Report on the Fourth GISDECO Conference

Pretoria, 1-2 October, 1998


G. Brent Hall, Faculty of Environmental Studies, University of Waterloo, Ontario, Canada

Pieter van Teeffelen, Faculty of Geographical Sciences, University of Utrecht, Utrecht, The Netherlands

Paul van Helden, Centre for Geoinformation Science, University of Pretoria, South Africa


This paper summarises the content of the GIS in DEveloping COuntries (GISDECO) Conference, recently held in South Africa.  The conference is the fourth to be organised on this general theme, with a specific focus, in this instance, on GIS tools for effective planning (in developing countries).  First, some general comments are made on the purpose of the GISDECO conferences and on the rapid pace of development of GIS technology in developed countries during the seven year period since the first GISDECO was organised. Next, the twenty-one papers presented at the conference are summarised around the key themes of 'Information for Development', 'Examples of GIS as a Tool', 'Decision Support Systems', and 'New Tools to Improve GIS Efficiency'.  The paper concludes with a summary of the role of GISDECO in the development process and plans for the next conference in two years time.

1           Introduction

The GISDECO Conference has been run every two years since its inception in 1992.  The fourth conference, held this year in Pretoria, South Africa, is the first that has been held outside of The Netherlands and its organisation was a collaborative effort by the Faculty of Geographical Sciences of the University of Utrecht, which organised the first three conferences, and the Department of Town and Regional Planning at University of Pretoria, South Africa.  The purpose of GISDECO is to provide a forum for the exchange of ideas and progress in problem solving for GIS researchers and practitioners, either from and working in developing countries or from developed countries and working on the problems of GIS technology transfer and spatial analysis applications in developing countries.  The general objective of these two broad groups is common, namely to assist the development process and improve the quality of life of those populations who suffer from lagging economic and social development, while safe-guarding the fragility of the bio-physical realm in truly and newly underdeveloped countries.

With this in mind, the themes of the four GISDECO conferences reflect a gradual evolution toward maturity in GIS research in developing countries and on development problems.  The first GISDECO conference, held in 1992, had 'Possibilities and Constraints' as its theme.  This was followed in 1994 with the theme of 'Problems and Challenges' and in 1996 with 'The Practice of Applications', leading to the most recent theme, noted above.  From the outset, it was unclear what possibilities existed for successful GIS implementation in developing countries, as the constraints of success seemed to countervail the likelihood of achieving successful problem solving.  This theme evolved two years later into a focus on the problems that existed in achieving the goals of development (e.g. equity in the access of all populations to basic goods, such as an adequate diet and shelter, and services, such as health care and education; preservation of natural environments through practising sustainable resource exploitation) and the challenges that confront the quest to overcome these problems.  Subsequently, the third conference promoted presentation of research that encouraged practical problem solving through the use of GIS technology in developing countries.

During the seven year period that GISDECO has been operational the nature of computing technology and indeed the nature of the GIS software industry has changed a great deal.  At the time of the first conference, ESRI's ArcView desktop GIS had not been released (this software is now up to version 3.1, with significantly enhanced functionality which has changed the nature of desktop GIS); IDRISI software, which has a strong presence in developing countries (largely because of its (formerly) low price and its support through the United Nations Development Program), was running only in DOS (Version 2 for Windows is now released);  MapInfo desktop GIS was in release 1.0 and its functionality in this version could be described as embryonic, at best; Microsoft Windows 3.1 operating system (OS) had just been released and was being adopted in developed countries as the standard OS for personal computers (yet most GIS software at this time ran only in DOS); Microsoft NT was not released (this and Windows 95 are now the standard OS for much of the commercial GIS market); macro GIS programming was significantly limited in its functionality on personal computers (primarily to ESRI's Simple Macro Language, running in DOS and MapInfo's MapBasic); remote sensing applications were limited, largely, to the UNIX and VMS domains; and, finally, the Internet did not exist as we now know it and the World Wide Web (WWW) had yet to make an appearance on the computing scene.  Now, the situation has changed dramatically. For example, it is not uncommon to find, in developing countries, very powerful personal computers running at least Windows 95 (and often NT).  The GIS software market has also grown, with most vendors porting their UNIX software to parallel NT development, and particularly in the areas of end-user applications programming with the availability of improved macro languages (such as Avenue in ArcView) and GIS-based mapping libraries (OCXs), such as ESRI's Map Objects and MapInfo's MapX.

