Abbaszadeh et al. (2026) GNSS evaluation of GRACE-assimilated water storage models over 89 river basins worldwide

Identification

• Journal: Scientific Reports
• Year: 2026
• Authors: Majid Abbaszadeh, Tonie van Dam
• DOI: 10.1038/s41598-025-31887-1

Research Groups

• Faculty of Science, Technology and Medicine, University of Luxembourg, Luxembourg.
• College of Science, Geology and Geophysics, University of Utah, Salt Lake City, USA.

Short Summary

This study evaluates two global GRACE-assimilated (GA) hydrological models, GLWS2.0 and CLSM-DA, using vertical displacement data from over 9,000 GNSS stations across 89 river basins. The results indicate that CLSM-DA generally provides a more accurate representation of seasonal water storage variations, particularly regarding amplitude in tropical regions and phase timing in mid-latitudes.

Objective

• To assess the quality and consistency of GRACE-assimilated terrestrial water storage (TWS) datasets (GLWS2.0 and CLSM-DA) by comparing their predicted hydrological loading displacements against independent GNSS geodetic observations at a global and river-basin scale.

Study Configuration

• Spatial Scale: Global, with specific clustering and analysis across 89 major river basins using 9,163 GNSS stations.
• Temporal Scale: January 2004 to December 2019 (16 years).

Methodology and Data

• Models used:
  • GLWS2.0: Global Land Water Storage release 2.0 (assimilating GRACE into the WGHM2.2e model).
  • CLSM-DA: Catchment Land Surface Model GRACE Data Assimilation (assimilating GRACE into CLSM-F2.5).
  • Loading Model: Preliminary Reference Earth Model (PREM) Green’s functions used to convert water mass to vertical elastic displacement.
• Data sources:
  • GNSS: Daily vertical coordinate time series from the Nevada Geodetic Laboratory (NGL) and IGS repro3.
  • Satellite Gravimetry: GRACE and GRACE-FO terrestrial water storage anomalies (TWSA).
  • Ancillary Data: HydroBASINS (river basin boundaries), ESMGFZ (non-tidal atmospheric and ocean loading corrections), and ERA5 (precipitation reanalysis).

Main Results

• Inter-model Discrepancy: The mean annual amplitude of equivalent water thickness (EWT) between the two models differs by >25 mm over 40% of the land area, with phase shifts >30 days across 50% of the domain.
• Regional Performance: CLSM-DA outperformed GLWS2.0 in 25 basins (notably in Africa, Southeast Asia, and South America), while GLWS2.0 was superior in 10 basins (primarily in the Eastern U.S. and Western Europe).
• Phase Accuracy: GLWS2.0 showed a 30–60 day phase lead (early peak) in the Western U.S. and Eastern Europe; CLSM-DA’s delayed phase aligned better with GNSS observations in these regions.
• Amplitude Variations: CLSM-DA estimated 100–300 mm larger annual water variations in tropical basins, which more closely matched the high-amplitude displacements recorded by GNSS.
• Extreme Events: Both models struggled to capture the full magnitude of extreme events, underestimating water storage loss during the 2011–2016 Western U.S. drought by approximately 30–50%.

Contributions

• Provides the first comprehensive global evaluation of GRACE-assimilated models using a massive, high-density GNSS dataset (NGL) rather than limited IGS core stations.
• Establishes a robust framework for regional model assessment by clustering geodetic stations within river basin boundaries.
• Identifies specific systematic biases in current GA products, such as the phase lead in GLWS2.0 and the general underestimation of sub-seasonal and extreme hydrological signals.

Funding

• Luxembourg National Research Fund (FNR), grant reference: INTER/DFG/20/15193324/GlobalCDA/Van Dam.

Citation

@article{Abbaszadeh2026GNSS,
  author = {Abbaszadeh, Majid and Dam, Tonie van},
  title = {GNSS evaluation of GRACE-assimilated water storage models over 89 river basins worldwide},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-025-31887-1},
  url = {https://doi.org/10.1038/s41598-025-31887-1}
}

Generated by BiblioAssistant using gemini-3-flash-preview (Google API)