Guo et al. (2026) Implementation and Evaluation of Parallel Computing Approaches for Large‐Domain, Process‐Based Hydrologic Simulations

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

Identification

• Journal: Journal of Advances in Modeling Earth Systems
• Year: 2026
• Date: 2026-01-01
• Authors: Junwei Guo, Martyn P. Clark, Wouter J. M. Knoben, Kasra Keshavarz, Kyle Klenk, Ashley Van Beusekom, Victoria Guenter, Raymond J. Spiteri
• DOI: 10.1029/2025ms005064

Research Groups

Not explicitly mentioned in the abstract, but likely computational hydrology and Earth system modeling groups.

Short Summary

This study implements and compares various parallelization approaches (MPI, OMP, Actor Model) within the SUMMA hydrologic solver to enable high-performance computing for large-domain simulations. It finds that SUMMA-MPI scales linearly up to 1,024 cores with load balancing improving utilization to 95%, while OMP is for smaller core counts and the Actor Model provides excellent fault tolerance.

Objective

• To implement and provide detailed guidelines for various parallelization approaches (Message Passing Interface (MPI), Open Multi-Processing (OMP), and the Actor Model) within the process-based hydrologic solver SUMMA.
• To compare the scalability, computational cost, input/output performance, and coupling capabilities of these parallel approaches against the original sequential approach using a North American hydrologic simulation as a case study.

Study Configuration

• Spatial Scale: North America
• Temporal Scale: Not explicitly mentioned in the abstract; "historical runs" are used for load balancing calibration.

Methodology and Data

• Models used: SUMMA (Structure for Unifying Multiple Modeling Alternatives)
• Data sources: Not explicitly mentioned for the primary simulation data; historical runs are used for load-balancing calibration.

Main Results

• SUMMA-MPI exhibits linear scaling up to 1,024 cores.
• SUMMA-OMP is recommended only for smaller numbers of cores.
• The MPI approach initially showed a straggler effect, resulting in core utilization of only 80%.
• A load-balancing calibration, based on historical runs, increased SUMMA-MPI core usage to 95%, mitigating the straggler effect.
• MPI is the most effective for large-scale simulations involving multiple nodes and extensive core counts, supporting strong coupling and synchronization.
• The Actor Model demonstrates excellent fault tolerance, allowing automatic modification and recommencement of specific Grouped Response Units (GRUs) without restarting the entire simulation in case of failure.

Contributions

• Provides detailed guidelines for implementing MPI, OMP, and the Actor Model parallelization schemes within a process-based hydrologic solver (SUMMA).
• Offers a comprehensive comparison of the scalability, computational cost, input/output performance, and coupling capabilities of these parallel approaches for large-domain hydrologic simulations.
• Introduces and demonstrates a load-balancing calibration method to improve core utilization and mitigate the straggler effect in MPI-based hydrologic simulations.
• Documents the advantages and limitations of multiple parallelization schemes, offering insights for advancing computational hydrology in the Earth System Science community.

Funding

Not mentioned in the abstract.

Citation

@article{Guo2026Implementation,
  author = {Guo, Junwei and Clark, Martyn P. and Knoben, Wouter J. M. and Keshavarz, Kasra and Klenk, Kyle and Beusekom, Ashley Van and Guenter, Victoria and Spiteri, Raymond J.},
  title = {Implementation and Evaluation of Parallel Computing Approaches for Large‐Domain, Process‐Based Hydrologic Simulations},
  journal = {Journal of Advances in Modeling Earth Systems},
  year = {2026},
  doi = {10.1029/2025ms005064},
  url = {https://doi.org/10.1029/2025ms005064}
}