Saemian et al. (2026) A Machine Learning approach for Total Water storage anomaly eXtension back to 1980 (ML-TWiX)

⚠️ 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: 2026
• Date: 2026-01-29
• Authors: Saemian, Peyman, Tourian, Mohammad J., Douch, Karim, Foster, James, Gou, Junyang, Wiese, David, Aghakouchak, Amir, Sneeuw, Nico
• DOI: 10.3929/ethz-c-000795546

Research Groups

Not explicitly stated in the provided text.

Short Summary

ML-TWiX is a global dataset of monthly total water storage anomalies (TWSA) reconstructed from 1980 to 2012, extending the GRACE-era record by using an ensemble of machine learning models trained on global hydrological and land surface model simulations.

Objective

• To reconstruct a long-term, continuous global dataset of monthly total water storage anomalies (TWSA) for the pre-GRACE period (1980-2012) to overcome the limited temporal span of satellite gravity missions.

Study Configuration

• Spatial Scale: Global, 0.5 degrees x 0.5 degrees grid.
• Temporal Scale: Monthly, from 1980 to 2012.

Methodology and Data

• Models used: Ensemble of three machine learning models: Random Forest, XGBoost, and Gaussian Process Regression.
• Data sources:
  • Input for ML models: Global hydrological and land surface model simulations.
  • Validation data: Satellite laser ranging, storage deduced from water mass balance closure, and global mean sea level budget estimates.

Main Results

• Development and release of ML-TWiX, a global dataset of monthly total water storage anomalies (TWSA) from 1980 to 2012.
• The dataset provides a continuous reconstruction of TWSA on a 0.5° x 0.5° grid with spatially explicit uncertainty estimates.
• ML-TWiX was validated against multiple independent datasets, demonstrating its reliability for long-term hydrological and climate studies.

Contributions

• Provides the first long-term, continuous global reconstruction of monthly TWSA for the pre-GRACE period (1980-2012), significantly extending the existing satellite-derived record.
• Offers a valuable resource for long-term climate and hydrological studies, and water resource assessment, addressing a critical data gap.
• Introduces an ensemble machine learning approach for reconstructing historical TWSA from model simulations.

Funding

Not explicitly stated in the provided text.

Citation

@article{Saemian2026Machine,
  author = {Saemian, Peyman and Tourian, Mohammad J. and Douch, Karim and Foster, James and Gou, Junyang and Wiese, David and Aghakouchak, Amir and Sneeuw, Nico},
  title = {A Machine Learning approach for Total Water storage anomaly eXtension back to 1980 (ML-TWiX)},
  journal = {Repository for Publications and Research Data (ETH Zurich)},
  year = {2026},
  doi = {10.3929/ethz-c-000795546},
  url = {https://doi.org/10.3929/ethz-c-000795546}
}