Scholarlink

This study uses machine learning models, namely, support vector machine (SVM) and multiple linear regression (MLR), to predict cumulative runoff load (CRL) and cumulative sediment load (CSL) from laboratory-based rainfall-runoff-sediment yield data generated on a model catchment while taking into account the effects of varying rainfall intensity, vegetal cover, gradient, antecedent moisture content (AMC) of soil, and soil types. The results demonstrated that the quantum of surface runoff and sediment generation varied substantially depending on rainfall intensity and catchment characteristics. When compared to the MLR approach, which yielded R2 0.81 and 0.87 for calibration data and R2 0.71 and 0.77 for validation data, SVM model based on Radial Basis Kernel revealed high CRL and CSL retrieval accuracies, yielding R2 0.87 and 0.86 for calibration data and R2 0.85 and 0.86 for validation data, respectively.

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

Runoff and Sediment Yield Prediction Using Machine Learning Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Mohammad Khalid Nasiry,
  • Saif Said

摘要

This study uses machine learning models, namely, support vector machine (SVM) and multiple linear regression (MLR), to predict cumulative runoff load (CRL) and cumulative sediment load (CSL) from laboratory-based rainfall-runoff-sediment yield data generated on a model catchment while taking into account the effects of varying rainfall intensity, vegetal cover, gradient, antecedent moisture content (AMC) of soil, and soil types. The results demonstrated that the quantum of surface runoff and sediment generation varied substantially depending on rainfall intensity and catchment characteristics. When compared to the MLR approach, which yielded R2 0.81 and 0.87 for calibration data and R2 0.71 and 0.77 for validation data, SVM model based on Radial Basis Kernel revealed high CRL and CSL retrieval accuracies, yielding R2 0.87 and 0.86 for calibration data and R2 0.85 and 0.86 for validation data, respectively.