THE DEVELOPMENT OF THE SOUTH AFRICAN NATIONAL TOPOGRAPHIC INFORMATION SYSTEM

Transcription

1 THE DEVELOPMENT OF THE SOUTH AFRICAN NATIONAL TOPOGRAPHIC INFORMATION SYSTEM Vorster, P.J. Chief Directorate: Surveys and Mapping, Department of Land Affairs, Private Bag X10, Mowbray, South Africa. Website: ABSTRACT The Chief Directorate: Surveys and Mapping (CDSM) is South Africa s National Mapping Agency and is mandated, amongst others, to provide and maintain national mapping coverage of the Republic. The primary national mapping series is the 1: series, consisting of 1912 map sheets, each covering a ¼ by a ¼ degree square (approximately 25km by 25km). It is this mapping series that was to eventually form the basis for the National Topographic Information System (NTIS). The primary development of the NTIS spans a period of 11 years, commencing in 1986 and being completed in In developing the NTIS the CDSM would implement two distinct processes, the first being a scanning and vectorising process, and the second being a GIS population and structuring process. The Geographic Information System (GIS), acquired in June 1988, was a locally developed package, named ReGIS, running on an AOS/VS platform. UNIX, OS/2, Windows 3.1 and then Windows NT would later replace the AOS/VS operating system. Intergraph s GeoMedia Professional 3.0 would replace the ReGIS GIS in The paper will provide a chronological overview of the two stages of development of the NTIS. The first stage was a pilot project, where the 1:500,000 national map series features were scanned, vectorised and populated in a GIS to provide a completed sample database for evaluation purposes. The second stage would build on the strategy and process refinements of the pilot in order to establish the NTIS through the scanning, vectorising and GIS population of the major features of the 1: mapping series. A brief description of the difficulties encountered, and the effects of these on the datasets produced, will be incorporated. An overview of the various digital processes that have been put in place to upgrade the data, which is cartographic in nature, to that of a topographic dataset will be presented. Here the process environments, and the relevant systems and technologies in use, will be identified. The NTIS primary data classification will be listed. Examples of NTIS application, where this database has proven itself to be a valuable national asset, will be highlighted. The conclusion will then summarise proposed future developments that will affect the NTIS. 1. INTRODUCTION The Chief Directorate: Surveys and Mapping (CDSM), South Africa s National Mapping Agency, is responsible for the provision and maintenance of national topographic map products, an orthophoto mapping series, national control survey networks and a national imagery database. The national topographic line map products are produced at scales of 1:50,000, 1:250,000 and 1:500,000. The orthophoto map series is produced at 1:10,000 scale and covers metropolitan and national development areas. The control survey networks comprise national three-dimensional and vertical networks. The three-dimensional networks comprise a trigonometric network, consisting of approximately 30,000 beacons, an active GPS base station network, named TrigNet (1), currently consisting of 34 base stations, and town survey mark schemes, present in more than 120 towns and cities throughout South Africa. The vertical network comprises a national precise levelling network, consisting of approximately 20,000 benchmarks. The national imagery database currently consists of aerial photography dating back to The CDSM provides additional cartographic map products, at various scales, of each of the 9 provinces of South Africa, South Africa, Southern Africa and Africa (south of the Sahara). The International Civil Aviation Organisation maps, covering Southern Africa, at scales of 1:500,000 and 1:1000,000, are produced and maintained by the CDSM on behalf of the Civil Aviation Authority. The CDSM also maintains a topographic names database, affiliated to the South Proceedings of the 21 st International Cartographic Conference (ICC) Durban, South Africa, August 2003 Cartographic Renaissance Hosted by The International Cartographic Association (ICA) ISBN: Produced by: Document Transformation Technologies

