Akl et al. (2025) Global Groundwater Drought Assessment Revisited: A Holistic Re‐Evaluation of the GRACE‐Groundwater Drought Index Across Major Aquifers

Identification

• Journal: Water Resources Research
• Year: 2025
• Date: 2025-12-01
• Authors: Mohamed Akl, Brian F. Thomas, Peter J. Clarke
• DOI: 10.1029/2025wr040389

Research Groups

• Civil and Geospatial Engineering, School of Engineering, Newcastle University, Newcastle upon Tyne, UK
• Public Works Engineering Department, Faculty of Engineering, Tanta University, Tanta, Egypt
• Department of Earth Sciences, University College London, London, UK

Short Summary

This study holistically re-evaluates the GRACE-Groundwater Drought Index (GGDI) across 37 major aquifers by integrating multi-model GRACE-derived groundwater storage anomaly (GRACE-GWA) estimates. It reveals that variability among these estimates introduces substantial uncertainty into groundwater drought indicators and aquifer memory, compromising the reliability of single-model GGDI assessments.

Objective

• To assess the influence of multi-model GRACE-GWA realizations on key groundwater drought indicators.
• To assess aquifer memory variability for perceived groundwater drought.

Study Configuration

• Spatial Scale: 37 large aquifer systems globally, each exceeding 100,000 square kilometers (km²).
• Temporal Scale: April 2002 to December 2022.

Methodology and Data

• Models used:
  • GRACE-based solutions: Five GRACE/GRACE-FO solutions (CSR-SH, JPL-SH, GFZ-SH, CSR-M, JPL-M) for Terrestrial Water Storage Anomalies (GRACE-TWSA).
  • Land Surface Models (LSMs): NOAH (1° × 1°, 0.10° × 0.10°), Variable Infiltration Capacity (VIC) (1° × 1°), Catchment Land Surface Model (CLSM) (1° × 1°) from NASA Global Land Data Assimilation System (GLDAS V2.1) and Famine Early Warning System Network (FEWS NET) Land Data Assimilation System (FLDAS V001) for Snow Water Equivalent Anomalies (SWEA) and Soil Moisture Anomalies (SMA).
  • Hydrological Models: WaterGAP Global Hydrology Model v2.2e (WGHM) (0.5° × 0.5°) for SWEA and SMA.
  • Reanalysis: ERA5-Land (0.10° × 0.10°) for SWEA and SMA.
  • GGDI framework: Utilizes GRACE-GWA to calculate Groundwater Storage Deviation (GSD) and then standardizes it to derive GGDI.
• Data sources:
  • Satellite observations: GRACE and GRACE Follow-On missions.
  • Modeling and reanalysis systems: NASA GLDAS, FEWS NET FLDAS, WaterGAP, ERA5-Land.
  • Surface water storage: GloLakes, Copernicus Climate Change Service (lake water-level observations), Hydrological Data and Maps Based on Shuttle Elevation Derivatives at Multiple Scales (HydroSHEDS) for lake/reservoir extents, GLOBathy for lake bathymetry.

Main Results

• Uncertainty Propagation: Even modest discrepancies in GRACE-GWA methodologies lead to considerable uncertainties in drought indicators and aquifer memory. Normalization within GGDI mitigates some amplitude discrepancies but cannot fully resolve model divergences, especially during extreme hydrological events.
• Drought Indicator Variability: Maximum observed intra-basin discrepancies across 37 aquifers reached:
  • Number of drought events: 11 events.
  • Maximum duration: 122 months.
  • Average duration: 63.33 months.
  • Severity: 24.47.
  • Maximum intensity: 5.4.
• Aquifer Memory Variability: Aquifer memory, inferred from GGDI autocorrelation, showed pronounced variability, with estimates ranging from 3 to 61 months in the Nubian Basin.
• Climate Zone Influence: Humid basins generally experienced more frequent drought events with shorter, more consistent durations and severity. Arid basins recorded fewer events but with prolonged durations and more extreme intensity values, accompanied by high variability.
• Climatology Removal Impact: Removing seasonal climatology substantially influenced aquifer memory. In humid basins, it increased both aquifer memory and its variability by exposing slower, interannual fluctuations. In arid/semi-arid basins, it tended to reduce aquifer memory variability.
• Aquifer Memory and Drought Characteristics: Aquifers with higher memory generally experienced fewer drought events and lower severity, but those droughts were typically longer and more intense. Spearman correlation analysis showed average negative correlations between aquifer memory and drought event frequency/severity, and average positive correlations with maximum/average drought duration and maximum intensity, though with notable basin-specific exceptions.

Contributions

• Provides the first holistic re-evaluation of the GRACE-Groundwater Drought Index (GGDI) by systematically integrating multi-model GRACE-GWA estimates.
• Quantifies the substantial uncertainty introduced into groundwater drought indicators (number of events, duration, severity, intensity) and aquifer memory due to variations in GRACE processing methods and water budget component assumptions.
• Highlights the critical deficiencies and unreliability of previous single-model approaches for GRACE-based groundwater drought assessment.
• Offers a transparent and probabilistic multi-model framework for groundwater drought characterization, enhancing interpretability and credibility, especially in data-sparse regions.
• Illuminates the complex relationship between aquifer memory and groundwater drought characteristics, showing how memory influences drought frequency, duration, and intensity.

Funding

• PhD scholarship (Grant ID: MM59/19) awarded to the first author by the Egyptian Ministry of Higher Education and Scientific Research, represented by the Egyptian Bureau for Cultural and Educational Affairs in London.

Citation

@article{Akl2025Global,
  author = {Akl, Mohamed and Thomas, Brian F. and Clarke, Peter J.},
  title = {Global Groundwater Drought Assessment Revisited: A Holistic Re‐Evaluation of the GRACE‐Groundwater Drought Index Across Major Aquifers},
  journal = {Water Resources Research},
  year = {2025},
  doi = {10.1029/2025wr040389},
  url = {https://doi.org/10.1029/2025wr040389}
}