Application of machine learning in predicting sources of water pollution in the Euphrates and Tigris rivers in Iraq

Abstract

New evaluation and control methods are required to address the ecological, economic, and public health concerns raised by the contamination of the rivers Tigris and Euphrates. To minimize negative effects on ecosystems, our research built and implemented a machine learning framework to track down and foresee potential water contamination hotspots. To examine the causes of pollution and its consequences on aquatic ecosystems, researchers combined data from multiple sources, such as aerial photographs, field surveys, and official government documents. Predictive models encompass significant attributes such as pesticides, mineral composition, suspended particulates, diversity of macroinvertebrates, and habitat quality. Feature selection techniques, including LASSO regression and recursive feature elimination, ensured dependable model construction. Four machine learning algorithms of MCP, K-nearest neighbors, decision tree, and multi-layer perceptron were employed for pollution source recognition and impact prediction. The models correctly identified significant pollution sources, including untreated sewage, agricultural runoff, and industrial discharges. The concentration and distribution patterns of pollutants were elucidated by clustering and regression techniques. The results indicated reduced biodiversity, habitat degradation, and toxic algal blooms, as well as identified significant pollution areas. This research shows that machine learning can transform environmental monitoring and water resource management. The study's practical findings, which integrate ecological and computational methodologies, can assist policymakers and water resource managers.