2 African Geographical Names Council (SAGNC) and provides in-house training to students studying in the fields of Surveying, Cartography and Geographic Information Systems. It is important to note, from an organisational point of view, that up until 1997, the functions and responsibilities of the CDSM were incorporated within the Chief Directorate: Surveys and Land Information (CDSLI). The CDSLI would also be entrusted with the management of the cadastre via four satellite Surveyors General Offices. The development of the NTIS would, therefore, be one of three Information Systems initiated by the CDSLI. The other two databases being the Cadastral Information System (CIS), a GIS of all cadastral land parcels, and the State Land Information System (SLIS), a GIS of all state owned land parcels. These three databases would collectively be referred to as the National Land-Oriented Information System (NLOIS). In 1997 the CDSLI would be re-engineered resulting in the creation of the Chief Directorate: Cadastral Surveys (CDCS), entrusted with the management of the cadastre, and the CDSM. For the sake of simplicity I have referred to the CDSM and the CDCS and not the CDSLI. The primary national mapping series is the 1: series, consisting of 1912 map sheets, each covering a ¼ by a ¼ degree square (approximately 25km by 25km). Technological developments in the fields of digital photogrammetry, mapping and GIS prompted the CDSM to replace its manual mapping methods with their digital counterparts. To this end it was decided to develop a national topographic database, based on the national 1:50,000 topographic map series product, to become the National Topographic Information System (NTIS). The development of the NTIS would entail two distinct processes, the first being a scanning and vectorising process, and the second being a GIS population and structuring process. The NTIS development programme, spanning 11 years, commenced in 1986 with the acquisition of a scanning and raster to vector digital conversion system. The first stage of development was a pilot project, consisting of the digital conversion of the 1:500,000 national map series, allowing the CDSM to refine processes and strategies prior to commencement of the second stage. The GIS data capture would commence in 1988 with the acquisition of the first GIS system, named ReGIS. The pilot project would be completed in 1992 and would serve as a prototype for the eventual NTIS. Maintenance of this pilot 1:500,000 database would cease in The second stage of development consisted of the creation of the NTIS proper, consisting of the raster to vector conversion of the 1:50,000 national map series. The second stage would overlap the first stage where this stage s raster to vector conversion process would commence in 1988 with the GIS population and structuring commencing in 1992 and being completed in Post 1997 has seen continued maintenance of the system where newly compiled topographically captured data has replaced the initial data captured, which is cartographic in nature. Various feature types, omitted in the initial capture, have been captured and general maintenance and partial clean-up procedures have been implemented. The current focus is threefold, firstly, to implement error detection and correction processes to ensure a topologically correct relief and hydrological national dataset, secondly, to define a CDSM data model, incorporating all omitted datasets, where this data model will be implemented and populated accordingly and, thirdly, to continuously replace the NTIS cartographic source content with newly acquired topographically sourced data. 2. STAGE 1: THE DIGITAL CONVERSION OF THE 1:500,000 MAP SERIES PILOT The first stage would consist of the digital conversion of the 1:500,000 map series. Here, for each map sheet, the relevant map separate film positives would be scanned and the raster points, lines and polygons converted to their vector equivalent as a first process. The second process would entail the GIS population and structuring of the data within the defined classification schema. The 1:500,000 map series contains 23 map sheets, Figure 1, where the base cartographic material of each sheet consists of five film separates being the black, blue, brown, red and green plates. Figure 1. 1:500,000 Map series index.

3 2.1 Scanning and Vectorising The first stage of the development of the NTIS commenced in July 1986 with the implementation of a SCITEX (Scientific Textiles) system, running on a proprietary operating system. The system configuration is set out in Table 1. Table 1. SCITEX system configuration. Component Units Sub-Components Scitex Scanner 1 Scitex Super Plotter 1 Scitex Response R-280 (HP 1000 computers-32kb RAM) 3 2 x graphic workstation 1 x processing workstation The Scitex Scanner would be used to scan the cartographic film separate positives and the Scitex workstations, in turn, would vectorise the line work contained within the scanned raster images and encode these for later import into the GIS. The Scitex workstations would serve a dual role in that revised edition cartographic film separates would be edited and prepared for plotting to the Scitex Super Plotter for use in the printing of new edition maps. Production shifts were introduced in 1987 in order to maximise the raster to vector production programme. Two shifts, of two persons each, would realise 14 hour production days. The five cartographic film separates scanned, vectorised and encoded were:! The Black plate: roads, power lines, railways, spot heights, built-up areas, etc.;! The Brown plate: Contours;! The Blue plate: hydrological features (rivers, pans, dams, pipelines, furrows, aqueducts etc.);! The Red plate: roads etc.;! The Green plate: Land use features (cultivated land, conservation areas, recreation areas etc.); The process entailed:! Creating film positives for each of the film separates via manual reprographic processes;! Scanning film positives (@20 lines per mm 77 minutes per scan [drum speed 300 revs. per min.]);! The registration, referencing and vectorisation of the black plate;! For subsequent scanned plates, the registration to the black plate and the vectorisation of these;! Feature coding and limited edge matching of vector data;! Archiving of raster and resultant vector data. The primary scanning, vectorisation and encoding process of the 1:500,000 map features would be completed in 1988, with some additional work continuing through to GIS population and structuring In 1988 the CDSM acquired GIS technology for the first time. The original technology plan, as set out in Table 2, was to be phased in over a period of five years. Table 2. GIS Technology Plan (1988). Component Units Sub-Components Base System 1 Host Computer (1440 MB disk storage) 1 Local area network (connectivity) Workstations 4 Raster edit workstations 11 Graphic workstations (for editing) 6 Graphic workstations (for digitising and editing) Terminals 1 Graphic query terminal 12 Alphanumeric terminals Peripherals 5 Colour screen copiers 2 Printers The CDSM took delivery of the first phase of the GIS system in July The components of this first phase are listed, in Table 3, in order to emphasise the advances in computer technology in the last two decades. The host would house the entire GIS database with each of the Graphic workstations downloading /uploading sheet components of this dataset, via the 10 base 5 network, but effectively operating in a standalone environment. The GIS software was a locally developed package, named ReGIS. The operating system was AOS/VS. In 1989 the existing