Clearly, we are now dealing with an era of GIS applications work that is significantly different to what was possible when GISDECO was first launched.  However, despite this rapid transformation of the industry, many of the critical issues that created the need for a conference such as GISDECO are just as valid, today, as they were close to a decade ago.  That is, although the technology has improved substantially, our ability to solve basic problems of development through technology have not been able to keep pace.  In particular, concerns with spatial data availability and standards in developing countries, with end-user training and administrative awareness of the possibilities that exist with this technology, and with moving beyond the development aid and assistance industry syndrome to self-management and sustainability in the use of GIS technology, are still major problems in most developing countries.  Hence, GISDECO remains as important as it was at its inception as a forum for discussing the trends in GIS use in developing countries and the contributions of GIS technology to the development process.

In total, twenty one papers, excluding the opening and closing addresses, were presented at GISDECO IV and twenty three papers are published in the conference proceedings[1].  The papers are categorised into four sections headed, respectively, 'Information for Development', 'Examples of GIS as a Tool', 'Decision Support Systems', and 'New Tools to Improve GIS Efficiency'.  Ten posters, of excellent quality, were exhibited at the conference with all but one (which displayed work in progress on the integration of Radar imagery with ground-based GIS to detect the morphology and spread of pockets of poverty in Rosario, Argentina) dealing with South African GIS topics.

There were approximately 200 registrants for the conference, the vast majority of whom were from South Africa.  Other countries represented by registration included The Netherlands, Portugal, Greece, Austria, Canada, the Philippines, Vietnam, Kenya and Zambia.  The absence of more participants from African countries, other than South Africa, was somewhat of a disappointment to the organisers, but generally this is a reflection of the language barrier (especially from French speaking Sahelian countries) and the limited funding opportunities available to GIS professionals from truly developing countries in Africa, Latin America and Asia to attend and participate in conferences such as GISDECO.

Certainly, the sophistication of GIS use in developing countries has improved during the life of GISDECO, as reflected in the summary of papers presented in the following section.  There remains, however, a great left to be accomplished and this is reflected in the proposed theme of 'Access to GIS for all through education and technology use' for the next GISDECO, proposed to be held at the International Institute for Rice Research headquarters in the year 2000, in the Philippines.  In moving the venue of the conference to Asia, it is hoped that greater participation will be achieved from GIS professionals in the truly developing countries of this continent, as well as embracing more of the community of scholars from elsewhere whose work is focused on GIS technology-based solutions to the problems of the development process.

2           Summary of Papers

In the opening address to the conference, Pieter van Teeffelen (van Teeffelen and Kwant, 1998) placed the contributions of GIS as a tool for effective planning somewhere between 'promise and reality'.  He argued that GIS technology (including hardware software and databases) has certainly diffused from the core regions of North America and Europe, largely through aid efforts and the work of international development agencies, to the developing world.  Yet, only 10% of all licensed GIS software internationally is found outside of the core regions (including Australia and New Zealand).  Of this 10%, 55.4% of the licenses are in Asia, 33.7% are in Latin America and Africa lags far behind with only 10.9%, or roughly 1% of all international software licenses.  van Teeffelen noted that it is difficult to suggest that any one of the three developing continents is doing better than the others with respect to GIS use, but there are pockets on a subcontinental scale (at the individual country level) where discernible progress is being made.  This was discussed by example with specific reference to each continent.

In Africa, much of the GIS work is focused on problems of resource management.  However, the classic problem of aid agency donors dictating which GIS software is purchased and used tends to preclude the opportunity for slowly building indigenous expertise in GIS technology and undertaking proper needs assessments prior to implementation. The problems to be investigated, the hardware and software environments to be used and the nature of training and post-project support must be decided as much by local participants as by expatriate or foreign experts.  Hence, GIS projects in the truly developing countries of Africa (which includes virtually the entire continent) have a tendency to fall somewhat short of their promise through difficulties in sustainability, as they are not client driven in the same way they are in the developed world.  The clear exception to this rule is South Africa, which has institutionalised GIS instruction at the tertiary level throughout the country and also in most national and local government offices.  The nature and breadth of GIS use in this country is exemplified in most of the papers presented at GISDECO IV.  However, it is incorrect to suggest that South Africa is without its own GIS teething problems, especially in the areas of sub-national data availability and data quality.

