Zhang et al. (2025) Unraveling threshold effects and feature contributions in hydrological-soil-vegetation response to Budyko parameters: A data-driven perspective

Identification

• Journal: Journal of Hydrology
• Year: 2025
• Date: 2025-11-19
• Authors: Yiran Zhang, Tingxi Liu, Asaad Y. Shamseldin, Xin T. Tong, Limin Duan, Tianyu Jia, Yongzhi Bao, Shuo Lun
• DOI: 10.1016/j.jhydrol.2025.134612

Research Groups

• State Key Laboratory of Water Engineering Ecology and Environment in Arid Area, Inner Mongolia Agricultural University, Hohhot 010018, China
• Inner Mongolia Key Laboratory of Ecohydrology and High-Efficient Utilization of Water Resources, College of Water Conservancy and Civil Engineering, Inner Mongolia Agricultural University, Hohhot 010018, China
• Autonomous Region Collaborative Innovation Center for Integrated Management of Water Resources and Water Environment in the Inner Mongolia Reaches of the Yellow River, Hohhot 010018, China
• Department of Civil and Environmental Engineering, The University of Auckland, Auckland 1010, New Zealand

Short Summary

This study investigated the threshold effects and feature contributions of hydrological, soil, and vegetation elements on watershed ecohydrological processes in a sandy region using a data-driven approach. It identified critical thresholds for key ecohydrological variables within the Budyko framework, providing a quantitative basis for ecosystem restoration.

Objective

• To understand the effects of sandy-region hydrological, soil, and vegetation elements on watershed ecohydrological processes for sustaining fragile dryland ecosystems, specifically by identifying critical thresholds of key ecohydrological variables through an interpretable machine learning approach within the Budyko framework.

Study Configuration

• Spatial Scale: A closed-flow basin in the Horqin Sandy Land.
• Temporal Scale: Long-term data from 1985 to 2022.

Methodology and Data

• Models used: FAO-56 Penman-Monteith equation, abcd hydrological model, XGBoost-SHAP (interpretable machine learning).
• Data sources: Long-term observational data from the study basin.

Main Results

• The study region is a typical water-limited system, with the Budyko parameter ω predominantly ranging between 1.5 and 2.
• Meteorological factors exerted significant negative effects on the Budyko parameter ω, while hydrological, vegetation, and soil variables had positive influences.
• All factors exhibited a consistent S-shaped response, approximately following a "three-stage" pattern (steady state I–transition–steady state II), indicating nonlinear response mechanisms.
• Critical thresholds were quantified for key ecohydrological variables: wet-day frequency (57–85 days per year), groundwater depth (1.7–2.8 meters), shrub dominance (7.5 %–11 %), and Soil Ratio Index (SRI) (1–1.7).

Contributions

• Quantifies critical thresholds for ecohydrological variables in sandy ecosystems using a data-driven, interpretable machine learning approach (XGBoost-SHAP) within the Budyko framework.
• Provides a quantitative basis and scientific support for targeted restoration and management of fragile sandy ecosystems.
• Reveals the nonlinear, S-shaped response mechanisms of sandy ecohydrological systems to environmental disturbances, enhancing understanding of their complex dynamics.

Funding

• Not specified in the provided text.

Citation

@article{Zhang2025Unraveling,
  author = {Zhang, Yiran and Liu, Tingxi and Shamseldin, Asaad Y. and Tong, Xin T. and Duan, Limin and Jia, Tianyu and Bao, Yongzhi and Lun, Shuo},
  title = {Unraveling threshold effects and feature contributions in hydrological-soil-vegetation response to Budyko parameters: A data-driven perspective},
  journal = {Journal of Hydrology},
  year = {2025},
  doi = {10.1016/j.jhydrol.2025.134612},
  url = {https://doi.org/10.1016/j.jhydrol.2025.134612}
}