Scholarlink

<p>Accurate streamflow simulation is crucial for effective water resource management and flood control, particularly in vulnerable regions like the Northeastern Himalayas region of India, which are susceptible to severe hydrological events. While traditional hydrological models and machine learning techniques are useful, they face limitations in terms of data requirements and predictive accuracy. To address these challenges, this study proposes a novel hybrid model that integrates the physics-based HEC-HMS hydrological model with an LSTM network, aiming to enhance streamflow prediction accuracy in the Tlawng River basin. The goal is to develop a more robust predictive tool that leverages the strengths of both the physics-based HEC-HMS model and the LSTM network. The hybrid model was trained using daily streamflow and meteorological data from June 2017 to December 2020, and its performance was validated on an unseen dataset from January 2021 to April 2022. The integration of HEC-HMS simulated discharge and additional meteorological inputs into the LSTM network was evaluated using five statistical metrics. The validation results on the new unseen dataset showed the following performance: HEC-HMS achieved an RSR of 0.62, PBIAS of 8.65%, RMSLE of 0.64, and MAE of 17.49, while the LSTM model demonstrated improved performance with an RSR of 0.56, PBIAS of 8.54%, RMSLE of 0.51, and MAE of 15.80. In contrast, the hybrid HEC-HMS-LSTM model outperformed the others with an RSR of 0.50, PBIAS of 7.34%, RMSLE of 0.49, and MAE of 14.64, effectively capturing peak discharges and temporal trends. The findings highlight the hybrid model’s potential as a practical and efficient tool for improving flood prediction and water resource management in the Tlawng River basin. The novelty of this research lies in the advanced hyper-parameter tuning applied in a mountainous region, which enhances the model’s performance and adaptability to the complex hydrological dynamics of the area.</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

Streamflow Simulation Using a Hybrid Approach Combining HEC-HMS and LSTM Model in the Tlawng River Basin of Mizoram, India

  • 期刊论文

  • 发表日期：

    2025年1月6日

查看全文
  • Sagar Debbarma,
  • Arnab Bandyopadhyay,
  • Aditi Bhadra

摘要

Accurate streamflow simulation is crucial for effective water resource management and flood control, particularly in vulnerable regions like the Northeastern Himalayas region of India, which are susceptible to severe hydrological events. While traditional hydrological models and machine learning techniques are useful, they face limitations in terms of data requirements and predictive accuracy. To address these challenges, this study proposes a novel hybrid model that integrates the physics-based HEC-HMS hydrological model with an LSTM network, aiming to enhance streamflow prediction accuracy in the Tlawng River basin. The goal is to develop a more robust predictive tool that leverages the strengths of both the physics-based HEC-HMS model and the LSTM network. The hybrid model was trained using daily streamflow and meteorological data from June 2017 to December 2020, and its performance was validated on an unseen dataset from January 2021 to April 2022. The integration of HEC-HMS simulated discharge and additional meteorological inputs into the LSTM network was evaluated using five statistical metrics. The validation results on the new unseen dataset showed the following performance: HEC-HMS achieved an RSR of 0.62, PBIAS of 8.65%, RMSLE of 0.64, and MAE of 17.49, while the LSTM model demonstrated improved performance with an RSR of 0.56, PBIAS of 8.54%, RMSLE of 0.51, and MAE of 15.80. In contrast, the hybrid HEC-HMS-LSTM model outperformed the others with an RSR of 0.50, PBIAS of 7.34%, RMSLE of 0.49, and MAE of 14.64, effectively capturing peak discharges and temporal trends. The findings highlight the hybrid model’s potential as a practical and efficient tool for improving flood prediction and water resource management in the Tlawng River basin. The novelty of this research lies in the advanced hyper-parameter tuning applied in a mountainous region, which enhances the model’s performance and adaptability to the complex hydrological dynamics of the area.