Van Teeffelen noted that GIS use is relatively widespread throughout Latin America, and in several countries, such as Chile, Peru and Brazil, most of the GIS development work is truly indigenous with considerable progress having been made in specific areas.  However, he went on to note that the same problems evident in Africa are also evident in this continent, albeit to a lesser extent.  Lack of finance and lack of trained personnel are less problematic in Latin America, and the GIS industry is more typical of  a community of relatively advanced users with constraints typical of advanced users elsewhere.  There remains considerable sub-regional differentiation, however, with six countries (Mexico, Venezuela, Brazil, Colombia, Argentina and Chile) accounting for 95% of the continental IT market of the region (including the Caribbean).

In Asia, GIS use is widespread throughout the continent, fuelled by very active promotion and development of information technology in certain countries.  For example, India, Thailand, Japan, and Korea all have highly developed GIS installations at national and sub-national scales, while the city states of Singapore and Hong Kong are highly advanced in their development and use of GIS for planning purposes.  Further, China is emerging as a major contributor to the international GIS industry despite the loss of many highly talented and skilled software programmers, especially to North America.  The same duality apparent in Latin America is also evident in Asia.

The opening address concluded with the observation that political readiness and the need for politicians and public servants to imbed GIS technology right into national and sub-national development planning is required, among other factors, to advance problem solving with this technology in developing countries.  With this in mind, attention is now turned to the individual paper presentations.

2.1          Information for Development

The first paper in this session, by Craig Schwabe (Schwabe, 1998), was heavily focused on data standards and the need to establish, for South Africa, a workable spatial information model with core, secondary and other useful data sets that, in turn, define its form and content.  Beyond this, the need to establish a protocol for the dissemination of spatial information that includes attendant spatial decision support software to encourage information use was discussed.  Several innovative  initiatives in South Africa to these ends were identified and importantly, Schwabe noted that 'to effectively implement a spatial information model for the reconstruction and development process in South Africa requires the development of a partnership between government, service providers and major research councils in the country'.  He illustrates in his paper some of the uses to which spatial information can be put in the context of South Africa.

Schwabe's paper intersects with and is built upon in the second paper, presented by Andre Brits (Britts, 1998).  In this paper, an integrated development information system (IDIS) for South Africa is described and discussed.  The appealing concept of 'unlocking' geo-spatial data is introduced in the context of spatial data liberation strategies elsewhere in the world and the National Spatial Information Framework  (NSIF) in South Africa.  Brits describes the creation of 'information co-operatives', based on the information communities model of the Open GIS Consortium (Buehler and McKee, 1998), to facilitate data sharing and the general issues involved with this.

The third and fourth papers, by Martin Landré and Anthony Walker move GIS problem solving down from national data/information issues to the local level.  First, Landré (1998) addressed the issue of data for local economic development and showed several examples of local-level GIS problem solving (with IDRISI).  One of the examples he presented is notable in that it was one of two instances in all papers presented that employed or proposed using multiple criteria analysis (MCA) in a loosely-couple GIS decision support application.  Walker (1998), on the other hand, presented several sets of census-based, socio-economic data analyses (using ArcView) that allowed visualisation of relative deprivation of the coloured population in Port Elizabeth, South Africa.

Gaby van Wyk, returned to general information concerns in the final paper in the first session, by challenging many of the conventions of cadastral mapping in relation to the 'parcel life cycle' and its attendant implications for maintenance in a GIS (van Wyk, 1998).  The life cycle of a parcel was presented in terms of three types of base mapping and the parcel concept was reconstructed into a more relevant and workable spatial entity than is often the case.

2.2          Examples of GIS as a tool

A problem that confronts urbanising areas in all developing countries is the issue of controlling the establishment and growth of informal or squatter settlements of landless migrants, many of whom arrive from rural areas.  The paper by Kevin Mearns (Mearns, 1998) addressed this problem in Harry Guala Township, East Guateng, South Africa.  IDRISI software was employed again in a simple but effective application to identify the land in the township best suited to settlement and thereby formalising the development of informal housing for landless families.

The second and third papers presented in session two, respectively by Cheri Green and Peter Schmitz on behalf of their co-authors (Green, Morojele, Maritz, 1998; Cooper, Schmitz, Potgeiter, 1998), each deal with the issue of crime and personal security in South Africa.  Green et. al. (1998) analyse the accessibility of populations in low income areas to police facility locations using a loosely coupled suite of GIS and data processing tools.  A nearest centre covering algorithm was used in their analysis to allocate population to police centres until capacity at a centre is met, subject to a maximum transport network (travel) constraint.  The approach requires manual identification of new police centre locations, but offers an effective means of visualising spatial coverage of and gaps in policing in high crime areas.  Similarly, the paper by Schmitz et. al. allows the locations of various types of crimes in Johannesburg to be visualised using a combination of ArcView and MapInfo GIS software.  A particularly interesting aspect of this paper, indicating the extent to which GIS application development has matured in South Africa, was the suggestion to use WWW-Internet map server delivery of up-to-date crime location maps to link suburban police stations with information stored in a central GIS Web server.

