Lu et al. (2026) Future water erosion on the Tibetan Plateau: Projections from coupled model intercomparison project phase 6 (CMIP6)

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2026
• Date: 2026-03-10
• Authors: Shaojuan Lu, Xixi Lu, Hui Shi, Tongtong Wang, Shengzhao Wei, Enwei Zhang, Zhongping Lai
• DOI: 10.1016/j.ejrh.2026.103331

Research Groups

• Institute for Interdisciplinary Innovation Research, Xi’an University of Architecture and Technology, Xi’an, Shaanxi, China
• Department of Geography, National University of Singapore, Singapore
• Institute of International Rivers and Eco-security, Yunnan University, Kunming, Yunnan, China
• Institute of Marine Sciences, Guangdong Provincial Key Laboratory of Marine Disaster Prediction and Protection, Shantou University, Shantou, Guangdong, China

Short Summary

This study projects future water erosion on the Tibetan Plateau using CMIP6 models, revealing that 'hot' models (high climate sensitivity) in unconstrained ensembles predict higher mean annual soil erosion rates (up to 30.37 t⋅ha⁻¹⋅a⁻¹) compared to constrained ensembles, highlighting the critical role of model selection and the need for adaptive soil conservation strategies.

Objective

• To evaluate the predictive performance of multiple CMIP6 climate models for future water erosion on the Tibetan Plateau.
• To examine the temporal change trends of water erosion under contrasting Shared Socioeconomic Pathway (SSP) scenarios.

Study Configuration

• Spatial Scale: The Tibetan Plateau, covering approximately 2.79 × 10⁶ km², with a mean elevation of 4500 m. Soil erosion was assessed at a 1 km spatial resolution.
• Temporal Scale: Historical data from 2001–2020. Future projections for three periods: near-term (2041–2060), medium-term (2061–2080), and long-term (2081–2100).

Methodology and Data

• Models used:
  • Climate Models: Ten single climate models from CMIP6 (ACCESS-CM2, CMCC-ESM2, EC-Earth3-Veg, GISS-E2-1-G, INM-CM5-0, IPSL-CM6A-LR, MIROC6, MPI-ESM1-2-HR, MRI-ESM2-0, UKESM1-0-LL). Two ensemble average models: 'constrained' (excluding models with equilibrium climate sensitivity (ECS) > 4 °C) and 'unconstrained' (incorporating all ten models).
  • Soil Erosion Model: Chinese Soil Loss Equation (CSLE).
  • Vegetation Prediction Model: Random Forest model.
  • Land Use Simulation: Future Land-Use Simulation System (FLUS).
• Data sources:
  • Climatic data: Historical (2001–2020) from the National Tibetan Plateau Data Center (1 km resolution); Future projections from selected CMIP6 models (GeoTIFF format).
  • Soil data: Basic soil property dataset for China (2010–2018, 1 km resolution, 0–5 cm depth).
  • Topographic data: Digital Elevation Model (DEM) from NASA (original 30 m resolution).
  • Fractional Vegetation Cover (FVC) data: China regional 250 m FVC dataset (2001–2020), resampled to 1 km.
  • Land use data: Global simulation dataset of land-use and land-cover change (1 km resolution) under SSP1-2.6 and SSP5-8.5 scenarios.

Main Results

• Projected mean annual soil erosion rates (MASER) across the Tibetan Plateau were consistently below 30.37 t⋅ha⁻¹⋅a⁻¹ across all models.
• Single CMIP6 models showed considerable variability in MASER projections, with biases ranging from 7.48% to 29.74% (mean 13.95% for SSP1-2.6, 17.50% for SSP5-8.5) relative to a reference model.
• Unconstrained ensemble average models (including 'hot' models with ECS > 4 °C) consistently projected higher MASER values (20.78–30.37 t⋅ha⁻¹⋅a⁻¹) than constrained ensemble models (20.54–28.90 t⋅ha⁻¹⋅a⁻¹), with a bias below 5% between the two ensemble types.
• Under the sustainability-focused SSP1-2.6 scenario, MASER values from ensemble models showed a marginal decline over time.
• Under the fossil-fuel-intensive SSP5-8.5 scenario, MASER values from ensemble models projected a moderate increase over time.
• Spatially, MASER consistently decreased from the southeastern to the northwestern Tibetan Plateau. High-risk erosion areas (high-high clusters) were predominantly concentrated along the middle and lower reaches of the Brahmaputra, the upper Salween, and the headwaters of the Yellow, Yangtze, and Mekong Rivers, exhibiting a primarily east-west orientation.
• Under SSP1-2.6, high-high cluster ellipses expanded over time, suggesting a broader dispersion of high-risk erosion areas. Conversely, under SSP5-8.5, these ellipses contracted, indicating a localized intensification of severe erosion.
• Variance partitioning analysis revealed that the vegetation cover and biological practice factor (B factor) had a greater influence on water erosion than the rainfall erosivity factor (R factor) across both SSP scenarios and all time periods, although the R factor's influence became progressively more prominent under SSP5-8.5.

Contributions

• Provided a systematic evaluation of the predictive performance of multiple CMIP6 climate models, including the implications of 'hot' models, for future water erosion on the Tibetan Plateau.
• Examined the temporal change trends and spatial shifts of water erosion under contrasting Shared Socioeconomic Pathway (SSP1-2.6 and SSP5-8.5) scenarios.
• Advocated for the adoption of ensemble average models, particularly unconstrained ensembles for risk-averse planning, to achieve more robust projections of future water erosion compared to single-model simulations.
• Highlighted the critical role of climate model selection in forecasting water erosion and underscored the urgent need for adaptive soil conservation strategies tailored to anticipated future climate trajectories.

Funding

• National Natural Science Foundation of China (No. 42107364)
• Ministry of Education, Singapore (No. A-8001249-00-00)

Citation

@article{Lu2026Future,
  author = {Lu, Shaojuan and Lu, Xixi and Shi, Hui and Wang, Tongtong and Wei, Shengzhao and Zhang, Enwei and Lai, Zhongping},
  title = {Future water erosion on the Tibetan Plateau: Projections from coupled model intercomparison project phase 6 (CMIP6)},
  journal = {Journal of Hydrology Regional Studies},
  year = {2026},
  doi = {10.1016/j.ejrh.2026.103331},
  url = {https://doi.org/10.1016/j.ejrh.2026.103331}
}