Scholarlink

<p>Predicting surface water quality is essential for environmental conservation, public health protection, and sustainable resource management. It empowers stakeholders to make well-informed choices concerning water systems, especially in the context of pollution and climate variability. The synergy between time series analysis and machine learning techniques marks a major advancement in the predictive modeling of water quality. This methodological integration not only boosts precision but also strengthens proactive measures for addressing water-related challenges. Decomposing a time series into basic and fluctuating parts is fundamental because it reveals the structure of the data, enables tailored-modeling of each component, improves forecast accuracy, and provides actionable insights for better decision-making in various applications. In this study, a combination of time series analysis and machine learning was used to predict the monthly surface water quality. The models used in this research combine the sensitive nonlinear iterative peak (SNIP) method and two models, a convolutional neural network (CNN) and an artificial neural network (ANN). After modeling and predicting the quality parameters of the surface water, the obtained results were compared with three evaluation criteria: coefficient of determination (<i>R</i><sup>2</sup>), root-mean-square error (RMSE), and Nash–Sutcliffe coefficient (NSE). Evaluations and graphical charts showed that the SNIP-CNN combined model has high performance on average for all three stations with <i>R</i><sup>2</sup> (0.883), RMSE (3.53), and NSE (0.842). By integrating the SNIP method with the CNN model, the error decreased by 35.52%, while the ANN model saw a 47.7% reduction. These results highlight the significant role of time series analysis using the SNIP method in enhancing prediction accuracy. According to the results obtained from the integration of the two methods, it is possible to make assessments and predictions with high reliability for the surface water quality of other basins.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Predictions of Surface Water Quality Through the Integration of Deep Learning and Time Series Analysis Techniques

  • 期刊论文

  • 发表日期：

    2025年7月14日

查看全文
  • Anis Hasani,
  • Mohammad Ali Ghorbani,
  • Sabereh Darbandi

摘要

Predicting surface water quality is essential for environmental conservation, public health protection, and sustainable resource management. It empowers stakeholders to make well-informed choices concerning water systems, especially in the context of pollution and climate variability. The synergy between time series analysis and machine learning techniques marks a major advancement in the predictive modeling of water quality. This methodological integration not only boosts precision but also strengthens proactive measures for addressing water-related challenges. Decomposing a time series into basic and fluctuating parts is fundamental because it reveals the structure of the data, enables tailored-modeling of each component, improves forecast accuracy, and provides actionable insights for better decision-making in various applications. In this study, a combination of time series analysis and machine learning was used to predict the monthly surface water quality. The models used in this research combine the sensitive nonlinear iterative peak (SNIP) method and two models, a convolutional neural network (CNN) and an artificial neural network (ANN). After modeling and predicting the quality parameters of the surface water, the obtained results were compared with three evaluation criteria: coefficient of determination (R2), root-mean-square error (RMSE), and Nash–Sutcliffe coefficient (NSE). Evaluations and graphical charts showed that the SNIP-CNN combined model has high performance on average for all three stations with R2 (0.883), RMSE (3.53), and NSE (0.842). By integrating the SNIP method with the CNN model, the error decreased by 35.52%, while the ANN model saw a 47.7% reduction. These results highlight the significant role of time series analysis using the SNIP method in enhancing prediction accuracy. According to the results obtained from the integration of the two methods, it is possible to make assessments and predictions with high reliability for the surface water quality of other basins.