Predicting Groundwater Potentials Using Case Based Reasoning

Abstract

Water is one of the basic necessities of life but the available surface water is inadequate to meet up with human’s demand for water. Hence, groundwater which is the subsurface water beneath the earth’s surface is exploited to complement surface water based on different parameters. These parameters include its potentiality, amount of water stored/ found in a particular media at a time and the ability to recharge. The potential to recharge is the major indicator in groundwater search; this is done by carrying out extensive geological and geophysical survey of the area. This paper proposes the prediction of groundwater potentials using Case Based Reasoning (CBR). Geological and geophysical survey data including electrical resistivity survey data also known as Vertical Electricity Survey (VES) was acquired from three local government areas in the southern part of Oyo State, Nigeria. The problem attributes of each case include the coefficients of resistivity, VES data, geo-electric section data while the solution attribute is the groundwater potential with values poor, moderate or high. Given a new set of problem attributes, our CBR technique retrieves the most similar case from our casebase as a potential solution. The retrieved solution is then adapted as solution to the current problem is the similarity meets a particular threshold. An accuracy of 80% was obtained from our experiments where moderate and high groundwater potential were assumed to be identical. We intend to experiment with data from other parts of the country to further validate the CBR technique and investigate the effects of a three-way classification, instead of the two solution classes used in this study.