A GIS-integrated fuzzy rule-based inference system for land suitability evaluation in agricultural watersheds

Abstract

Land suitability evaluation in water scarce agricultural watersheds consists of assessment of land potential for various crops as well as surface water potential to identify the scope for supplementary irrigation. A large amount of information related to the crop land suitability can be conveyed through linguistic terms. Capability of fuzzy sets in modeling involving uncertainty and vagueness is made use of in fuzzy rule-based systems, where various decision making criteria in linguistic terms are expressed as fuzzy rules. In the present study, a fuzzy rule-based inference system is developed in Geographic Information System (GIS) environment to assess the land suitability pertaining to the specified crop, considering both land potential and surface water potential. When large numbers of attributes are involved in decision making, representation of the attributes in a common scale, aggregation of the attributes and design of the rule-base becomes difficult tasks. In order to model the heterogeneous land suitability criteria involving large number of attributes, a new approach is proposed in this study in which the attributes are systematically classified into different groups to estimate the intermediate suitability indices. Weighted linear aggregation method and Yager’s aggregation method are used for estimating the aggregated effect of the attributes in each group and the results are compared. Further, the rule-base is developed by using the intermediate land suitability indices. The model has been applied to a subwatershed of Gandheshwari area in West Bengal (India). The input attributes are prepared in raster map format in the GIS environment by using ERDAS IP ver. 9.1 and the output is generated in the form of thematic map showing the suitability of each cell (20 m × 20 m) for the selected crop. For the land suitability evaluation problem in the case study area, Yager’s aggregation method has been found more appropriate than the commonly used weighted linear aggregation method. From the analysis, 23% of the existing paddy fields have been found less suitable/ not suitable for paddy due to the poor surface water potential or unsuitable terrain conditions of the area. The method, integrated with GIS, is found efficient in handling large amount of attribute information, and is useful in the land suitability assessment in agricultural watersheds.