Advancing sedimentation modeling in large reservoir systems: Insights from multi-scale process coupling and machine learning

Advancing sedimentation modeling in large reservoir systems: Insights from multi-scale process coupling and machine learning

Study region: The lower Jinsha River, China, one of the largest cascade reservoir systems in the world. Study focus: Sedimentation directly affects reservoir operation and management. The study area is a typical mountainous basin with complex sediment transport dynamics, which challenge both the acc...

Full description

Saved in:
Bibliographic Details
Main Authors: Yuning Tan, Huaixiang Liu, Yongjun Lu, Zhili Wang, Wenjie Li
Format: Article
Language:English
Published: Elsevier 2025-08-01
Series:Journal of Hydrology: Regional Studies
Subjects:
Sediment transport
Cascade reservoir
Factors
Mountainous basin
Jinsha River
Online Access:http://www.sciencedirect.com/science/article/pii/S2214581825003982
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Study region: The lower Jinsha River, China, one of the largest cascade reservoir systems in the world. Study focus: Sedimentation directly affects reservoir operation and management. The study area is a typical mountainous basin with complex sediment transport dynamics, which challenge both the accuracy and efficiency of modeling efforts. To address this issue, this study proposes a novel modeling framework that integrates multi-scale physical processes (sediment sources, inflow, and deposition) with machine learning (ML) techniques. New hydrological insights for the region: The proposed framework firstly reduced the total sedimentation error from 53.42 % to 3.44 % and the maximum group-wise error from 90.88 % to 13.46 %, highlighting the dominant influence of tributary sediment inputs and flocculation factor on reservoir sedimentation. The subsequently developed ML models effectively captured short-term sedimentation rate variations (R2 > 0.83, KGE > 0.85 in testing). Random Forest outperformed XGBoost, Support Vector Regression, and Artificial Neural Network, demonstrating the most robust performance and the clearest feature attribution. Our climate scenario simulations revealed uneven sedimentation patterns and projected declining sedimentation rates over the next 30 years. This study can offer valuable insights for modeling sediment dynamics in mountainous basins with large reservoir systems.
ISSN:2214-5818

Similar Items