Assessment and Modeling of Water Quality in the Niger River: A Comprehensive Study Using Water Quality Indices and Intelligent Techniques

Abstract

The Niger River is essential to the livelihoods of those in Bamako and its neighboring areas, providing vital resources for drinking water, agriculture, livestock, industry, and fishing. Given its significance across these sectors, there is a pressing need to establish an effective water resource management strategy that incorporates a thorough qualitative assessment of the river's water quality. This research seeks to characterize the water quality of the Niger River by employing Water Quality Indices (WQI) and intelligent modeling techniques. To fulfil this objective, various physicochemical parameters, including pH, electrical conductivity (EC), nitrate (NO3-), nitrite (NO2-), and iron (Fe), were collected from 40 sampling points along the river during three distinct periods: December 2017, March 2018, and July 2018. The study utilized a weighted arithmetic approach to compute the WQIs, while the predictive models were developed using two of the most famous and effective modeling techniques namely Multiple Linear Regression (MLR) and Artificial Neural Networks (ANN). In order to evaluate the predictive performance of the models, the dataset was partitioned into three distinct segments, allocating 60% for training purposes, 20% for validation, and the remaining 20% for testing, with the segments organized in a sequence from upstream to downstream. The performance of both models was evaluated using metrics such as the correlation coefficient (r²), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). The computed Water Quality Indices (WQIs) vary from 0.44 to 1887.40, indicating a diverse range of water quality across the samples analyzed. The classification of these samples reveals that 62.5% are considered excellent, while 15% are categorized as good, another 15% as poor, 2.5% as very poor, and 5% as unsuitable for consumption. Furthermore, the results derived from ANN with five inputs, one hidden layer (13 neurons) and one output (WQI) demonstrates superior efficiency in assessing water quality.