Sharon Biermann (Biermann and Whisken, 1998) presented next a model for the calculation of the potential cost of bulk infrastructure, such as water, electricity and sanitation, provision during the early stages of land development planning.  Although no results are of the application of their model are presented and discussed in the paper published in the proceedings, the results of an application of the methodology for the Greater Pretoria area were shown in the presentation.  Consideration of the cost estimates of service provision, taking into account the effects of land and location (among other factors), is particularly important at the strategic planning stage of land development in developing and developed countries alike.

The first day of the conference was brought to an end with two papers, presented by Andries Naude and Juanita Moolman on behalf of their co-authors (respectively, de Jong, Naude and van Teeffelen, 1998 and Moolman and Quibell, 1998).  In the first paper de Jong et. al. used a loosely coupled suite of GIS and related tools, similar to that discussed by Green et. al. (1998) to examine the accessibility, by catchment calculation, of rural residents to services such as education and health care.  Moolman and Quibell, on the other hand presented a GIS-based approach to calculating sediment yield in a river basin in Northern Province, South Africa.  While the former paper shows promise for refinement and future tighter coupling of the tools used into an integrated MCA-based decision support tool for service planning, the latter paper suffered somewhat from the small map scale at which the data were assembled (1:250000 (land cover) and 1:500000 (soils)).

2.3          Decision support systems

The second day of the conference began with six paper presentations focussing on the use of GIS as decision support systems.  The first paper, presented by C. T. Hoanh (Hoanh and Kam, 1998), described an expert system-based approach, using IDRISI for windows as the mapping and spatial data platform, to analyse the balance between rice supply and demand in several Asian countries.  Results were presented for North Vietnam that compared the sensitivity of the analysis to grid cell size in a raster GIS database.  The paper showed that GIS has a potentially important contribution to make in analysing rice yields, but serious questions remain concerning spatial data inputs.

Following the paper by Hoanh and Kam, Adlai Davids (Davids and Grundlingh, 1998) described a simple but effective visual analysis, using ArcView, of the dislocation between health centre supply (location) and revealed access to care for women using birth control services in the Eastern Cape of South Africa.  While the results of their analysis were presented at the conference, their written paper regrettably considers only health care data issues, which were shown to be considerable in the analysis displayed in their presentation. Stefan Kollarits (Kollarits, 1998), presented in the third paper an overview on issues relating to transportation planning and the use of GIS in developing countries.  Issues in this area in developed countries were used as a means of contrasting the challenges faced by transportation planning in developing countries.  This was followed by a complementary paper by Rodney Steinhofel (Steinhofel, 1998) on the role of GIS in transport planning in Cape Town, South Africa.  Again, unfortunately numerous useful examples of GIS use in route and service planning were shown and discussed in Steinhofel's presentation, however the written paper considers only general issues without empirical support.

The same pattern for the previous paper was evident in the contribution by Peter Gildenhuys (Gildenhuys, 1998), in that the tabled paper contains, at best, a summary of the work presented to the conference attendees.  Gildenhuys presented an environmental management framework containing numerous spatial data sets that, combined, allow the sensitivity of areas to be scoped in terms of their development potential.  A particularly interesting aspect of Gildenhuys work is the development of a dedicated distributable data browser (written in Visual Basic using Map Objects) for end use.  With the advent of ESRI's Internet Map Server, future development of this work could easily see data sets made available for end use over the Internet.

The morning session of papers was rounded out by a paper presented by Mike Webster (Webster and Hounsome, 1998) on the use of GIS as part of a strategic environmental assessment with an application for Cato Manor settlement, located inside metropolitan Durban.

2.4          New tools to improve GIS efficiency

The afternoon and final session of day two featured three remote sensing papers and a final paper dealing with data-sharing through the WWW.  The first paper, by Thomas Blaschke (Blaschke, 1998) was a review of the potential contributions remote sensing imagery can make to addressing development problems.  The paper provided a comprehensive summary of the range of remotely sensed imagery available for GIS-based analysis and provides several general examples of the types of end-uses to which such data can be put in developing countries.  Papers two and three built upon this overview.  First, Suan Pheng Kam presented a paper co-authored by herself and 5 colleagues (Kam, Minh, Tuong, Hoanh, Liew and Chen, 1998) on the use of remote sensing data for the analysis of changes in rice cropping systems along the Mekong River delta in south Vietnam.  This paper showed clearly, the useful contribution of radar imagery to detect the various rice cropping systems used in the study area.  However, the authors also note that integration of remote sensing analysis with GIS and detailed local knowledge would enhance significantly the usefulness of their approach.

