Lauretta Burke
Abstract
This paper provides an overview of issues and considerations for implementing a GIS to support urban and municipal planning in developing countries, supported by descriptions of project experience. Topics include data availability and access, organizational and institutional issues, staffing and training, hardware and software selection and support. Project experience includes a rapid land use assessment application in four cities in the Philippines, and a municipal planning project in San Pedro Sula, Honduras.
Introduction
There is a casually quoted statistic that roughly half of all GIS implementations fail. Most failures are related to institutional issues, resistance to change, lack of political support, insufficient funding, and the fact that GIS innovation results in a radical change in information flow within an organization (NCGIA, 1990). Most examinations of GIS implementation success have focused on developed countries, where user support for hardware and software, availability of trained GIS professionals, and access to a reliable power supply are not problems. The considerations relevant for any GIS implementation are compounded by additional circumstances and constraints in developing countries. This paper presents a short summary of many of the key considerations for implementing an urban GIS within a developing country. Issues presented are drawn from the collective experience of a team of GIS professionals working for a consulting firm specializing in international development. Issues include project design, institutional considerations, people (both technical staff and end users), training, physical site condition, hardware and software selection and support, data availability and access, application development, operating budget, and data sharing. The paper concludes with project experience which highlights some of the benefits of appropriate use of GIS and satellite information technology for urban planning in developing countries.
KEY CONSIDERATIONS FOR GIS IMPLEMENTATION
Project Design
The spatial information component of a project needs to be an integral part of overall project design from the beginning. Insertion of a GIS into a project design as an afterthought or add-on will rarely be successful. GIS implementation is difficult. Even when a GIS can be well executed from a technical point of view, project design strongly influences whether the information products generated will be effectively used. Timing of the user needs assessment, training, data collection, pilot project, and full project implementation are critical to gaining institutional support and to ultimate project success. An awareness of spatial data products and analysis capabilities typically needs to be cultivated in end-users early-on in order for these products to be used to their fullest.
The user needs assessment is a vital component of GIS implementation within a municipality. Thoroughly exploring potential data sources, integrating the GIS with more traditional information management within the municipality, and promoting an understanding of spatial information and analysis capabilities early-on are critical to project success. It is also important to explore whether there have been previous GIS initiatives within a city, and whether there are any project remnants (equipment, trained staff, or digital data sets) which can be utilized.
Institutional Issues
Whether the GIS will be controlled by a single existing institution, shared by several institutions, or be housed within a new entity established for particular project is an important consideration. Initially, GIS installations are probably most successful when centrally managed, but are later more effective with a decentralized structure. An effective structure involves the mayor of the city establishing a new body which draws upon the resources and staff of core groups around the city. This structure facilitates data sharing and use. GIS is still a new technology, and resistance to change is natural and widespread. There needs to be sufficient political support within the host institution to make the GIS installation a welcome change from existing information management. Pilot projects and pretty maps are often useful for generating institutional support. The availability of appropriately qualified staff within one or more institutions for GIS training and to support the GIS installation on an ongoing basis is another important issue. Staff selection and availability for training and implementation can be a highly political issue. Those implementing the project need to have reasonable control over staff selection.
People
In examining the people to be involved with a GIS project, one needs to consider both technical GIS staff and the planners, managers, and other end users of the information. Appropriate training needs to be provided for both.
It is a fallacy that GIS technologies are too complex to be assimilated in remote, developing country environments. Basic competency in PC-technology is widespread. Experience with the basics of mapping are common, while experience in digital cartography is not. Overall, the background necessary for learning to use GIS tools is present.
Consideration must be given to what provides the motivation for staff trained in GIS to stay with an institution, rather than joining the private sector. Long-term incentives for skilled technical staff must be a part of the project design. A single key individual can make or break a project.
Training
Within developing countries, there is a lack of higher-level training in spatial data analysis and GIS. Universities and other institutions of higher education must become more involved in GIS. Technical seminars, exchange and mentorship programs are being initiated by the private sector on an increasing basis. More development aid funds need to be allocated for educational programs, rather than for GIS implementation by outside consultants.
Application-based training is an effective means of fostering expertise, developing a spatial data base, and generating local commitment to a project. An effective approach involves providing introductory GIS training, followed by a database development which is structured as training. Window's-based spatial data set query and display software (such as ArcView version 1 or Idrisi) provides a valuable means of introducing basic spatial data and GIS concepts.