4 graphic workstations (DS7500) had a RAM upgrade to 6Mb as the initial 4Mb configuration proved inadequate for data processing requirements! An additional digitising and editing workstation was acquired in June This workstation would be the first acquisition of a 386 microcomputer (6Mb RAM, 20Mhz CPU speed and 150Mb HDD), running on a Disk Operating System platform (DOS), an A1 GTCO digitiser and the ReGIS software. Table 3. GIS Implementation first phase. Component Units Sub-Components Description Base System 1 Host Computer MV (3 magnetic tape units, 8Mb RAM, 592 Mb HDD and ReGIS file server software) 1 Local area network (connectivity) Supplied (10 Mbaud) Workstations 2 Raster edit workstations The Scitex HP 1000 s were already in service. 1 Graphic workstations (edit) DS 7500 (4Mb RAM, 160Mb HDD, 737Kb floppy, 21Mb cassette and ReGIS software) 1 Graphic workstations (digitise/edit) DS 7500 (4Mb RAM, 160Mb HDD, 737Kb floppy, 21Mb cassette and ReGIS software and Calcomp digitiser) Terminals 1 Graphic query terminal Data General Dasher (D461) 3 Alphanumeric terminals Data General Dasher (D215) Peripherals 1 Colour screen copiers Plotmaster 2 Printers 180 and 350 characters per second. A second such system was purchased, early In 1991, an additional four editing workstations were acquired. These workstations would also be 386 microcomputers running on an OS/2 platform (8Mb RAM, 20Mhz CPU speed and 150Mb HDD). The OS/2 operating system was preferred to DOS as it provided multi-tasking capabilities with improved networking capability, user-friendliness and performance. These workstations would allow the CDSM to gradually retire the DS7500 workstations from the production environment. The vectorised and encoded data was imported and geo-referenced within the GIS by means of a customized Scitex Importer utility. Here the data was structured with respect to the classification tree, topology was added to shared segments, as chains, and edge-matched. The database consisted of a project file per 1:500,000 map sheet. The first stage of the development of the NTIS, being the 1:500,000 GIS database pilot, containing all the feature instances of that map series, was completed in July STAGE 2: THE DIGITAL CONVERSION OF THE 1:50,000 MAP SERIES The second stage would consist of the digital conversion of the 1912 map sheets of the 1:50,000 map series, utilizing the lessons learnt during the pilot first stage, in order to establish the NTIS itself. An initial test capture of a typical 1:50,000 map sheet realized a total conversion and GIS structure period of 6 months, implying a period of 900 personyears (or in excess of 100 years with available resources) to establish the NTIS. Client requirements (primarily the State Departments of Water Affairs, Environmental Affairs, Development Aid and Geological Surveys) demanded that the conversion and GIS population of the NTIS be completed in a 5 to 7 year period. To realize this goal a partial capture was adopted where only the major features would be converted and form the basis for the NTIS. In contrast to the first stage, this stage would require that only the black, blue and brown plate map separate film positives be scanned where the resultant raster points, lines and polygons of selected features would be converted to their vector equivalent as a first process. The second process would entail the GIS population and structuring of this data within the defined classification schema. 3.1 Scanning and Vectorising In 1988, on completion of the first stage, the SCITEX system commenced with the raster to vector conversion and encoding of the selected 1:50,000 map feature instances. The partial capture necessitated the scanning, vectorising and classification of the listed features within the relevant three cartographic film separates as set out Table 4. A Data General Aviion 410 server would replace the MV15000 host in The AOS/VS operating system would be discontinued and RAM and disk storage increased. The Aviion 410 ran on a Unix (DG/UX) operating system, had 16Mb RAM and 2Gb HDD. The Aviion 410 would, again, only be used as a data archive, with the GIS workstations operating in a standalone environment and networked to this server.

