Scholarlink

<p>Managing salinity in arid rivers is impeded by sparse monitoring, relying on low-frequency grab samples that miss hydrological event dynamics. Here, interpretable machine learning is applied to a 50-year monthly archive (1968–2018; Discharge, major ions, pH) from three stations on Iran’s Karkheh River. Gradient Boosting Regression achieves high predictive skill for Total Dissolved Solids (TDS)/Electrical Conductivity (EC) (test-set R<sup>2</sup> = 0.94/0.97; RMSE = 55&#xa0;mg L<sup>−1</sup>/56 µS cm<sup>−1</sup>), validated via time-aware cross-validation. SHAP-based feature attribution reveals that Na<sup>+</sup> and SO<sub>4</sub><sup>–2</sup> are the strongest contributors to TDS, while Na<sup>+</sup> and Cl<sup>−</sup> dominate EC, consistent with conservative salinity sources under baseflow conditions. A reduced-input decision tree (four predictors) retains R<sup>2</sup> = 0.81–0.87, enabling minimal-sensor monitoring. Flow-regime partitioning and STL (Seasonal-Trend decomposition using Locally estimated scatterplot smoothing)-detrended event composites reveal low-flow salinization and ion-specific post-flood recovery (Cl<sup>−</sup>: 1–2&#xa0;months; Na<sup>+</sup>: 2–3&#xa0;months), guiding targeted sampling. This framework extracts predictive power, process associations, and operational guidance from legacy grab-sample archives, scalable to data-limited basins worldwide.</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

Interpretable machine learning for river salinity dynamics in arid basins

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月20日

查看全文
  • Hossein Amini,
  • Reza Shakeri,
  • Narjes Ghaderi,
  • Farshid Fakheri,
  • Khosro Morovati,
  • Banafsheh Zahraie,
  • Reza Ahmadian

摘要

Managing salinity in arid rivers is impeded by sparse monitoring, relying on low-frequency grab samples that miss hydrological event dynamics. Here, interpretable machine learning is applied to a 50-year monthly archive (1968–2018; Discharge, major ions, pH) from three stations on Iran’s Karkheh River. Gradient Boosting Regression achieves high predictive skill for Total Dissolved Solids (TDS)/Electrical Conductivity (EC) (test-set R2 = 0.94/0.97; RMSE = 55 mg L−1/56 µS cm−1), validated via time-aware cross-validation. SHAP-based feature attribution reveals that Na+ and SO4–2 are the strongest contributors to TDS, while Na+ and Cl− dominate EC, consistent with conservative salinity sources under baseflow conditions. A reduced-input decision tree (four predictors) retains R2 = 0.81–0.87, enabling minimal-sensor monitoring. Flow-regime partitioning and STL (Seasonal-Trend decomposition using Locally estimated scatterplot smoothing)-detrended event composites reveal low-flow salinization and ion-specific post-flood recovery (Cl−: 1–2 months; Na+: 2–3 months), guiding targeted sampling. This framework extracts predictive power, process associations, and operational guidance from legacy grab-sample archives, scalable to data-limited basins worldwide.