Aghababaei et al. (2025) Development and Comparison of Methods for Identification of Baseflow-Dominant Periods in Streamflow Records

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Water
• Year: 2025
• Date: 2025-10-28
• Authors: Amin Aghababaei, Norman L. Jones, Gustavious P. Williams, Eniola Webster-Esho, Ryan van der Heijden, Xueyi Li, T. Prabhakar Clement, Donna M. Rizzo
• DOI: 10.3390/w17213083

Research Groups

Not explicitly stated in the provided text.

Short Summary

This study defines baseflow-dominant (BFD) periods and develops an expert-labeled dataset from 182 USGS stream gages to evaluate automated BFD identification methods. It demonstrates that a machine learning model (RF-BFD) significantly outperforms other approaches, achieving an F1 score of 0.92 and 92% accuracy, thereby establishing benchmarks for improved large-scale hydrological assessments.

Objective

• To define baseflow-dominant (BFD) periods as flow conditions with minimal quickflow contribution.
• To develop a comprehensive, expert-labeled dataset of BFD periods for diverse hydrological settings.
• To evaluate and benchmark various automated BFD identification methods, including new machine learning, gradient-based, and statistical approaches, against established techniques.

Study Configuration

• Spatial Scale: 182 USGS stream gages across diverse hydrological settings in the continental United States.
• Temporal Scale: Not explicitly stated for the duration of records, but covers historical streamflow.

Methodology and Data

• Models used: Machine learning classifier (RF-BFD), gradient-based method, statistical method, BN77 method, Strict Baseflow method.
• Data sources: Expert-labeled dataset of BFD periods derived from streamflow records of 182 USGS stream gages.

Main Results

• The machine learning model (RF-BFD) demonstrated superior performance in identifying BFD periods compared to all other evaluated approaches.
• The RF-BFD model achieved an F1 score of 0.92 and an accuracy of 92%.
• The study characterized challenges inherent in identifying BFD periods and established performance benchmarks for future improvements in large-scale hydrological studies.

Contributions

• Provides a clear definition of baseflow-dominant (BFD) periods, including conditions dominated by bank flow, groundwater interaction, or residual flow routing.
• Creates a novel, comprehensive, expert-labeled ground-truth dataset for BFD periods from 182 USGS stream gages.
• Introduces and evaluates three new automated BFD identification methods (machine learning, gradient-based, statistical).
• Establishes the machine learning model (RF-BFD) as the most robust and accurate method for BFD identification, setting new benchmarks.
• Offers a pathway for more robust and scalable BFD identification techniques, enhancing low-flow forecasting and groundwater-surface water interaction assessments.

Funding

Not explicitly stated in the provided text.

Citation

@article{Aghababaei2025Development,
  author = {Aghababaei, Amin and Jones, Norman L. and Williams, Gustavious P. and Webster-Esho, Eniola and Heijden, Ryan van der and Li, Xueyi and Clement, T. Prabhakar and Rizzo, Donna M.},
  title = {Development and Comparison of Methods for Identification of Baseflow-Dominant Periods in Streamflow Records},
  journal = {Water},
  year = {2025},
  doi = {10.3390/w17213083},
  url = {https://doi.org/10.3390/w17213083}
}