Scholarlink

<p>Rivers are essential natural resources, but managing their sedimentation processes poses significant challenges. Suspended Sediment Concentration (SSC) is crucial in understanding water quality and ecosystem health. However, direct measurement of SSC is often impractical and costly for many river systems. This study explores machine learning approaches for predicting daily SSC in the Ganga River in Banaras. We compare the performance of a Feed-Forward Backpropagation Neural Network (FFNN) and a Support Vector Machine (SVM) using one year of daily data. Evaluation metrics such as Root Mean Square Error (RSME), coefficient of determination (R), and Nash-Sutcliffe Efficiency (NSE) were employed to assess model accuracy. Our results indicate that the FFNN model outperforms the SVM model, achieving superior accuracy with better RSME, R, and NSE values compared to SVM's RSME, R, and NSE values. These findings highlight the potential of machine learning techniques, particularly FFNN, in accurately predicting sediment levels. Such predictive capabilities are essential for informed decision-making in river management and water resource planning.</p> Graphical abstract <p></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

Comparative analysis of machine learning models for daily suspended sediment concentration prediction in environmental monitoring

  • 期刊论文

  • 发表日期：

    2025年3月24日

查看全文
  • Goldi Jarbais,
  • Pon Harshavardhanan

摘要

Rivers are essential natural resources, but managing their sedimentation processes poses significant challenges. Suspended Sediment Concentration (SSC) is crucial in understanding water quality and ecosystem health. However, direct measurement of SSC is often impractical and costly for many river systems. This study explores machine learning approaches for predicting daily SSC in the Ganga River in Banaras. We compare the performance of a Feed-Forward Backpropagation Neural Network (FFNN) and a Support Vector Machine (SVM) using one year of daily data. Evaluation metrics such as Root Mean Square Error (RSME), coefficient of determination (R), and Nash-Sutcliffe Efficiency (NSE) were employed to assess model accuracy. Our results indicate that the FFNN model outperforms the SVM model, achieving superior accuracy with better RSME, R, and NSE values compared to SVM's RSME, R, and NSE values. These findings highlight the potential of machine learning techniques, particularly FFNN, in accurately predicting sediment levels. Such predictive capabilities are essential for informed decision-making in river management and water resource planning.

Graphical abstract