Physical Site Conditions
Access to a reliable power supply is still a major problem in many developing countries, though this is improving in some urban areas. The use of voltage regulators and universal power supply (UPS) units is critical to safeguarding hardware and mitigating work loss and stoppages. In environments where adequate office space may be scarce and heating and cooling systems may be less than adequate, working conditions can be troublesome.Hardware Issues
Until recently, hardware in developing countries was expensive, and difficult to purchase and maintain. Developments in the PC market have led to faster and cheaper machines that support multiple operating systems. Peripherals remain costly and difficult to repair. Maintenance and technical support continue to be problems, although the development of local markets has begun to help. PC-technology is most often the appropriate choice for municipal-scale projects in developing countries. As with any project, it is important to evaluate the user needs, and pick hardware appropriate to the project, the long-term goals of the installation, and which can be realistically supported.Software Issues
GIS and image processing software are still not up to par with other software. Software vendors are beginning to address this, largely due to market and user forces. GIS software is becoming less expensive and easier to use. The growing desktop environment, especially Windows, has forced traditional GIS vendors to change and prices are beginning to fall. Local technical support in developing countries is inconsistent, though local firms are increasingly emerging to address this need.In some cities, hardware and software support will not be much of a problem, though few firms have extensive experience with GIS software. In general, some external method of information access - home office support or communication with support services is a good idea. In the past, too much of the emphasis for GIS development is given to software issues. Long-term flexibility and compatibility are key considerations. Software selection should take into account local experience and existing digital data (if any), in addition to short-term needs of the project and long-term goals of the GIS installation.
Data Availability and Access
In general, there is a lack of accurate, current data for urban areas in developing countries. Projects often rely on a catch-as-catch-can basis, where the best available data of a reasonable scale from a variety of sources is integrated. This is essentially a make-do approach.Satellite imagery is a significant alternative source of data for development and maintenance of data layers for a municipal GIS. Digital satellite data is often of overwhelming size and format for installations in developing countries. The costs are also often prohibitive. Manual interpretation from hardcopy images is a valuable alternative, which requires as little image processing as possible. This technique is useful for developing land use classifications, for developing road and hydrology networks which can serve as a base reference for integrating additional data sets, and for the development of point-based property information systems. Low-tech application of high-tech solutions is often appropriate. The availability of accurate sources for real world coordinate data is often a problem. GPS is invaluable for image rectification, data integration, and data collection. GPS can be effectively utilized in developing country environments.
Satellite Data Issues
There is a growing need for high-spatial-resolution data. High resolution sources exist, such as SPOT 10 and 20 meter resolution data and Russian KVR 2-3 meter resolution photography. These are valuable data sources for many areas. Cloud cover remains a problem for space-borne sensors in the tropics. Beyond frequent revisits and dry-season collection, there is no real solution. The cost of both digital and hardcopy high resolution satellite data products is still prohibitive for most developing nation users.Application Development
GIS installations in developing countries are mostly donor designed and implemented exclusively by consultants. Local authorities are demanding and receiving more control over project design and implementation. Applications are becoming more the product of local entities as local markets become more developed, local staff become better educated, and as "imported" applications fail, end-users need to be thoroughly engaged in the design process. Within donor-funded municipal GIS projects, the technical capabilities most often established relate to basic competency in GIS--database development and maintenance, simple analysis methods, and map production. The larger and lengthier investment in training required to transfer a full understanding of more sophisticated spatial analysis methods is usually beyond the scope of short-term, tightly-funded GIS components of larger projects.Operating Budget
There is a lack of ongoing funding and institutional commitment supporting the use of GIS and satellite data. Too often, GIS installations are only funded for the life of a project, and become word processing and game centers once the visiting consultants return home. The development of government-independent funding mechanisms, such as trust funds and foundations has helped to provide continued support. Also in response to this problem, there has been an emergence of GIS development sites which are public / private joint ventures. GIS components of projects are to often under-funded and are sometimes structured as a luxury option' within a larger project. Overall, there is too little investment in information development. Decision makers must be educated about the nature of spatial data and the need for and benefits of continuity, so they can understand the need for long-term support.Communications and Information Sharing
