Rahman et al. (2026) Unmasking human-subsidized resilience through hydrological drivers of water use efficiency during compound droughts across the North China Plain

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2026
• Date: 2026-03-31
• Authors: Khalil Ur Rahman, Shahanshah Abbas, Deqiang Mao, Songhao Shang, Quoc Bao Pham, Aftab Haider Khan, Anwar Hussain
• DOI: 10.1016/j.ejrh.2026.103394

Research Groups

• School of Civil Engineering, Shandong University, China: Khalil Ur Rahman, Shahanshah Abbas, Deqiang Mao
• Department of Hydraulic Engineering, Tsinghua University, China: Songhao Shang, Aftab Haider Khan
• Faculty of Natural Sciences, Institute of Earth Sciences, University of Silesia in Katowice, Poland: Quoc Bao Pham
• Department of Economics and Development Studies, University of Swat, Pakistan: Anwar Hussain

Short Summary

This study quantifies water use efficiency (WUE) in North China Plain (NCP) provinces and analyzes its resilience and vulnerability to multi-type droughts, revealing that human-subsidized resilience temporarily masks meteorological deficits but collapses non-linearly under deep groundwater depletion.

Objective

• To evaluate the response of water use efficiency (WUE) under different compound drought events (Standardized Precipitation Evapotranspiration Index–Standardized Soil Moisture Index (SPEI–SSMI), SSMI–Standardized Water Availability Index (SWAI), and SSMI–GRACE-based Groundwater Drought Index (GGDI)).
• To quantify the contribution of each drought type severely impacting the WUE in the SPEI–SSMI–SWAI–GGDI nexus.
• To assess the sensitivity of WUE to each drought at district scale.
• To quantify the positive proportion of droughts that enhance the WUE and their prospective trends.

Study Configuration

• Spatial Scale: Five provinces/municipalities in the North China Plain (Beijing, Tianjin, Shandong, Hebei, Henan), covering approximately 5.412 × 10^5 square kilometers. Data aggregated to district scale, with a spatial resolution of 0.1° × 0.1° for most datasets.
• Temporal Scale: 2002–2024. Monthly data for water use efficiency and drought indices, analyzed at 1-, 3-, 6-, and 12-month time scales.

Methodology and Data

• Models used:
  • Bayesian Generalized Additive Models (BGAM) and reduced-BGAM frameworks.
  • Extreme Gradient Boosting (XGBoost) machine learning model (for GRACE data downscaling).
  • Random forest algorithm (for remote sensing data bias correction).
  • Long Short-Term Memory (LSTM) (for in-situ data gap filling).
  • Thornthwaite equation (for potential evapotranspiration (PET) estimation).
• Data sources:
  • In-situ data: Daily climate variables (precipitation, maximum and minimum temperature, land surface temperature, pan evaporation, surface pressure, wind speed) from 64 stations via the China Meteorological Data Service Centre.
  • Remote sensing and reanalysis:
    • ERA5-Land (ECMWF): Precipitation, maximum and minimum temperature, relative humidity, surface pressure, evaporation, soil moisture (0–7 cm and 7–28 cm layers).
    • MODIS (MOD17A2): Gross Primary Productivity (GPP) (8-day, 1 km resolution).
    • MODIS (MOD13A2): Normalized Difference Vegetation Index (NDVI) (16-day, 1 km resolution).
    • MODIS (MOD16A2): Evapotranspiration (ET) (8-day, 500 m resolution).
    • GRACE/GRACE-FO (CSR RL06.3): Terrestrial Water Storage (TWS) (0.25° × 0.25° resolution).
    • Global Land Data Assimilation System (GLDAS v2.2): Snow Water Equivalent, Surface Water Storage, Canopy Water Storage, Soil Moisture Storage (0.25° × 0.25° resolution).
    • Landsat-Derived Global Rainfed and Irrigated-Cropland Product (LGRIP): Proportion of irrigated and rainfed cropland.
    • MODIS (MOD16A2GF.061): Crop Water Stress Index (CWSI).
  • Drought Indices: Standardized Precipitation Evapotranspiration Index (SPEI), Standardized Soil Moisture Index (SSMI), Standardized Water Availability Index (SWAI), GRACE-based Groundwater Drought Index (GGDI).

Main Results

• Drought propagation is non-linear and significantly altered by anthropogenic interventions, which temporarily decouple crop physiology from meteorological deficits.
• At the 1-month scale, SSMI and SWAI explain 40–60% of WUE variance in irrigated districts, demonstrating an anthropogenic masking effect.
• Agro-ecosystem resilience is strictly conditional on groundwater dynamics; deep groundwater depression cones (e.g., Hebei) exhibit rigid physiological points where WUE anomalies collapse steeply and non-linearly.
• Transient WUE stability is maintained by continuous anthropogenic surface subsidies (e.g., South-to-North Water Diversion Project, groundwater pumping), but long-term resilience is primarily constrained by deep groundwater storage and aquifer integrity.
• Positive proportions of drought (where WUE increases) range from 2.5% to 69.6% at the 1-month scale, with both positive proportion and mean slope increasing significantly at the 6-month scale, indicating cumulative hydrological memory.
• GGDI is a dominant driver of WUE variability, explaining over 50% of variance at the 1-month scale in some districts (Jizhou, Jining, Shangqiu) and 45–60% at the 6-month scale in others (Huairou, Dezhou, Hengshui, Baoding, Zhengzhou, Shangqiu).
• Sensitivity of WUE to SPEI is low-to-moderate, while sensitivity to SSMI is markedly higher across irrigated plains. GGDI sensitivity is highest and most spatially coherent in groundwater-dependent core regions (central/northern Hebei, southwestern Henan, eastern Shandong), intensifying at longer timescales.

Contributions

• Challenges the assumption of natural drought propagation by empirically demonstrating how anthropogenic interventions (e.g., South-to-North Water Diversion Project, irrigation) temporarily decouple meteorological deficits from actual hydrological signals, thereby questioning the reliability of standalone drought indices in highly managed catchments.
• Introduces a non-linear Bayesian Generalized Additive Model (BGAM) framework to identify abrupt ecological thresholds and "physical tipping points" where human-subsidized resilience collapses into acute hydraulic failure, providing quantitative assessment of conditions under which groundwater buffering becomes insufficient.
• Addresses severe multi-collinearity in multi-index drought studies using a Reduced-BGAM framework, disentangling the independent physiological impacts of deep groundwater versus root-zone soil moisture and revealing deep groundwater deficits as a distinct silent stressor that can trigger sharp WUE declines even with adequate topsoil moisture.
• Provides a comprehensive, district-scale quantification of drought contributions, sensitivity, and prospective trends on WUE, offering actionable insights for optimizing irrigation and water resource management under climate change.

Funding

• National Natural Science Foundation of China (Grant No. 51839006)
• Basic Research Grant from the Shandong University (Grant No. 31410061340019)

Citation

@article{Rahman2026Unmasking,
  author = {Rahman, Khalil Ur and Abbas, Shahanshah and Mao, Deqiang and Shang, Songhao and Pham, Quoc Bao and Khan, Aftab Haider and Hussain, Anwar},
  title = {Unmasking human-subsidized resilience through hydrological drivers of water use efficiency during compound droughts across the North China Plain},
  journal = {Journal of Hydrology Regional Studies},
  year = {2026},
  doi = {10.1016/j.ejrh.2026.103394},
  url = {https://doi.org/10.1016/j.ejrh.2026.103394}
}