Analysis and ANN Modeling of Water Quality of Ramsar Site of Assam

This paper presents the water quality status of Deepor Beel which is a perennial freshwater lake and the Ramsar site in Assam. Now, its ecological health is being affected by developmental works; this study is aimed to find out the status of water quality so that remedial measures can be initiated to protect it. Various analyses including physical and chemical quantification of important parameters of the water samples were collected from different locations of Deepor Beel. Samples were found to have higher concentration of alkalinity, BOD, iron, TS, TDS, and TSS, in some locations, and DO was found less than the minimum requirement in few locations. Based on the correlation coefficient, alkalinity, acidity, total solids, total suspended solids, DO, and chloride were identified as sensitive parameters which were influencing the BOD level. Iron is the parameter which is least correlated with any of the parameters. An attempt was made to model the ecosystem study by using artificial neural network using the identified sensitive parameters as inputs and BOD as output. The number of hidden neurons was 10, and the transfer functions were tangent and logarithmic sigmoidal functions in the input layer and in the output layer, respectively, for the best configured neural network model.