Scholarlink

<p>Estimating the suspended sediment load (SSL) in watersheds is a fundamental challenge in surface water resource management and hydraulic structure design. This study utilized data from the Haraz, Babol Rud, Talar, and Neka Rud watersheds in Mazandaran province, Iran, encompassing discharge, daily sediment load, and key physical characteristics. These data were organized in three different combinations as inputs to several machine learning models. Among these, the input scenario M<sub>3</sub>, which combines streamflow data with physical watershed characteristics, demonstrated the best performance across all models. The machine learning models employed were Support Vector Regression (SVR), Long-Short Term Memory (LSTM), Random Forest (RF) and Extreme Gradient Boosting (XGBoost), the performance of which was enhanced using two optimization algorithms: Particle Swarm Optimization (PSO), and the Flow Direction Algorithm (FDA). Results indicated that FDA-optimized models consistently outperformed their PSO counterparts. Specifically, the XGBoost-FDA hybrid model exhibited superior performance, achieving high accuracy in both training (RMSE = 2.86 ton/day, MAE = 2.20 ton/day, Pearson <i>R</i> = 0.93, KGE = 0.88, NSE = 0.86) and testing phases (RMSE = 3.05 ton/day, MAE = 2.31 ton/day, <i>R</i> = 0.93, KGE = 0.89, NSE = 0.86). This performance represents a substantial improvement over a baseline simple linear regression model (<i>R</i> = 0.74). The findings of this study have significant implications for engineers and policymakers in the design of hydraulic structures and water resources management.</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 suspended sediment load forecasting using a physics-based optimized hybrid model: a regional study in the Mazandaran basins

  • 期刊论文

  • 发表日期：

    2026年5月4日

查看全文
  • Ali Akbar Eatesam,
  • Khosrow Hoseini,
  • Hojat Karami

摘要

Estimating the suspended sediment load (SSL) in watersheds is a fundamental challenge in surface water resource management and hydraulic structure design. This study utilized data from the Haraz, Babol Rud, Talar, and Neka Rud watersheds in Mazandaran province, Iran, encompassing discharge, daily sediment load, and key physical characteristics. These data were organized in three different combinations as inputs to several machine learning models. Among these, the input scenario M3, which combines streamflow data with physical watershed characteristics, demonstrated the best performance across all models. The machine learning models employed were Support Vector Regression (SVR), Long-Short Term Memory (LSTM), Random Forest (RF) and Extreme Gradient Boosting (XGBoost), the performance of which was enhanced using two optimization algorithms: Particle Swarm Optimization (PSO), and the Flow Direction Algorithm (FDA). Results indicated that FDA-optimized models consistently outperformed their PSO counterparts. Specifically, the XGBoost-FDA hybrid model exhibited superior performance, achieving high accuracy in both training (RMSE = 2.86 ton/day, MAE = 2.20 ton/day, Pearson R = 0.93, KGE = 0.88, NSE = 0.86) and testing phases (RMSE = 3.05 ton/day, MAE = 2.31 ton/day, R = 0.93, KGE = 0.89, NSE = 0.86). This performance represents a substantial improvement over a baseline simple linear regression model (R = 0.74). The findings of this study have significant implications for engineers and policymakers in the design of hydraulic structures and water resources management.