Al-Khuzaie Marwah M, Abdul Maulud Khairul Nizam, Wan Mohtar Wan Hanna Melini, Yaseen Zaher Mundher
Department of Civil Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, 43600 UKM, Bangi, Selangor, Malaysia.
Civil Engineering Department, College of Engineering, University of Al-Qadisiyah, Al-Qadisiyah, Iraq.
Sci Rep. 2025 Jan 2;15(1):51. doi: 10.1038/s41598-024-84072-1.
Pollution monitoring in surface water using field observational procedure is a challenging matter as it is time consuming, and needs a lot of efforts. This study addresses the challenge of efficiently monitoring and predicting water pollution using a GIS-based artificial neural network (ANN) to detect heavy metal (HM) pollution in surface water and effect of wastewater required discharge on the Euphrates River in Al-Diwaniyah City, Iraq. The study established using 40 water sampling stations and incorporates Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-OES) to assess HM levels. An ANN model suggested to estimate Heavy Metal Pollution Index (HPI) considering physiological and chemical factors. It formulates six scenarios to enhance HPI prediction accuracy, utilizing ANN in MATLAB for modeling and GIS statistical tools with inverse distance weighted (IDW) methods for a comprehensive assessment. The developed approach predicted HP concentration in the Euphrates River basin in an actual case study. The validation of the predictive maps between the theoretical and practical part is performed by monitoring 16 stations and conducting laboratory analyses, resulting in acceptable coefficients of determination (R), observations standard deviation ratio (RSR), and Nash-Sutcliffe efficiency coefficients of 0.999, 1, and 0.99, respectively indicates that reliable forecast results closely match observed data from monitoring stations. The study identifies that nickel, iron, and cadmium concentrations exceeded Iraqi and World Health Organization (WHO) standards, leading to a heavy pollution index peak of 150.38 in the Euphrates River branch. In this study, the HPI is used to identify areas with high pollution levels, validate the accuracy of the ANN model for prediction, and generate a pollution map to visualize pollution levels.
采用实地观测程序对地表水进行污染监测是一项具有挑战性的工作，因为它既耗时又需要付出大量努力。本研究应对了利用基于地理信息系统（GIS）的人工神经网络（ANN）有效监测和预测水污染这一挑战，以检测伊拉克迪瓦尼亚市幼发拉底河地表水的重金属（HM）污染以及废水排放要求对其的影响。该研究利用40个水采样站开展，并采用电感耦合等离子体原子发射光谱法（ICP - OES）来评估重金属水平。提出了一个考虑生理和化学因素来估算重金属污染指数（HPI）的人工神经网络模型。它制定了六种情景以提高HPI预测准确性，在MATLAB中利用人工神经网络进行建模，并使用反距离加权（IDW）方法的GIS统计工具进行全面评估。在一个实际案例研究中，所开发的方法预测了幼发拉底河流域的HP浓度。通过监测16个站点并进行实验室分析，对理论部分和实际部分的预测图进行了验证，得到的决定系数（R）、观测标准差比（RSR）和纳什 - 萨特克利夫效率系数分别为0.999、1和0.99，这表明可靠的预测结果与监测站的观测数据紧密匹配。该研究确定镍、铁和镉的浓度超过了伊拉克和世界卫生组织（WHO）的标准，导致幼发拉底河支流的重度污染指数峰值达到150.38。在本研究中，HPI用于识别高污染水平区域、验证人工神经网络模型预测的准确性，并生成污染图以直观显示污染水平。