The second remote sensing paper, presented by Steven de Jong on behalf of three colleagues (Bagré, Deursen, de Jong and van Teeffelen, 1998), examined the use of multi-spectral (as opposed to radar), multi-temporal imagery to identify trends in urban growth and land use for Ouagadougou, Burkina Faso.  Their use of SPOT-XS imagery with a SPARK (SPAtial Re-classification Kernel) classifier produced promising results at a general level of land use classification, but the spatial resolution of 20 metres was found to be insufficient to identify clearly areas of high spatial variability and heterogeneous land use.  Certainly, however, both papers (Kam et. al. and Bagré et. al.) indicate that, finally, interoperability of GIS and remote sensing approaches may yield promising avenues of work in developing countries.

The third and final paper presented at GISDECO IV by Martie Pienaar (Pienaar, 1998) illustrated the possibilities that exist for spatial data sharing via the WWW.  It could be argued that the need still exists to create good quality and current spatial data bases in most developing countries and only after this need is satisfied can data be distributed via a medium such as the WWW to end users.  South Africa, as evidenced by several of the papers presented at this conference, is at a stage where dissemination of some databases can be initiated in certain areas via the Web.  However, in this regard, South Africa is very different from most other newly and truly developing countries.  Issues that concern data standards and data availability (as discussed by Schwabe, 1998 and others at GISDECO IV) still need to be settled before this becomes a widespread reality.

A final paper on the contribution of GIS to regional planning in Greece (Karnavou, 1998) is tabled in the conference proceedings, but was not presented.  The closing plenary of the fourth GISDECO presented by Brent Hall is summarised in the current paper.

3           Conclusion

The overall quality of the research presented at GISDECO IV reflects, more than anything else, the relatively advanced stage of GIS use in South Africa.  This should not give us occasion to fall into a sense of false self-congratulation and believe that achievements are similar elsewhere in the developing world.  There is certainly a great deal to do, particularly at the respective ends of the public spectrum (grassroots and bureaucratic decision making), to bring the potential contributions of GIS technology to fruition in developing countries.

As noted above, it was announced at the conference that GISDECO V is slated to be held in the Philippines in two years, with the hope of encouraging greater participation from Asian researchers.  The issue of the theme for the next conference was discussed in the closing plenary and several interesting observations were made by conference participants.  First, the issues of greater awareness or bringing people back into GIS were discussed.  A common point made, not only in the forum of developing countries, is that knowledge of the potential contributions that GIS technology can bring to the development process is not well enough disseminated and understood among, at one end, the recipients of planning and development (i.e. the public at large) and, at the other end, those who govern the planning and development process (i.e. senior administrators or bureaucrats and politicians, who wield considerable power but typically lack technical knowledge).

One means of achieving increased GIS awareness is through education and training for all and facilitating this is a reasonable objective for GISDECO V to strive for, in conjunction with providing a forum for exchange of research completed and underway on GIS use in developing countries.  Hence, it is proposed to preface GISDECO V with a series of workshops targeted at different levels of users and various end-use applications.  This would include hands-on, highly focused three day training workshops for the higher level end-user in, say, the use of macro programming in Avenue script to customise ArcView or the use of Visual Basic code to create Web-ready browsers for spatial data delivery across the Internet.  For those who have no intention of becoming a GIS user, but who want to understand better what this technology can and cannot do, a complementary set of workshops is envisioned on, for example, what a decision maker can expect from a GIS for given levels of expenditure of time and other scarce resources.

The training workshops would be followed by the standard two day conference presentations and attendees could arrange their time to participate in both a selected workshop(s) and conference, or simply attend one or the other.  The brings a new dimension to the way in which GISDECO has been traditionally organised but perhaps better reflects the types of needs that exist in the parts of the world the conference is targeted at.

4           References

Buehler K. and L. McKee, 'The OpenGIS Guide; introduction to interoperable geo-processing and the OpenGIS specification', 1998.



[1]  Individual papers or the full conference proceedings are available at cost of production and mail from: Professor Paul van Helden, Centre for Geoinformation Science, University of Pretoria, Pretoria, South Africa.