5 Table 4. Partial Capture feature list. BLACK PLATE Roads National freeway, National route, Arterial route, Main road, Secondary road, Other road, Street in built-up area, Road under construction. Railway Standard railway, Narrow gauge railway, marshalling yard, old disused railway, service railway, station, rail under construction. Tunnels Tunnel, Tunnel on National Freeway, Tunnel on National route, Tunnel on Arterial route, Tunnel on Main road, Tunnel on Secondary road, Tunnel on Other road, Tunnel on Street in BUP, Tunnel on Standard railway, Tunnel on Narrow gauge railway. Bridges Bridge, Bridge on National Freeway, Bridge on National route, Bridge on Arterial route, Bridge on Main road, Bridge on Secondary road, Bridge on Other road, Bridge on Street in BUP, Bridge on Standard railway, Bridge on Narrow gauge rail. Other Power lines, Built-up-areas. BROWN PLATE Contours Index contour, Intermediate contour, Index depression contour, Intermediate depression contour. Other Cliff (feathered contour), Eroded area, Terrace. BLUE PLATE Rivers Single perennial river, Single non-perennial river, Double perennial river, Double non-perennial river, Single dry river, and Double dry river. Dams etc. Dam water line, Dam wall, Lagoon, Lake, Marsh, Vlei. Pans Very small non-perennial pan, Perennial pan, Dry pan. Other Aqueduct, Canal, Furrow, Pipeline, Siphon, Coastline. The features not vectorised, classified and captured within the NTIS are listed in Table 5. Table 5. Not captured feature list. BLACK PLATE Line work Building, Hut, Large Building, Ruin, Silo, Fence, Track/Footpath, Aerial Cableway, Conveyor belt, Cutting, Embankment, Excavation, Digging, Mine dump, Slimes dam, Wall, Weir, Airstrip, Airport, Anti erosion wall, Stock pen, Stadium, Large Reservoir, Sewage works, Coastal rock. Symbols Trigonometrical Beacon, Wreck, Wind pump, Cemetery, Cave, Spot height, Battlefield, Monument, Benchmark, Magnetic Station, Marine beacon, Communication tower. GREEN PLATE Line work Cultivated land, Saaidamme, Orchard, Vineyard, Woodland, Forest, Single tree, Row of trees, Recreational ground, Golf course, Racecourse, Nature reserve boundary, Fire break. RED PLATE Symbols Lighthouses, Marine Lights. BROWN PLATE Line work Eroded area, Rocky outcrop, Sand area, Sand dune, Mud flat, Sand bank. BLUE PLATE Line work Reservoir, Fountain, Hot spring, Water tower, Waterfall, Rapid, Cataract, Salt works. An analysis of Scitex production, from 1988 to 1990, revealed that a digital raster to vector conversion rate of approximately 120 map sheets per year could be maintained, requiring an estimated 15 year program to complete the conversion of the 1:50,000 map series. The system, as stated previously, would also be required to produce cartographic map separates for the printing of new edition map sheets, extending this estimate. The CDSM acquired a Laserscan VTRAK system, in January 1992, to complement the Scitex system and ensure completion within the 5 to 7 year window. The VTRAK system came into full production in July It would be the first true client server system introduced at the CDSM. The VTRAK system configuration is set out in Table 6. Table 6. VTRAK system configuration. Component Units Description Master Node 1 VAX Server 4000 (20Mb RAM, 600Mb HDD, 296 Mb streamer tape, (1/2 1660BPI) tape drive) VTRAK Workstation 4 VAX Station 3100 (20Mb RAM, 665Mb HDD)

6 The VTRAK system had a proprietary VAX/VMS operating system and could vectorise in excess of 250 map sheets per annum, with two shifts operating. A customised internal file format (IFF) converter would be created in order to transfer the VTRAK vectorised data to the ReGIS system. The VTRAK system would primarily be utilised for the raster to vector conversion and encoding of the blue and brown plate features, and the Scitex system would, similarly, convert and encode the black plate features. 3.2 GIS population and structuring The GIS population and structuring of the 1:500,000 pilot database was completed in The NTIS, consisting of a GIS database containing a subset of the 1:50,000 cartographic map features would commence soon after. The long-term goal was to attain countrywide coverage where the 1912 sheets would be edge-matched into a seamless topographic (cartographic) dataset. This would allow the creation of user defined map sheets that did not necessarily have to follow the boundaries of the hardcopy 1:50,000 map series and where multiple map sheet coverages could be combined into a single map. An example application would be the Gauteng (Johannesburg/Pretoria) metropolis, which is contained within four separate 1:50,000 map sheets. Hardware and software capability did not support this ideal and the NTIS would initially consist of a database file per map sheet. In January 1993 the 8 existing ReGIS workstations were upgraded and an additional 6 workstations were purchased. The 14 ReGIS workstations would run on a Windows 3.1 operating system with the accompanying compatible ReGIS (version 4.4) software purchase/upgrade. The ReGIS software utilized a proprietary database for spatial information where attribute data was stored within a DBASE III database. During 1993 and 1994 the entire ReGIS workstation hardware fleet would be upgraded to 486 microcomputers (8Mb RAM, 33Mhz CPU speed, 200Mb HDD, 16 bit architecture). In May 1993 the DG Aviion 410 NTIS server host was replaced by a DG Aviion 5500, (33 Mhz RISC processor, 32Mb RAM, 6Gb HDD), in order to increase disk storage capacity and improve database accessibility. The server would, however, continue to act as a data repository serving the ReGIS workstations. The local area network(lan) had been upgraded to 10baseT at this time. In 1994 the Scitex Super Scanner (and Plotter) experienced increased maintenance problems, resulting in significant downtime. To alleviate this problem an Intergraph Mapsetter 5000 plotter/scanner was purchased and put into operation in January The Mapsetter 5000 specifications are set out in Table 7. Table 7. Mapsetter 5000 Production Specifications. Operation Production Specifications Scanning Resolution: 12.5µm (±5 µm) Speed: 2 megapixels per second (80 min.) Plotting Resolution: 12.5µm Speed: 8 megapixels per second (20 min.) The Mapsetter 5000 would be commissioned at a 20 µm scanning resolution resulting in an average period of 45 minutes per map separate scanned, realising a 40% gain in production time (at double the resolution) when compared to the Scitex Scanner. In 1996 the 14 workstations would be upgraded to Pentium microcomputers operating on the Windows NT operating system (32Mb RAM, 120MHz CPU speed, 1Gb HDD, 32 bit architecture) in order to support the ReGIS software upgrade to this platform. These hardware developments allowed the creation of multi-sheet, edge-matched databases for the first time. Each map sheet required approximately 40Mb HDD storage. The adjacent four files would be merged and edged-matched to form a ¼ degree square block (160Mb HDD storage) and then adjacent blocks would be edge-matched (320Mb HDD storage) until a full degree square was completed, Figure Project Rapid Update In 1996 the CDSM was approached by the Central Statistical Services (CSS) (later to be re-named Statistics South Africa) to provide digital topographic base data for South Africa. The CSS wished to integrate Figure sheets per Deg.Sq.

