Cao et al. (2026) Assessing the response of terrestrial water storage to climate warming in China by coupling CMIP6 multi-model ensembles, hydrological model, and machine learning algorithms

Identification

• Journal: Climatic Change
• Year: 2026
• Date: 2026-01-29
• Authors: Xiaoyan Cao, Jiali Ju, Chuanhao Wu, Pat J.-F. Yeh, Min Shi, Ashraf Dewan, Yongze Song, Xueyuan Zhang, Tian Yao, Yufei Jiao, Qiongfang Li, Shanshui Yuan, Xiaolei Fu, Bill X. Hu
• DOI: 10.1007/s10584-025-04082-4

Research Groups

• School of Water Conservancy and Environment, University of Jinan, Jinan, Shandong, China
• School of Water Resources and Environment, China University of Geosciences, Beijing, China
• The National Key Laboratory of Water Disaster Prevention, Hohai University, Nanjing, China
• Yangtze Institute for Conservation and Development, Hohai University, Nanjing, China
• Department of Civil Engineering, School of Engineering, Monash University Malaysia Campus, Bandar Sunway, Malaysia
• School of Environment and Climate, Jinan University, Guangzhou, China
• Spatial Sciences Discipline, School of Earth and Planetary Sciences, Curtin University, Perth, WA, Australia
• School of Design and the Built Environment, Curtin University, Bentley, Australia
• College of Earth and Environmental Sciences, Lanzhou University, Lanzhou, Gansu, China
• College of Hydraulic Science and Engineering, Yangzhou University, Yangzhou, China

Short Summary

This study develops a new framework for comprehensive terrestrial water storage (TWS) projection and attribution in China by integrating CMIP6 multi-model ensembles, a hydrological model, and machine learning algorithms. It projects an intensified water cycle and increasing TWS in most of China, but a declining TWS in western China, with Global Climate Model (GCM) uncertainty dominating projections.

Objective

• To assess the spatiotemporal changes and trends of TWS in response to climate change across China.
• To quantify the relative contributions of GCM and Shared Socioeconomic Pathway (SSP) scenario uncertainties to the total uncertainty in TWS projections.
• To explore the dominant meteorological and hydrological factors influencing projected future TWS changes.

Study Configuration

• Spatial Scale: China, specifically ten major river basins, at a 0.25° × 0.25° grid resolution.
• Temporal Scale: Historical baseline (1985–2014), near future (2030–2059), and far future (2070–2099) under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios.

Methodology and Data

• Models used:
  • CMIP6 Global Climate Models (18 GCMs)
  • Variable Infiltration Capability (VIC) hydrological model (version 4.1.2d)
  • Multi-Layer Perceptron Artificial Neural Network (MLP-ANN) model
  • Hierarchical Sensitivity Analysis Framework (HSAF)
  • Extreme Gradient Boosting-Shapley Additive exPlanations (XGBoost-SHAP) model
  • Empirical Quantile Mapping (for precipitation bias correction)
  • Variance-based Bias Correction (VB-BC) method (for temperature bias correction)
• Data sources:
  • CMIP6 GCM outputs (daily precipitation, maximum, minimum, and mean temperature)
  • China Meteorological Forcing Dataset (CMFD) (daily precipitation, maximum, minimum, and mean temperature, 0.1° resolution)
  • GRACE and GRACE-FO satellite data (CSR-M and JPL-M mascon products for monthly TWS anomaly)
  • Reconstructed GRACE data (Zhong et al. 2019) for data gaps
  • Observed runoff data from 17 hydrologic stations in China

Main Results

• Climate warming is projected to intensify the water cycle in China (2030–2099), with precipitation increasing by 5.59 to 21.09 mm/10a and evapotranspiration by 2.48 to 11.61 mm/10a.
• Western China is projected to experience a greater water cycle intensification rate, with precipitation and evapotranspiration increasing by 50% and 70% respectively by the end of the century under SSP5-8.5.
• Terrestrial Water Storage (TWS) shows a significant increasing trend (2030–2099) under SSP1-2.6 (~1.32 mm/10a) and SSP2-4.5 (~1.81 mm/10a), but not under SSP5-8.5 (0.79 mm/10a).
• Western China is expected to experience a declining TWS trend (2030–2099) under all SSP scenarios, most pronounced under SSP5-8.5 (3.21 mm/10a), indicating increased water scarcity.
• Global Climate Model (GCM) uncertainty is the primary source of TWS projection uncertainty, contributing over 57% of the total uncertainty across the ten river basins, decreasing from the near future to the far future.
• Shared Socioeconomic Pathway (SSP) scenario uncertainty is smaller (18–43%) but becomes more pronounced over time.
• Attribution analysis indicates that precipitation dominates future TWS variability, followed by soil moisture and temperature, while snow water equivalent shows the least impact.
• The contribution of precipitation to TWS variability is projected to decrease under all SSP scenarios relative to the historical period.

Contributions

• Developed a novel, comprehensive framework for TWS projection and attribution by coupling CMIP6 multi-model ensembles, a distributed hydrological model (VIC), and machine learning algorithms (ANN, XGBoost-SHAP).
• Provided improved reliability for large-scale TWS projections in China under future climate change scenarios.
• Quantified the relative contributions of GCMs and SSP scenarios to the total uncertainty in TWS projections, highlighting the dominance of GCM uncertainty.
• Identified the dominant meteorological and hydrological factors influencing future TWS changes in China and analyzed their changing contributions under different SSP scenarios.

Funding

• Basic Research Program of Jiangsu (Grant No. BK20252047)
• CHINA SCHOLARSHIP COUNCIL (202408370234)
• Shandong Provincial Natural Science Foundation (ZR2023QD090)
• Study on the mutual feed mechanism and coupling regulation of ecological water quantity and quality (IWHR-SKL-KF202318)
• Study on the interaction mechanism and coupling simulation of the surface water and groundwater (XBS2460)
• IWHR Research & Development Support Program (Grant No. JZ0199A022021)
• National Key Research and Development Program of China (Grant No. 2024YFC3082200)

Citation

@article{Cao2026Assessing,
  author = {Cao, Xiaoyan and Ju, Jiali and Wu, Chuanhao and Yeh, Pat J.-F. and Shi, Min and Dewan, Ashraf and Song, Yongze and Zhang, Xueyuan and Yao, Tian and Jiao, Yufei and Li, Qiongfang and Yuan, Shanshui and Fu, Xiaolei and Hu, Bill X.},
  title = {Assessing the response of terrestrial water storage to climate warming in China by coupling CMIP6 multi-model ensembles, hydrological model, and machine learning algorithms},
  journal = {Climatic Change},
  year = {2026},
  doi = {10.1007/s10584-025-04082-4},
  url = {https://doi.org/10.1007/s10584-025-04082-4}
}