"Open" user communities are forming that are based on data sharing and technical exchange. Data is still, too often, viewed as a commodity, and is hoarded. Information is power. The project design needs to consider disincentives for data hoarding and incentives for sharing. Arrangements where contribution to the data pool affords access to the resulting data pool is the type of incentive which needs to be negotiated. Also, participation by NGO's and the private sector in the use and production of spatial products needs to be encouraged. In several cities, private entities have purchased municipally-developed digital data. Users often generate non-standardized, un-shareable data sets. Data standards, dictionaries, and national data catalogs must be developed and will evolve out of need. Local governments must communicate their needs and experiences with the national agencies developingAPPROPRIATE METHODOLOGIES
Simplified and relatively low-tech approaches to data generation, automation and consolidation are often effective in developing country environments. PADCO has developed and applied streamlined methodologies for land use classification and land information system (LIS) development.Rapid Land Use Assessment
A streamlined approach to land use classification called the Rapid Land Use Assessment (RLA) methodology assesses and quantifies land use through a combination of satellite imagery, local field knowledge/ expertise and resources, and simple GIS techniques. This methodology has been used by PADCO in Honduras, Indonesia, Pakistan, and in four cities in the Philippines. In the case of the Philippines, USAID sponsored a study that has utilized 2-meter-resolution Russian satellite data to inventory land use patterns over large areas and integrate this information into a GIS. Such high-resolution data is valuable for developing information on land use and roads. In rural areas, the imagery could also be used for identifying property locations.In Honduras the RLA methodology allowed the municipality of San Pedro Sula to determine and quantify by area 18 categories of land use for a 900 square kilometer area in three weeks using SPOT 10 meter satellite data and local staff. Image maps of the satellite data were processed in the US and taken to Honduras to be used as black and white ortho maps for interpretation. Utilizing simple aerial photographic techniques, field surveys, and local knowledge of the area, the land use database was constructed. Once the data had been interpreted it was then transferred digitally (through a process known as digitizing) by trained local municipal staff into a simple, PC-based GIS. This resulted in an accurate and up-to-date land use classification for the municipality. Road networks and surface hydrology data layers can be developed through a similar process. These preliminary data layers can serve as a base for development of a rich database for environmental and land use planning.
Rapid Land Information System Development - Lots by Dots
PADCO has developed a streamlined approach to LIS that incorporates key data into a land registration structure, subsequently transferring them into a fully automated information system. Under this approach, a land parcel is not stored as a polygon or area in the LIS and hence is not used as the base framework for the related database. Instead, a single point feature representing each property is used as the identifier and geographic locator - hence, "Lots by Dots".This is a critical difference from the polygon-based approach which attempts to reconcile geometry and compile all land parcels together into a contiguous map of polygons. This is practically impossible to achieve even in the United States since individual surveys of land parcels or 'lots' are often, not entirely accurate and do not actually reconcile with one another. In developing regions, property boundaries for land parcels are generally ill-defined with no accurate survey information for the parcel boundary. Parcels can, however, be rapidly located in the field and integrated into an LIS as a point. Subsequently, the cadastral information can be associated to the point location. In this way, the related plot survey information (not the graphic boundary polygon) is used to describe the essential features of the property.
This point-based property database can be developed from hardcopy very high resolution satellite imagery or aerial photographs, or through ground-based collection using Global Positioning Systems (GPS) techniques. As the field person collects the geographic (point) position of each land parcel, the property identifier number as well as physical characteristics of the land such as: land cover, soil condition, and number of structures can also be logged. This effectively allows an LIS database to be built in the field during survey. Additional complimentary data can also be integrated into the LIS, such as scanned property documents. The benefits of this point-based LIS versus a polygon-based system include reduced time for input and reduced processing overhead as the point databases are smaller and faster to manage, analyze, update and use than polygon databases. All relevant cadastral information is stored within the database just as it would be with a polygon-based system, only no resources are expended on resolving land geometry problems during database and system development.
Conclusions
Successful GIS implementation in developing countries is dependant upon good project planning, cooperation, and common sense. Externally-initiated (donor-funded) GIS projects and installations must be supported in a sustainable fashion, and must be driven by the needs of local agencies themselves.Acknowledgments
Thanks to Peter Rabley, Melissa Connor and Roman Pryjomko for sharing their GIS implementation experiences with me, and to Carol Baumann for a timely review of this paper.End Notes
A description of PADCO Geographic Information Management Group's projects will be made available at the Esri User Conference. Information is also available directly from PADCO at the address listed below.Reference
NCGIA, 1990. Core Curriculum for Geographic Information Systems, (Michael Goodchild and Karen Kemp, eds.), National Center for Geographic Information and Analysis, Santa Barbara, CA.