7 this data with the Chief Directorate: Cadastral Surveys (CDCS) Cadastral Information System (CIS) in order to generate a geo-spatial database of enumerator areas relating to the 1996 census. The CSS 1996 census data was captured by means of alphanumeric text descriptions accompanied by sketch plans (2). The Independent Electoral Commission (IEC) would then use the enumerator dataset in order to determine Voting Districts to be used in the 1999 elections. The Municipal Demarcations Board (MDP), constituted in 1998 (3), would then determine municipal and other boundaries from these voting districts, Figure 3. Figure 3. IEC and Census data capture. The cadastral database would be the primary source for the determination of enumerator areas in formal urban areas. The enumeration of rural areas, however, utilized both cadastral and topographic features as boundaries and this required the NTIS database to spatially reference these enumerator areas. An assessment of the progress made in the development of the NTIS database, table 8, revealed that approximately 1000 of the :50,000 map sheets would have to be captured in order to adequately support the CSS. A project named Rapid Capture was formulated at the CDSM, where the NTIS database would be sufficiently completed to serve as reference for the enumerator area determination. This project would commence in June 1996 and be completed in June The project would entail (refer Tables 4 and 8):! Black Plate: The capture of road features only, where the built-up-areas and power lines were excluded;! Blue Plate: these features would be captured, but area features narrower than 750m (i.e. Pans) would be excluded;! Brown Plate: the capture of these features received low priority;! Topology: shared segments would not be structured and edge matching received low priority. Table 8. Progress from 1989 to 1995 (To do: Rapid Capture ). Early in 1996, the Scitex Scanner (after eight years of service) broke down and was removed from the production line. The Mapsetter 5000 could not provide all the necessary scanning required for the Rapid Capture project, and a Contex scanner (optical resolution 300dpi / interpolated resolution 600dpi) was sourced, on loan, to fill the void. The Contex scanner would provide a scanning resolution of 600dpi(~43µm) and a production rate of 10 minutes per scanned film map separate. Project Rapid Capture would be completed in July In parallel, the CDCS would rapidly complete the population of its CIS through Project Hope (to complete the country-wide capture of rural cadastre) and Project Miracle (to complete the capture of urban cadastre).

8 3.4 Project Merge In June 1997, the CDSM embarked on Project Merge to assist the CSS in creating a GIS of enumerator areas. The target date was 31 December In May 1997 a DG Aviion 3600 NT server (24Gb raid capacity) was acquired and ReGIS (Gserver) management software loaded allowing for the creation of Multi-User Geobases (MUG s). This configuration would provide client server functionality and introduce quad tree indexing, allowing the creation a countrywide seamless NTIS for the first time. Within Project Merge the CDSM would combine, structure and edge match the relevant degree square topographic databases referenced to 1:500,000 map sheet areas, Figure 1, thereafter these databases would be combined structured and edge matched within the four MUG s (NW, SW, NE and SE), Figure 4. To avert data volume and operational difficulties a single countrywide database was not contemplated. Figure 4. The four MUGS. The CDSM would then successfully merge and vertically integrate the rural topographic and cadastral datasets and then derive the ±6500 (of the ±86000 total) enumerator area polygons. In addition, the CDSM produced digital colour orthophotos covering the informal housing areas in major urban areas, to be used in the field, by CSS contract staff, to identify and delineate the boundaries of those enumerator areas. The CSS would experience difficulties in completing the latter and the CDSM would assist by carrying out the field identification and delineation of ±3000 enumerator areas in the Eastern Cape rural, and the Cape Town and Gauteng informal housing areas. Project Merge was completed in the first quarter of Lester (4) provides an excellent summary of the establishment of the CSS enumerator database and its successful application in support of the 1999 South African national general elections. Post Project Merge the CDSM resumed the capture of contours and correction of the NTIS content through error detection and correction techniques. 3.5 The New South African Datum: Hartebeesthoek94 and transfer to an Oracle Spatial platform Concurrent to the development of the NTIS, the CDSM was in the process of re-defining the South African horizontal geodetic datum, to be referenced to the WGS84 ellipsoid, by complete re-computation of this geodetic network. This new datum, to be named the Hartebeesthoek94 Datum, would officially be adopted on the 1 st of January 1999 (5). The NTIS would require a datum transformation, from the Cape Datum (referenced to the modified Clarke 1880 ellipsoid) to this new datum. To complicate matters, AutoDesk acquired the ReGIS software source code, in 1995, with the intention to re-engineer the code from a 16 bit to a 32 bit application, redesign user interfaces, implement a new data structure and rename it Autodesk World. Autodesk World would not support MUG s. The ReGIS product was not WGS84 enabled and would not receive any further development. The CDSM had not only created the NTIS on the ReGIS application suite, but had also integrated its photogrammetric data capture processes to it through customised software routines. Adopting the Autodesk World product would require the re-development of these customised software routines. In March 1999 the CDSM replaced the ReGIS GIS application software with Intergraph GeoMedia Professional 3.0, which would manage the NTIS on an Oracle Spatial (relational-8i) database platform. Here the NTIS data was exported in ReGIS feature file format where these feature files were then subjected to a datum transformation from the Cape to the Hartebeesthoek94 Datum, Table 9.

