GIS for Urban Planning in Developing Countries

The future of GIS for planning in developing countries is dependant on building tools that meet the local needs of planning practices for data manipulation, decision-support, visualization and policy analysis (Brail and Klosterman, 2001). Strategic information resource management and the integration of appropriate technologies for effective planning are, therefore, key to the diffusion of GIS and GIS-based capacity building for planning applications. The availability of both qualitative and quantitative data remains a major constraint to planning efforts in the City of Beira. City planners can overcome some of these problems by using simple techniques such as controlled surveys and extrapolation to build a GIS that will allow them to gain a snap-short of prevailing conditions, identify hotspots, and, to monitor change. A complementary approach would be the use of remotely sensed images of land use types (e.g., housing) in conjunction with existing official records such as city infrastructure layout maps to determine access to facilities such as water, sanitation and electricity, provided the capacity exists to manipulate and interpret such data.

The potential for geographic information systems to assist urban planners in data poor environments that are experiencing rapid change has been demonstrated with this research. For example, valuable spatial information about housing quality, population density, access to critical infrastructure, services, and land use/land cover patterns has been presented. In the case of water, GIS buffering shows poor access in densely populated neighborhoods and generally poorly serviced squatter communities. There is also a concentration of population growth near the city center and this compounds the strains of urban growth and change. There is also a clustering of population near transportation nodes and networks because of the poor transportation options available to Beira residents. The Beira GIS also indicates significant urban agriculture, and this is consistent with informal housing growth and a steady migration from rural areas. These gardens are an important source of nutrition for residents and planners must recognize this spatial reality and the importance of urban agriculture for peoples' livelihoods.

These examples suggest that GIS technology is appropriate for local planners and planning agencies in Beira. The hardware and software are already available locally and CIDDI-UCM has become a high quality depository for digital spatial information for Beira and all of Central Mozambique. Our conclusion is that GIS is an appropriate technology for city planning in Beira and similar urban areas in developing countries.

Data Issues

Several data issues faced this project, both anticipated and unanticipated. The expectation of many northern institutions that readily available data exists for cities in developing countries cities is inaccurate. The data needs for the list of urban indicators drawn up by the United Nations can be easily met in the developed world, but the case is very different in most developing country situations. This is especially relevant in data-poor African contexts. The following are the most significant data issues faced in this project.

Incomplete meta-data: Directly related to issues of poor data quality are those regarding meta-data. In many cases the data was simply not explained. Where coverages had complete databases, the explanation of the database was absent. Documentation on data creation methods is completely absent and documentation of data quality is being generated, as it also does not exist. The absence of this information, while not proving to be insurmountable, contributes significantly to delay and generation of the information is more time consuming at this later stage than at the time that data are created.

Data quality: Where data exist, they are often of poor quality. A simple satellite underlay showed gaping holes and significant errors in the data.Cross-compatibility is a second data quality issue. Several coverages simply could not be opened together. Projection errors were most commonly the case of poor cross compatibility, but some coverages were captured in different reference systems (about one-fourth of the coverages were captured in the Clarke reference system while half  were captured in a WGS 84 reference system).

A second major data quality issue is that of incomplete data sets. While the GIS features were present in the coverages (line, polygons, points), much of the database relating to these features was un-populated. No data existed for much of the transportation and water coverages. The land use and houses/building coverages had databases only one-fourth to one-half complete. Populating these data sets from the air photography has proven to be time consuming.

Language: Language differences were minimized in that four members of the team at CIDDI spoke English, however a lack of Portuguese speakers on the team at WVU resulted in inevitable bottle-necks as translation from Portuguese to English had to go through the project administrator. Data sets were largely in Portuguese and translation into English was time consuming as each question and variables, in the socio-economic survey for instance, had to be translated. While issues of language will remain, methods should be investigated as to reducing language barriers.