Using indigenous knowledge, remote sensing and gis for sustainable development

Joseph A. Tabor
Charles F. Hutchinson

Information describing the natural resources of any region (e.g. soil, water and vegetation) forms the base upon which sustainable development must be built. The twin challenges of providing information that can lead to sustainable development are to keep acquisition costs low and utility high. In this article an alternate approach based on several techniques of data acquisition, interpretation and management is presented. It is intended to enhance the utility and lower the cost of surveys by complementing- -not replacing--conventional survey techniques.

Survey costs and utility
Especially in developing countries, personnel, logistics and classification systems are three reasons for the high costs of resource surveys:

• Personnel: Surveys typically are conducted by highly paid expatriate experts who are not always familiar with the area, its resources, the local culture and resource management practices. Thus, part of their expensive time is devoted to learning.
• Logistics: Surveys are built around ground sampling. The movement of personnel and equipment is expensive.
• Classification systems: Many surveys conducted in developing countries are based on classification systems developed elsewhere. Thus, surveyors devote considerable time to making resources fit into classification systems that may not reflect local conditions. Survey progress may be further impeded because most 'objective' or scientific classification systems (e.g. soils) depend on measurable features (e.g. soil chemical and mechanical composition). They may require laboratory analyses that cannot be conducted in the field (e.g. USDA's soil classification system's dependence on laboratory analysis as compared to that of the French system).

• Accessibility: Objective classification systems are based on criteria that are unfamiliar to many potential users (e.g. planners, farmers, foresters, land managers). Thus, they cannot interpret the resource survey without expert assistance.
• Credibility: Most classification systems do not consider local resource values and management practices. Thus, surveyors may include units that are unimportant or, worse, ignore those that are extremely valuable to local resource managers (e.g. farmers, herders, foresters).

An alternate approach
The approach outlined is based on several techniques of data acquisition, interpretation, and management. These techniques are indigenous knowledge and classification systems, remote sensing and satellite navigation, and geographic information systems (GIS). The approach is intended to enhance the utility and lower the cost of surveys by complementing--not replacing--conventional survey techniques.

Local resource managers possess a detailed understanding of resource values and management practices. A large part of this understanding is contained in the classification systems they use to describe resources (Sturtevant, 1964; Conklin, 1969; Hunn, 1982). For example, in the drought-prone Senegal River Valley, three riparian landscapes are recognized by local farmers. One is flooded by the Senegal during the fall and the other two by local rains in the summer. Although each landscape is composed of the same soils, they are distinguished by the frequency, period and time of year they are flooded. As a result, each soil is managed differently. These soils cannot be recognized in a conventional soil survey because distinctions are based on value and management criteria rather than physical properties.

Unlike objective classification systems, indigenous systems embody local values and management systems as well as the relationships among other resources (e.g. soil/vegetation associations). Moreover, because they are derived from local terminology, they provide a common language for exchanging information.

There may be disparities between indigenous and objective classification systems. Yet, because physical properties of the resource ultimately determine its potential uses, these divergences tend to be superficial from a classification perspective (Behrens, 1989). Because of their inherent similarity, it is possible to move among systems--indigenous and objective--drawing on the different knowledge contained in each. In addition, the differences in classification criteria provide insight into local values and management practices. Finally, development policy that unknowingly undermines or contradicts viable indigenous resource management strategies is unlikely to be sustainable.

Aerial sampling, aerial photography, aerial video and satellite images are tools for producing conventional objective resource survey maps. Because indigenous systems of resource classification are based on observable characteristics, remote sensing can play a useful role. In the approach suggested here, the designation of the units to be mapped comes from the resource manager rather than from the surveyor or remote sensing specialist.

Natural resources are spatially variable and can be difficult to describe without a dense net of sample points. Because of this variability, boundaries between resource units are more easily recognized on images than from ground samples alone. Depending on the scale of the survey, remote sensing data can be used to develop reasonably homogeneous sampling 'strata' to which ground samples are allocated. Remote sensing data from different dates, platforms (e.g. satellites and aircraft) and different scales can be used to further supplement ground samples (Hutchinson, 1991). Images acquired for the resource survey are also historical documents that describe conditions at the beginning of the project. Subsequent changes resulting from the project can be measured and documented.

Every resource survey involves ground sampling, but this can be speeded up by new satellite technologies that can provide ground navigation. Particularly in remote areas without good base maps, roads or reference points, determining simple geographic location is a major task. Satellite-based Global Positioning System (GPS) can provide highly accurate positions for ground sample points quickly and inexpensively.

A geographic information system (GIS) permits the capture, storage, manipulation, analysis, and display of spatial data. These may be in the form of maps (e.g. land use), images (e.g. satellite), point data (e.g. rainfall) or tabular data associated with geographic areas (e.g. census). A GIS is capable of storing, processing and displaying all these types of data.

A GIS must be able to manage and analyze map features in two dimensions (e.g. soil units), but also their associated attributes (e.g. soil type, texture and depth). Thus, a database management system is also required for entering and managing tabular data as well as performing certain statistical and logical operations.

Resource information commonly is presented as maps, but additional interpretive information is carried in tables that are referenced explicitly to map units. In the GIS, interpretive tabular data (e.g. infiltration rates by soil type) can be transformed into maps that are more easily displayed and understood.

