Thébault et al. (2026) Comparing multi-model mosaic and multi-model combination methods to simulate streamflow across the contiguous USA

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-06-29
• Authors: Cyril Thébault, Wouter Johannes Maria Knoben, Nans Addor, Andrew J. Newman, Diana Spieler, Nicolás Vasquéz, Yalan Song, Gaby J. Gründemann, Shaun T. Carney, Mukesh Kumar, Katie van Werkhoven, Chaopeng Shen, Andrew W. Wood, Martyn Clark
• DOI: 10.5194/hess-30-3945-2026

Research Groups

• Schulich School of Engineering, University of Calgary, Calgary, Alberta, Canada
• Fathom, Bristol, UK
• Geography, University of Exeter, Exeter, UK
• RAL, NSF National Center for Atmospheric Research, Boulder, Colorado, USA
• Civil and Environmental Engineering, The Pennsylvania State University, University Park, Pennsylvania, USA
• Research Triangle Institute, Research Triangle Park, North Carolina, USA
• Civil, Construction, and Environmental Engineering, University of Alabama, Tuscaloosa, Alabama, USA

Short Summary

This study compares multi-model mosaic and combination methods to simulate streamflow across the contiguous United States. The results show that a carefully chosen single model from a larger ensemble can closely approach the performance of more complex multi-model strategies.

Objective

• To compare the performance of multi-model mosaic and combination methods in simulating streamflow across the contiguous United States.
• To investigate the benefits and limitations of using multiple models to enhance predictive reliability and inform decision-making.

Study Configuration

• Spatial Scale: Continental (contiguous United States)
• Temporal Scale: Daily time step, with a calibration period from 1989-1998 and an evaluation period from 1999-2009

Methodology and Data

• Models used: FUSE (Framework for Understanding Structural Errors) ensemble of 78 hydrological models
• Data sources: CAMELS dataset, Daymet meteorological forcing product

Main Results

• The results show that a carefully chosen single model from the FUSE ensemble can closely approach the performance of more complex multi-model strategies.
• Multi-model combinations and mosaics exhibit broadly similar overall skill, although combination approaches deliver slightly higher performance and lower sampling uncertainty.
• However, per-catchment differences persist, indicating that no single multi-model strategy dominates everywhere.

Contributions

• This study provides a systematic comparison of multi-model mosaic and combination methods in simulating streamflow across the contiguous United States.
• The results highlight the potential benefits of using multiple models to enhance predictive reliability and inform decision-making.

Funding

• Not specified

Citation

@article{Thébault2026Comparing,
  author = {Thébault, Cyril and Knoben, Wouter Johannes Maria and Addor, Nans and Newman, Andrew J. and Spieler, Diana and Vasquéz, Nicolás and Song, Yalan and Gründemann, Gaby J. and Carney, Shaun T. and Kumar, Mukesh and Werkhoven, Katie van and Shen, Chaopeng and Wood, Andrew W. and Clark, Martyn},
  title = {Comparing multi-model mosaic and multi-model combination methods to simulate streamflow across the contiguous USA},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-3945-2026},
  url = {https://doi.org/10.5194/hess-30-3945-2026}
}