Patra et al. (2025) Long-term projections of global groundwater storage under future climate change scenarios using deep learning

Identification

• Journal: The Science of The Total Environment
• Year: 2025
• Date: 2025-11-29
• Authors: Sumriti Ranjan Patra, Hone‐Jay Chu, Tatas Tatas
• DOI: 10.1016/j.scitotenv.2025.181043

Research Groups

• Department of Geomatics, National Cheng Kung University, Tainan City, Taiwan.
• Civil Infrastructure Engineering Department, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia.

Short Summary

This study utilizes a deep learning model to project global groundwater storage (GWS) variations until 2100 under CMIP6 climate scenarios, identifying maximum temperature as the primary driver of depletion. The findings indicate that over 50% of the global population will reside in regions facing GWS decline by the end of the century, with tropical and temperate zones being the most vulnerable.

Objective

• To project long-term global variations in groundwater storage (GWS) under future climate change scenarios (SSPs) and identify the dominant climatic drivers influencing these changes.

Study Configuration

• Spatial Scale: Global (gridded analysis including major aquifers like the Ogallala Aquifer).
• Temporal Scale: Long-term projections from the present until 2100.

Methodology and Data

• Models used: Climate-induced AI model (Deep Learning/Neural Networks).
• Data sources: GRACE-derived (Gravity Recovery and Climate Experiment) groundwater storage data, CMIP6 (Coupled Model Intercomparison Project Phase 6) climate projections, and Shared Socioeconomic Pathways (SSPs).
• Variables: Maximum temperature ($T{max}$), minimum temperature ($T{min}$), and precipitation.

Main Results

• Model Performance: The AI model demonstrated high predictive accuracy with a Normalized Root Mean Square Error (NRMSE) < 0.1 and an Index of Agreement (IOA) > 0.9 across most global regions.
• Primary Drivers: Feature sensitivity analysis revealed that $T{max}$ is the dominant driver of GWS changes, followed by precipitation and $T{min}$.
• Scenario Projections: Under the high-emission SSP585 scenario, tropical and temperate regions face the most severe GWS declines, particularly when $T_{max}$ increases exceed 3 °C.
• Socioeconomic Impact: By 2100, more than 50% of the global population is projected to live in areas experiencing GWS loss.
• Aquifer Depletion: Major aquifers are at risk; specifically, the Ogallala Aquifer is projected to lose up to 40% of its storage under severe warming conditions.
• Regional Variations: While arid zones face moderately high losses, colder regions may experience slight gains in GWS.

Contributions

• Establishes a robust deep learning framework for global-scale groundwater projection that integrates GRACE satellite data with CMIP6 climate models.
• Quantifies the critical threshold of a 3 °C increase in $T_{max}$ as a tipping point for intensified groundwater depletion in specific climate zones.
• Provides a comprehensive assessment of the intersection between climate-induced groundwater stress and global population distribution.

Funding

• Not specified in the provided text.

Citation

@article{Patra2025Longterm,
  author = {Patra, Sumriti Ranjan and Chu, Hone‐Jay and Tatas, Tatas},
  title = {Long-term projections of global groundwater storage under future climate change scenarios using deep learning},
  journal = {The Science of The Total Environment},
  year = {2025},
  doi = {10.1016/j.scitotenv.2025.181043},
  url = {https://doi.org/10.1016/j.scitotenv.2025.181043}
}

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