Neural Network Implementations on the coastal water quality of Manora channel for the years 1996 to 2014.

Abstract

Neural Networks is an Important Part of Computational Intelligence, Systems Theory and Signal Processing and finds numerous important applications in Science and Engineering. Sea water quality contaminates due to the severe untreated domestic, sewage and industrial pollutants. Presence of ammonia in seawater causes the deterioration of coastal water in terms of diminution of oxygen levels which suffocates the marine lives, fishes and mangroves. Industrial, sewage and domestic effluents carried by Lyari River contaminate the Manora channel, Karachi. The aim of study is to make the clear and transparent step-wise use of Artificial Neural Networks for the data driven water quality parameters models of Manora channel (Lyari river outfall zone N 24-51-26, E 66-58-01), Karachi (Pakistan) as well as to compare the pollutant contaminant ratio with the national environmental quality standard limits and other sampling sites of Manora channel and southern east Karachi coast. In this study, Manora channel Physico-chemical water quality parameters are assessed by using Artificial Neural Network taking Biochemical Oxygen Demand (BOD), chemical oxygen Demand (COD), Bicarbonates, potential Hydrogen(pH) , Chloride(Cl) as input and Ammonia(NH3)as output. Mean Square Error and R square are used for the model assessments statistical metrics. The computational work has been done by using R-studio. This is also found that Manora channel has the contaminated level of ammonia along the other sampling stations of both southern Karachi coast (N 24-47-03 E 67-08-39) as well as the other sampling site of Manora channel Karachi coast (N 24-50-15, E 66-58-01). In spite of all contamination Ammonia is found to be within National Environmental Quality Standards limits of Pakistan.