Fuzzy modeling of farmers' knowledge for land suitability classification

In a case study, we demonstrate fuzzy modeling of farmers' knowledge (FK) for agricultural land suitability classification using GIS. Capture of FK was through rapid rural participatory approach. The farmer respondents consider, in order of decreasing importance, cropping season, soil color, soil texture, soil depth and slope as factors of suitability of their land for certain crops. Multi-class fuzzy sets using S-membership functions were generated for soil texture, soil depth and slope because of correlation or equivalence between farmers' definitions and scientific classifications of such land characteristics. In contrast, binary fuzzy relations, which are also fuzzy sets, were generated for cropping season and soil color because farmers' perceptions of such land characteristics are intrinsically binary. Despite variations in individual farmers' perceptions of land suitability, 12 unique FK rules for classifying land suitability were defined by hierarchical grouping of such different perceptions based on decreasing importance of factors. The FK rules form inference engines in combining fuzzy factor maps using appropriate fuzzy operators to create agricultural land suitability maps. Suitability maps resulting from application of Fuzzy AND and Fuzzy OR operators were found consistent with the FK rules. The FK-based suitability maps indicate either agreement or conflict with a land resource development plan (LRDP) for the case study area. Results of the study indicate usefulness of fuzzy modeling in FK-based classification of agricultural land suitability, which could provide useful information for optimum land-use planning.