Li et al. (2026) Reproducing GRACE Total Water Storage Change at Finer Spatial Scales

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

Identification

• Journal: Geophysical Research Letters
• Year: 2026
• Date: 2026-01-06
• Authors: F. Li, Jürgen Kusche
• DOI: 10.1029/2025gl119881

Research Groups

• Hydrology and Water Resources Research
• Remote Sensing and Satellite Geodesy
• Machine Learning and Data Science

Short Summary

This paper develops a novel machine-learning-based iterative downscaling method to enhance the spatial resolution of GRACE terrestrial water storage anomalies (TWSA) from approximately 3° to 0.25°, demonstrating improved agreement with in situ groundwater levels while preserving the original GRACE signal magnitude.

Objective

• To develop a machine-learning-based iterative downscaling method capable of reproducing terrestrial water storage anomalies (TWSA) at 0.25° resolution, while retaining nearly all of the original GRACE signals, using ERA5 soil moisture, precipitation, and temperature as inputs.

Study Configuration

• Spatial Scale: Global, with original GRACE data at approximately 3° resolution downscaled to 0.25° resolution.
• Temporal Scale: Long-term (implied by GRACE mission data, typically monthly anomalies).

Methodology and Data

• Models used: Machine-learning-based iterative downscaling method.
• Data sources:
  • Gravity Recovery and Climate Experiment (GRACE) satellite mission (Terrestrial Water Storage Anomalies - TWSA).
  • ERA5 reanalysis data (soil moisture, precipitation, temperature).
  • In situ groundwater levels (for validation).

Main Results

• The developed method successfully downscales GRACE TWSA to 0.25° resolution, retaining nearly all of the original GRACE signals.
• Downscaled TWSA shows improved agreement with in situ groundwater levels compared to original GRACE data:
  • Higher correlation at over 63% of wells globally.
  • Reduced root mean square error (RMSE) at more than 83% of wells globally.
• The downscaled TWSA retains an average correlation of 0.99 with original GRACE data at the basin scale.
• The method outperforms a previously released downscaling product.
• The downscaled TWSA dataset is publicly available.

Contributions

• Introduction of a novel machine-learning-based iterative downscaling method for GRACE TWSA that effectively addresses the coarse spatial resolution limitation.
• The method uniquely preserves the full magnitude of the original GRACE signals, a common challenge in existing downscaling approaches.
• Provides a higher-resolution (0.25°) global TWSA dataset with validated improvements against in situ groundwater observations.
• Demonstrates superior performance compared to existing downscaling products.
• Makes the high-resolution downscaled TWSA dataset publicly available, fostering further research and applications.

Funding

• Not specified in the provided abstract.

Citation

@article{Li2026Reproducing,
  author = {Li, F. and Kusche, Jürgen},
  title = {Reproducing GRACE Total Water Storage Change at Finer Spatial Scales},
  journal = {Geophysical Research Letters},
  year = {2026},
  doi = {10.1029/2025gl119881},
  url = {https://doi.org/10.1029/2025gl119881}
}