Aalstad et al. (2026) Evolving beyond collapse: an adaptive particle batch smoother for cryospheric data assimilation

Identification

• Journal: Geoscientific model development
• Year: 2026
• Date: 2026-09-16
• Authors: Kristoffer Aalstad, Esteban Alonso‐González, Norbert Pirk, Sebastian Westermann, Clarissa Willmes, Ruitang Yang
• DOI: 10.5194/gmd-19-8565-2026

Research Groups

• Department of Geosciences, University of Oslo (UiO), Norway
• Instituto Pirenaico de Ecología, Consejo Superior de Investigaciones Científicas (IPE-CSIC), Spain

Short Summary

This study presents an adaptive particle-based data assimilation scheme for cryospheric applications that leverages developments in importance sampling. The proposed approach combines the advantages of particle methods and iterative ensemble Kalman methods to improve resilience against ensemble collapse and enable early-stopping strategies.

Objective

• Develop a robust and reliable tool for cryospheric data assimilation that can handle complex cases with dense observational datasets.
• Improve upon existing algorithms by reducing computational cost while maintaining accurate uncertainty quantification.

Study Configuration

• Spatial Scale: Global to regional scales, focusing on snow-covered areas in high-latitude regions.
• Temporal Scale: Annual to multi-year timescales, considering seasonal and interannual variability.

Methodology and Data

• Models used: Flexible Snow Model (FSM2) and the Multiple Snow Data Assimilation System (MuSA) toolbox.
• Data sources: Satellite observations from the ESMSnowMIP project, ground-based measurements, and reanalysis datasets.

Main Results

• The adaptive particle batch smoother (AdaPBS) outperforms or matches the performance of other commonly used algorithms in cryospheric data assimilation experiments.
• AdaPBS successfully handles complex cases with dense observational datasets, demonstrating its robustness and reliability.
• The scheme's ability to adaptively adjust computational cost based on problem complexity reduces overall processing time.

Contributions

• This study contributes a novel adaptive particle-based data assimilation method for cryospheric applications, addressing the limitations of traditional particle methods.
• AdaPBS offers improved resilience against ensemble collapse and enables early-stopping strategies, reducing computational cost while maintaining accurate uncertainty quantification.

Funding

• This research was supported by the European Research Council (ERC) under grant agreement No. 725273 (DROUGHT-HEAT).
• Additional funding was provided by the Norwegian Research Council (NFR) through project No. 262622.
• The work of E.A.-G. was also supported by the Spanish Ministry of Science, Innovation and Universities (MCIU) under grant agreement No. PID2020-113765GB-I00.

Citation

@article{Aalstad2026Evolving,
  author = {Aalstad, Kristoffer and Alonso‐González, Esteban and Pirk, Norbert and Westermann, Sebastian and Willmes, Clarissa and Yang, Ruitang},
  title = {Evolving beyond collapse: an adaptive particle batch smoother for cryospheric data assimilation},
  journal = {Geoscientific model development},
  year = {2026},
  doi = {10.5194/gmd-19-8565-2026},
  url = {https://doi.org/10.5194/gmd-19-8565-2026}
}