9 Table 9. Datum Transformation Summary. The transformed feature files were then imported and structured within the Oracle Spatial database where these data where indexed to the Hartebeesthoek94 map sheet boundary references. The migration to the new system was completed in October 1999, with final acceptance taking place in February The advantages of the new system where/are:! A single centralised database on an Oracle Spatial platform;! Feature data validation tools;! Dynamic point snapping;! Improved graphic user interface (GUI) and image management;! Open system with extended input/output utilities to other data formats;! A refined hierarchical classification schema with 27 feature types and 233 related attribute types (Table 10). Table 10. Features and number of related attributes. The customised ReGIS photogrammetric data capture software would be replaced by usmart photogrammetric data capture software, running on a Microstation CAD platform. Intergraph s Modular GIS Environment software (MGE) would serve to facilitate data transfer to the NTIS Oracle Spatial database. In April 1999 the vectorisation of the remaining features, excluded from the Rapid Capture project (built-up-areas, power lines and contours), would be completed. The Scitex workstations would be decommissioned soon after. The Mapsetter 5000 was decommissioned in A Contex FSS scanner (optical resolution 400dpi / interpolated resolution 800 dpi) would be acquired in 2000 to fulfil scanning requirements. The VTRAK system would commence with the vectorisation of the 1:10,000 orthophoto contours in May 1999, until it was decommissioned in November Two Intergraph IGeovec workstations would be acquired in 2001 to continue with this work. The spot height data capture, from the 1:50,000 map series, commenced in 2000 and was completed in The NTIS has been updated with 130 (81,250 km 2 ) cartographically captured map sheets where all NTIS features have been captured. The core component, responsible for the maintenance of the NTIS, currently consists of 16 GeoMedia Professional workstations (13 for data maintenance; 3 for data management, checking and plotting), 2 IGeovec workstations (vectorisation of 1:10,000 contours and other miscellaneous tasks) and 3 DEM Workstations (Digital Elevation Models (DEM) computation).

10 4. CURRENT TOPOGRAPHIC DATA CAPTURE SYSTEMS AND PROCESSES UPDATING THE NTIS The CDSM introduced complete digital topographic data capture processes and techniques into its production environments in 1996 with the first map sheet, 3030 CA St. Faiths, commencing capture on the 29 th of April All NTIS features per map sheet, excluding relief, are topographically captured for population in the NTIS. To date 430 (270,000 km 2 ) 1:50,000 map sheet areas have been compiled where 220 (137,500 km 2 ) have replaced the original cartographic data in the NTIS. A brief overview of these systems and processes, and the technologies used, will be presented here. A simple process flow diagram of the current topographic data capture process is presented in Figure 5. Figure 5. Production Process. The reader will notice tentative workflow process pointers within the scanning and rectification environments. The scanning of aerial photograph film negatives is required for use in the digital photogrammetric workstation (DPW)/digital softcopy environment, and the rectification of these images are required where they will serve as source data for monoplotter workstations. Alternatively, hardcopy diapositives are utilized for stereoplotter (analytical and analogue) workstations. Careful management of the data capture workflow is carried out to ensure that all available resources are utilized efficiently. Aerial photography is captured on an annual basis covering 25% of the country in order to update the National Imagery database and to serve as source for topographic data capture using photogrammetric techniques. Approximately 120 sheets (75,000 km 2 ) of the 1:50,000 topographic map series are updated annually. The CDSM maintains a place and topographic names database. This database contains names that have been approved by the South African Geographical Names Council (SAGNC). The database was digitally converted, and spatially referenced, in 1998, and has served as source for the publication of Provincial Geographical Names Gazetteers for each of the nine provinces, in The National Digital Elevation Model (NDEM) database was re-designed in 2001 in order to reference it to the new South African Datum, Hartebeesthoek94, and to remove errors, deficiencies and data overlap. The previous database was not consistent (photogrammetrically captured from 1:150,000 photography, causing rifts at model edges), contained duplicated data (the 200m/400m DEM datasets had two degree overlaps within the South African projection system) and had sparse coverage in rural areas (at 200m/400m intervals). The new database will be populated within an

