Sjursen et al. (2025) Machine learning improves seasonal mass balance prediction for unmonitored glaciers

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

Identification

• Journal: Repository for Publications and Research Data (ETH Zurich)
• Year: 2025
• Date: 2025-11-17
• Authors: Sjursen, Kamilla Hauknes, Bolibar, Jordi, van der Meer, Marijn, Andreassen, Liss Marie, Biesheuvel, Julian Peter, Dunse, Thorben, Huss, Matthias, Maussion, Fabien, Rounce, David R., Tober, Brandon
• DOI: 10.3929/ethz-c-000788212

Research Groups

Not explicitly mentioned in the provided text.

Short Summary

This study introduces the Mass Balance Machine (MBM), an XGBoost-based machine learning model, to provide accurate, high spatio-temporal resolution regional-scale reconstructions of glacier mass balance, demonstrating superior performance over traditional models for seasonal predictions on unmonitored glaciers.

Objective

• To develop and evaluate a machine learning model (Mass Balance Machine - MBM) for accurate, high spatio-temporal resolution regional-scale reconstructions of glacier mass balance, particularly for unmonitored glaciers, and to compare its generalization capabilities with traditional glacier evolution models.

Study Configuration

• Spatial Scale: Regional-scale, covering 32 glaciers across heterogeneous climate settings in mainland Norway.
• Temporal Scale: Spanning from 1962 to 2021 (60 years), focusing on seasonal and annual point mass balance measurements.

Methodology and Data

• Models used: Mass Balance Machine (MBM) based on the XGBoost architecture. Its predictions were compared against regional-scale simulations from three unnamed glacier evolution models.
• Data sources: A dataset of approximately 4000 seasonal and annual point mass balance measurements collected from 32 glaciers.

Main Results

• MBM successfully predicted annual and seasonal point mass balance on independent test glaciers with a Root Mean Square Error (RMSE) ranging from 0.59 m w.e. to 1.00 m w.e. and a bias from -0.01 m w.e. to 0.04 m w.e.
• MBM outperformed traditional glacier evolution models in predicting seasonal mass balance across various spatial scales.
• The model reduced the RMSE by up to 46 % for glacier-wide winter mass balance and up to 25 % for glacier-wide summer mass balance compared to other models.

Contributions

• Introduction of the Mass Balance Machine (MBM), a novel data-driven, machine learning approach (XGBoost-based) for regional-scale glacier mass balance reconstruction.
• Demonstration of the capability of machine learning models to generalize effectively across diverse glaciers and climatic settings using relatively sparse mass balance data.
• Significant improvement in the accuracy of seasonal mass balance predictions compared to existing glacier evolution models, particularly beneficial for unmonitored glaciers.

Funding

Not explicitly mentioned in the provided text.

Citation

@article{Sjursen2025Machine,
  author = {Sjursen, Kamilla Hauknes and Bolibar, Jordi and van der Meer, Marijn and Andreassen, Liss Marie and Biesheuvel, Julian Peter and Dunse, Thorben and Huss, Matthias and Maussion, Fabien and Rounce, David R. and Tober, Brandon},
  title = {Machine learning improves seasonal mass balance prediction for unmonitored glaciers},
  journal = {Repository for Publications and Research Data (ETH Zurich)},
  year = {2025},
  doi = {10.3929/ethz-c-000788212},
  url = {https://doi.org/10.3929/ethz-c-000788212}
}