In the GIS, various features can be combined and compared to select regions of interest (e.g. habitat dictated by vegetation, soil type, and proximity to surface water). In addition, with the approach described here, it is possible to compare different classification systems--both logically and spatially (e.g. comparisons of indigenous vs FAO vs French vs U.S. soil classifications). These comparisons provide insight into alternative development initiatives under different management scenarios. For example, in the Senegal River Valley, karan karan soils are eroded, crusted, and considered useless. However, in an objective classification, these soils have good agricultural properties if water conservation or irrigation is applied.

Application
The general approach to resource survey outlined above can be used in at least three different ways to enhance the utility and reduce costs of conventional information gathering and management. In each, scale would be determined by the size of the management unit.

Reconnaissance surveys are performed in information-poor regions and provide an overview of conditions for relatively large areas. Principal users include planners and government managers. Using the techniques described above, reconnaissance surveys would enable these people to gain a general understanding of the resource base and management issues, capture the local resource taxonomy, describe local resource management practices and values, and build a framework for future work and for the identification of critical information gaps.

Management surveys are performed to identify critical issues that relate to the management of specific tracts of land. The audience for the information might include extension workers or project planners. The techniques employed would be the same as those outlined above, the products would be similar but site-specific. Thus, in addition to greater detail on the physical nature of resources, the management survey would contain comparable detail on other issues (e.g. tenure, markets and finance).

Reconciliation must be performed for information- rich regions; its purpose is to make information contained in multiple surveys accessible for reference. The audience would be planners and government managers. Techniques employed would be the same as those used for reconnaissance, and the products would be the same. The difference is that reconciliation would produce a common lexicon so that information could be shared. In addition, divergences in taxonomies can be extracted in order to highlight deficiencies in objective classifications.

Conclusions
The methods described here can contribute to sustainable development by enhancing the efficiency of conventional information-gathering efforts. They can also help to indicate the types of development that might actually be sustained by local populations. Using indigenous knowledge ensures that the most important factors determining resource value and management practices will be captured during the survey. In addition, use of local terminology ensures that the information generated can be easily disseminated.

The time required for a surveyor to learn about local resources is minimized if he/she starts with the indigenousknowledge base. The links between resource's properties and its social significance (e.g. resource tenure) are difficult to ascertain through individual experience alone. However, these relationships are inherent in indigenous classification systems. Also, basing surveys on indigenous knowledge ensures that effort is concentrated on the most valuable resource types.

The use of remote sensing, combined with a GPS-navigated ground sampling strategy, yields a more accurate and credible product (either estimates or maps) than can be produced easily through ground sampling alone. It is impossible to generate resource estimates (e.g. range production and animal census) and produce maps quickly, inexpensively and accurately using ground samples alone. Remote sensing and statistical methods are critical for balancing the improvement of information accuracy (i.e. more sampling) against the added cost of that improvement.

The GIS allows the spatially referenced data that managers might desire (e.g. crop production by district) to be compared with map- based resource information (e.g. land value). The GIS also allows products (e.g. maps and tables) to be prepared in the classification system of choice, permitting communication between surveyors and end users. In addition, some of the products suggested above (e.g. habitat based on vegetation, soil and water) cannot be easily generated using conventional (i.e. manual) techniques.

Maps in digital form are easily and inexpensively archived, updated and disseminated. Moreover, many useful products (e.g. overlays) are almost impossible to prepare manually. Despite differences in classification systems, site descriptions (point samples) from previous surveys can be located and referenced, thereby reducing the total number of sites that must be visited if the region has already been sampled.

These techniques for information gathering, management and interpretation can be combined to strengthen conventional surveys conducted for reconnaissance or management. They also can be used to reconcile and thus make good use of different surveys that might exist for the same area.

Joseph A. Tabor
Charles F. Hutchinson
Arizona Remote Sensing Center
College of Agriculture
The University of Arizona
Tucson, AZ 85721
USA

References
Behrens, C.A. (1989) 'The scientific basis for Shipibo soil classification and land use: changes in soil-plant associations with cash cropping', American Anthropologist 91:83-100.

Conklin, H.C. (1969) 'An ethnoecological approach to shifting agriculture', pp. 221-233 in A.P. Vayda (ed) Environment and Cultural Behavior. New York: The Natural History Press.

Hunn, E. (1982) 'The utilitarian factor in folk biological classification', American Anthropologist 84:830-847.

Hutchinson, C.F. (1991) 'Integration of remote sensing and farming systems research in West Africa', pp. 59-68 in A. Bishay and H.E. Dregne (eds) Advances in desert and arid land technology and development, Volume 5, Part 1. New York: Harwood Academic Publishers.

SEDAGRI (1973) 'Etude pedologique (texte)' and 'Etude pedologique et geomorphologique de la vallee et du delta du Senegal au 1/50.000', pp. 252 in Organisation pour la Mise en Valeur du fleuve Sénégal/FAO Etude hydro-agricole du bassin du fleuve Senegal. Paris: SEDAGRI.

Sturtevant, W.C. (1964) 'Studies in ethnoscience', American Anthropologist 66:99-131.

Tabor, J.A. 1993. 'Soils of the lower middle and upper Senegal River Valley', pp 33-50 in T.K. Park (ed) Risk and tenure in arid lands: the political ecology of development in the Senegal River Basin. Tucson: University of Arizona Press.

Tabor, J.A., A. Watson, A. Diagne, M. Daffe, M. Dieme and M. Khouma (1993) 'Annex du sols', pp. 91 data base on diskettes, 1:50.000 and 1:20.000 maps in Organisation pour la Mise en Valeur du fleuve Sénégal Plan directeur de développement de la haute vallée du fleuve Sénégal. Paris: Dakar/USAID/Dames & Moore.