11 Oracle database and will cover South Africa at a 25m interval. The elevation accuracy of the DEM will be ±2.5m (in areas derived from mass elevation data capture) and ±10m (in areas derived from 1:50,000 20m contour and spot height data). The various production systems utilized within the CDSM data capture environments are listed in Table 11. The CDSM is progressively phasing out its analogue stereoplotter workstations and replacing these with digital photogrammetric workstations. Analytical stereoplotter workstations will remain for the foreseeable future until such time that comparable image quality can be obtained within the digital environment. Table 11. CDSM Production Systems No. EQUIPMENT PRIMARY SOFTWARE TASK/S Aerial Triangulation Environment 2 PUG point transfer devices Transfer tie points on diapositives 1 BC2 stereoplotter Orima Aerial Triangulation 1 Z/I Imagestation Z3 ISDM and ASAE Aerial Triangulation and Mass Elevation Data* 2 Z/I Imagestation SSK ISAT Automated Aerial Triangulation* 1 A/T Adjustment Workstation PAT-B Block Adjustment with GPS airstations* Compilation Environment 4 AG1 stereoplotter usmart (Analogue) Topographic Data Capture 3 PG2 stereoplotter usmart (Analogue) Topographic Data Capture 2 A8 stereoplotter usmart (Analogue) Topographic Data Capture 1 Quasco PG2 stereoplotter usmart (Analytical) Topographic Data Capture 1 BC3 stereoplotter usmart (Analytical) Topographic Data Capture 7 usmart Softcopy usmart (Softcopy) Orthophoto Data Capture* 12 Monoplotter Workstations Geomedia Professional 5 Topographic Data Capture Database Maintenance Environment 16 NTIS Workstations Geomedia Professional 5 NTIS Data maintenance and structuring 2 Vectorising Workstations Igeovec Raster to vector conversion (contours etc.) 3 DEM Workstations 2xSCOP, Geomedia GRID Generate 25m DEM from mass elevation data and contours Scanning & Rectification Environment 1 Vexcel 4000 Scanner Vexcel proprietary Photogrammetric scanning of film negatives* 2 Rectification Workstations ERDAS Imagine 8.6 Rectification of scanned imagery using DEM* Orthophoto Mapping Environment 5 Orthophoto Workstations ArcView 3.2 Create Orthophoto Map Products Cartographic Mapping Environment 19 Cartographic Workstations LAMPS Create Topographic Map Products** * utilized in both the orthophoto and topographic data process workflows/** utilized for all scales of topographic map products 5. NTIS: SUCCESS STORIES The NTIS has been successfully utilized in a variety of applications and projects. I wish to highlight the most significant of these to demonstrate its usefulness and justify its creation. These are:! In 1995 the pilot (1:500,000) dataset was utilized as data source for the creation of maps for the publication of the Readers Digest Atlas of Southern Africa.! In 1997/8 the NTIS database served as reference for the determination of rural enumerator areas, in order to determine voting districts, culminating in the successful South African general election held in 1999.! The NTIS has, and is serving as base spatial reference for the development of the South African National Land Cover project 2000.! The South African component of the Global Map project, supplied in 2001/2, was sourced from the NTIS dataset and restructured to Global Map specifications.! The CDSM, in 2002, completed the digital conversion of Swaziland s entire 1:50,000 topographic map series, consisting of 23 map sheets, where the same procedures and processes, used to create the NTIS, were applied. 6. NTIS CONTENT: PROBLEMS AND SOLUTIONS The creation of the NTIS database required the capture of topographic feature entities within an area of 1,22 million km 2, at an accuracy of approximately 20 metres. Difficulties have arisen and these have been addressed within the constraints that existed at that time. Decisions made, inaccurate processes, technologies and other factors have led to erroneous and incomplete datasets within the NTIS.

