Assessment of the Suitability of Rain Water Harvesting Areas Using Multi-Criteria Analysis and Fuzzy Logic

Abstract

Rain Water Harvesting (RWH) is any system that encompasses methods for collecting, concentrating and storing various forms of runoff for various purposes. Agriculture in semiarid tropics depends on the vagaries of weather, especially of the rain. Without doubt, the greatest climatic risk to sustained agricultural production in these areas, including Botswana, is rainfall variability. RWH has the potential to mitigate spatial and temporal variability of rainfall. Many methods of evaluating suitability for RWH, however, have limitations and/or drawbacks. This study presents an approach that will enable water managers to assess suitability of RWH for any given area by taking advantage of the capabilities of Earth Observation (EO) techniques and fuzzy multi-criteria analysis. Literature shows that the incorporation of fuzzy logic to multi-criteria analysis can improve the results in suitability analysis hence the study to explore these capabilities in RWH. South East District of Botswana was used as the study area to identify suitable areas for macro RWH techniques using Analytical Hierarchical Process (AHP) and Fuzzy AHP integrated in GIS and RS. The study area was suitable for over 80% of the area, with AHP approach showing 87.1% suitable while Fuzzy AHP showing 92.2%, distributed between highly suitable (S1), moderately suitable (S2) and marginally suitable (S3). Validation process shows existing water bodies occupying only highly suitable area (44%) and moderately suitable (56%) and this was a good indication that the model has a good level of accuracy. Field visit showed an accuracy of 57% comparing model results with actual situation on the ground. In conclusion, even though AHP is widely used in the decision analysis, it is not capable of modeling the uncertainties inherent in the criteria and the confidence of the decision maker. Fuzzy AHP is seen to perform better as it incorporates the techniques of AHP, fuzzy numbers, fuzzy extent analysis, alpha cut and Lambda functions which are able to model the uncertainties inherent in the criteria and confidence of the decision maker since the process of decision making involves a range of criteria and a good amount of expert knowledge and judgments which in turn affect the outcome greatly.