Article ID: | iaor20032827 |
Country: | United States |
Volume: | 73 |
Issue: | 3 |
Start Page Number: | 233 |
End Page Number: | 260 |
Publication Date: | Sep 2002 |
Journal: | Agricultural Systems |
Authors: | Dounias I., Aubry C., Capillon A. |
Keywords: | developing countries, artificial intelligence: decision support |
Recent work on decision processes on French farms and irrigated systems in Africa has shown that farmers plan their cyclical (recurrent) technical operations, and that one can model this planning process. Taking the case of cotton crop management in North Cameroon, we show that with certain adjustments, modelling of this kind can also be done for rainfed crop farming in Africa. The adjustments are needed to take account of the differences in social status between different fields on one farm and the implications of the fact that farm work is primarily manual. This produces decision models with a similar structure to that described for technical management of an annual crop break in a temperate climate using mechanised implements. Not only do these models give us a detailed understanding of the variability of farming practices, we can also classify them into categories according to weather scenarios yield level as a function of weather scenario. We show that one can attribute farms to these types of model using simple indicators concerning work organization. By analyzing North Cameroon farmers' decision processes for managing cotton crops we can thus produce an effective tool for organizing technical supervision of farmers at the regional level: advisers can work with these decision model types by measuring some simple indicators at farm level to predict which types of model are applicable, without the onerous work of constructing individual decision models.