operated in a field with commercial alternatives.
The 1988 change also marks the point at which several previously separate traditions, including land-resource databases, remote sensing, automated mapping and commercial graphics systems, increasingly came under the shared GIS label. NZLRI had already supplied one of those traditions with a national database, experienced users and more than a decade of accumulated work.
Early spatial datasets
NZLRI and LADEDA were operating before the adoption of commercial GIS packages. New Zealand’s land-resource programme started before the term GIS was common in local professional use. The inventory was being digitised from 1977, and by the early 1980s a purpose-built computer environment was retrieving, processing and plotting national polygon data. Contemporary authors called the system GIS by 1985.
Software could only retrieve erosion classes or produce a land-resource plot because field and mapping teams had spent years defining units, interpreting photographs, checking ground conditions and coding descriptions consistently. The digital system reused that work. Its analytical capability depended on the quality and maintenance of the inventory beneath it.
The later 1:50,000 conversion continued the same pattern. New software and coordinate transformation did not remove the need to compare lines with aerial photographs, topographic maps and control points. Digital production changed how corrections could be made and propagated through the database, but it did not remove the judgement involved in deciding where a thematic boundary belonged. The national dataset became easier to query and reproduce while retaining the characteristics of the mapping from which it had been built.
By 1990 the NZLRI existed as a national digital land-resource database operating within a commercial GIS environment, with its earlier LADEDA lineage preserved in technical publications and institutional records. The paper sheets remained useful products, but they were no longer the only place where the inventory existed. The maintained spatial data could generate new maps, tables and analyses from the same national record. New Zealand had moved from mapping land resources onto sheets to maintaining the mapped units as spatial data.
Chapter source notes
1. The New Zealand Land Resource Inventory chronology is supported by official Water and Soil/NZLRI histories and later Manaaki Whenua documentation. The chapter uses 1973 for the beginning of the national NZLRI programme, 1977 to 1980 for first-edition digitisation and approximately 85,000 mapped units for the first-edition national dataset where the existing source trail supports those figures.
2. R. D. Williams, LADEDA Plotting Handbook, Hydrology Centre Publication No. 1, May 1982, is bibliographically secure even though the full handbook scan remains a physical/repository retrieval object. Commands and screens remain unverified while the handbook is unavailable.
3. P. R. van Berkel and R. D. Williams, "Introduction to LADEDA: a Geographic Information System", New Zealand Geographer, 1985, provides direct contemporary evidence that LADEDA was being called a geographic information system by its own practitioners.
4. Department of Conservation Science and Research Internal Report 88, together with the NZLRI section and later institutional histories, supports the historical description of NZLRI coding, LADEDA and the late-1980s move into commercial GIS. The manuscript uses 1988 cautiously for that migration and does not invent an exact cutover project where the evidence is thinner.
5. The 1987 to 1990 conversion from the one-inch mapping framework to 1:50,000 and subsequent second-edition work is supported by NZLRI institutional documentation. Chapter 7 uses the resulting data and classification system but does not retell this conversion history.