12 I will highlight each of these and state how the CDSM has decided to deal with theme:! Incomplete feature content: In 1992 the CDSM made the decision to capture only the major communication, relief and hydrological features to complete the development of the NTIS within a 5 to 7 year period. The Rapid Capture project would omit further features in order to attain higher production rates to realise completion in 5 years and assist in Project Merge ;! Limited metadata content: The NTIS is primarily a cartographic database, captured from scanned cartographic film separates, and, therefore, suitable and sufficient metadata was not available or was not captured;! Positional errors in feature content: The commissioning of the Contex scanner, during the Rapid Capture project, would introduce positional errors within the NTIS dataset. This Contex scanner had inadequate rollers with which to secure the cartographic film separates during the scanning process creating distorted raster images; registration ticks were affixed to the film with wax resulting in movement of these during the scanning process and causing registration inaccuracies (these would later be covered with tape to secure them). At a later stage the registration ticks were scribed onto the film material, thus alleviating this error;! Minimal topological content: The NTIS is effectively a spaghetti dataset that has been edge-matched and where shared segments have been aligned, but are duplicated within the component entities; The CDSM has embarked on a Clean-up project in order to correct the relief and hydrological components of NTIS. Errors and content deficiencies have and will be addressed, in the long term (approximately 10 years), through the maintenance program of the 1:50,000 national map series where new edition photogrammetrically captured map features will replace these erroneous features. Error detection and correction of manageable components will be addressed as the need arises. The topological structuring of the NTIS will be addressed within a Data model investigation where additional and omitted features will be identified for capture and the implications of introducing an object model will be assessed and implemented. 7. CONCLUSION The development of the NTIS has been successful and is justified through its use by clients on a daily basis. The CDSM must now focus on the improvement of the database content and structure in order to ensure that it complies with national and international standards and remains relevant to our clients. In this regard the CDSM has decided to focus, in the short to medium term, on two areas of concern, firstly, to implement error detection and correction procedures and processes to ensure a topologically correct relief and hydrological national dataset and, secondly, to redefine the CDSM data model, incorporating all omitted datasets, where this data model will be implemented and populated accordingly. A clean-up project, of the relief and hydrological datasets, commenced in April 2003 and will be completed in September The result of this project will see these datasets being individually error free and logical in relation to each other. The relief dataset will serve as raw data to complete the population of the 25m national NDEM, where DEM, derived from mass elevation data capture produced via the orthophoto process, does not exist. The CDSM data model investigation will commence in June 2003 and be completed in September Here a conceptual, logical and physical data model will be defined encompassing required and available disparate technical datasets. The Physical Model will serve as a guide in the creation, population and maintenance of an integrated CDSM corporate technical data warehouse and will comply with relevant South African standards (SANS 1876: Feature Instance Identification Standard, SANS 1878: South African Spatial Metadata Standard and SANS 1880: South African Geospatial Data Dictionary (SAGDaD) and Its Application). The implementation of the Physical model will commence in October ACKNOWLEDGEMENTS I would like to express my sincere appreciation to my colleagues, Messrs. D Clarke, P Lucas, R Duesimi, R Swart and K Scott for sharing their knowledge and experience in the preparation of this paper. 9. REFERENCES [1] Hedling G, Parker A, Wonnacott R, ION GPS 2000: 13th International Technical Meeting of the Satellite Division of the Institute of Navigation, TrigNet - The Network of Active GPS Base Stations for South Africa, (2000) [2] Statistics South Africa, Population Census, 1996, Chapter 2: Methodology used in Census 96, (1998) [3] Municipal Demarcation Board, South Africa, About Demarcation, Demarcating Metropolitan And District Municipality Boundaries, (1998)

13 [4] Lester K J, Cambridge Conference 1999, Shaping South Africa's Future with GIS - The 1999 General Election Experience (Paper 2.4), (1999) [5] Vorster P J, The New South African (Hartebeesthoek94) Datum, South African Journal of Geo-Information, Vol. 7, Part (1998) [6] CDSM technical files: ME 9/3 vol. 1, 5-9; ME 9/1 vol. 1-2; ME 9/4 vol. 1-2; PRE 2/2/1 vol. 1

14 THE DEVELOPMENT OF THE SOUTH AFRICAN NATIONAL TOPOGRAPHIC INFORMATION SYSTEM Vorster, P.J. Chief Directorate: Surveys and Mapping, Department of Land Affairs, Private Bag X10, Mowbray, South Africa. Website: Biography The author received a B.Sc. Survey degree, from the University of Cape Town, in He took up employment with the Armaments Corporation of South Africa (Armscor), at its Overberg Test Range facility, where he was responsible for the establishment of an extensive precise three-dimensional geodetic network, referenced exclusively to the World Geodetic System 1972 (WGS72) spheroid, to serve as the measurement framework for the test facility s real-time measurement technologies. Here he would be one of the first in South Africa to apply the Global Positioning System (GPS), as a geodetic measurement tool. In 1992 he took employment with the Kwazulu government service in order to gain cadastral experience and to complete his articles towards attaining professional status. Here he would participate in large operations, utilizing GPS technology, in order to delineate tribal authority boundaries. He completed his articles and was registered as a Professional Land Surveyor in In 1994 he would be employed with the Chief Directorate: Surveys and Mapping (CDSM), South Africa s National Mapping Agency (NMO). In 1995 he was promoted to the position of Deputy Director and would play an integral role in the establishment of South Africa s new South African Datum, based on the WGS84 spheroid, officially adopted in He is currently responsible for the evaluation and application of current and new developments within the geodetic, photogrammetric and related fields and technologies at the NMO. His management focus is within the field of Geographic Information Systems (GIS) where he has been responsible for projects relating the maintenance of the NMO s national topographic information system (NTIS) and the integration of national geospatial databases in the public domain. He is currently in his final year of study for a B.Sc. (Honours) Geoinformatics degree, affiliated to the international UNIGIS programme.