Classification of groundwater by applying simplified fuzzy adaptive resonance theory

Abstract

Groundwater quality assessment is primarily intended to determine whether the water in a particular area can be used for aquatic purpose or not. The assessment comprises analysis of physical, chemical and microbiological characteristics of groundwater samples. The quality of groundwater can be evaluated through few standard conventional methods viz. Water Quality Index, Canadian Council of Ministries Environmental Water Quality Index and Weighted Arithmetic Water Quality Index etc. In addition to the conventional methods, multivariate statistical methods like Principal Component Analysis, Factor Analysis and Cluster Analysis can also be used to assess the groundwater quality. As these methods are descriptive models, they are inadequate to predict the quality of unknown groundwater sample. Hence, an efficient predictive model is desirable to analyze the characteristic parameters of groundwater samples and predict the quality of an unknown sample. The sample may have both crisp and fuzzy values. Conventional supervised learning methods may not be suitable for constructing the required prediction model as they are not suitable for handling fuzzy input data. Therefore, simplified fuzzy adaptive resonance theory model is an appropriate choice for accomplishing the task of building the prediction model. The present work proposes to assess the quality of groundwater by applying the Weighted Arithmetic Water Quality Index method and Simplified Fuzzy Adaptive Resonance Theory model by considering 7 groundwater quality parameters. The accuracy of afore mentioned approach seems to be pleasing when compared to counter parts like Back propagation and Random Forests classifiers.