Gu et al. (2026) Nonstationary Runoff Evolution and Structural Regime Shifts in Cold-Region Plateau Rivers Under Climate Change

Identification

• Journal: Water
• Year: 2026
• Date: 2026-03-30
• Authors: Kaiye Gu, YANHUI AO, Yong Li
• DOI: 10.3390/w18070816

Research Groups

Key Laboratory of Integrated Regulation and Resource Development on Shallow Lakes, Ministry of Education, College of Environment, Hohai University, Nanjing 210098, China

Short Summary

This study investigates the nonstationary evolution of streamflow and extreme hydrological risks in the Yalong and Dadu River basins (upper Yangtze River) under future climate change (2017–2100) using a SWAT model and advanced time-frequency and spatial analyses. It reveals that precipitation is the primary streamflow driver, but temperature regulates seasonal intensity, with significant hydrological regime shifts projected for the mid-21st century and intensified, clustered extreme events, highlighting an upstream buffering-downstream sensitivity pattern.

Objective

• To apply multivariate wavelet coherence analysis to unravel the multiscale time-frequency relationships between streamflow and climatic drivers (precipitation and air temperature), identifying dominant driving mechanisms and lead-lag relationships.
• To employ spatial wavelet analysis to quantify the spatial heterogeneity of streamflow seasonality and elucidate the regulatory effects of different underlying surfaces along hydroclimatic gradients.
• To detect nonstationary change points and assess the evolution of extreme hydrological events (floods and droughts) under different levels of warming.

Study Configuration

• Spatial Scale: Yalong River basin (approximately 130,000 km²) and Dadu River basin (mainstem length 1062 km) in the upper Yangtze River, China, analyzed at the sub-basin level.
• Temporal Scale: Future projections for 2017–2100 under four Shared Socioeconomic Pathways (SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5). Model calibration and validation used historical data from 2006–2019.

Methodology and Data

• Models used:
  • Hydrological Model: Soil and Water Assessment Tool (SWAT)
  • Climate Models: Multi-model ensemble mean of CMIP6 Global Climate Models (EC-Earth3 and MRI-ESM2-0)
  • Analytical Methods: Multivariate Wavelet Coherence analysis (MWTC), Spatial Wavelet Transform, Moving-window Z-test, Pettitt test, Buishand test, DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm.
  • Bias Correction: Quantile Delta Mapping method for GCM outputs.
• Data sources:
  • Topography: Digital Elevation Model (DEM) (30 m × 30 m, 2009, Geospatial Data Cloud)
  • Land Use/Land Cover: GlobeLand30 (30 m × 30 m, 2020)
  • Soil Properties: World Soil Database (1 km × 1 km)
  • Meteorological Data: Daily precipitation, maximum/minimum temperature (1999–2021, Hydrological Yearbook, China Meteorological Administration)
  • Hydrological Observations: Daily discharge data from 8 hydrological stations (2006–2019, Hydrological Yearbook)
  • River Network: Vector data (0.1 m, 91 Weitu)

Main Results

• Annual average streamflow is projected to increase across all SSP scenarios, with the Yalong River basin showing larger changes (0.13% to 0.88% annual increment) than the Dadu River basin (0.24% to 0.58%). By 2081–2100, streamflow could increase by 13.45–50.11% in Yalong and 3.81–30.91% in Dadu relative to 1991–2014.
• Streamflow exhibits strong and rapid dependence on precipitation (Average Wavelet Coherence ≈ 0.90, Percentage of Area Significant Coherence > 85%, lag < 1 month), indicating a precipitation-limited regime. Temperature primarily modulates seasonal streamflow amplitude via snowmelt and evapotranspiration, with weaker coherence (PASC ≈ 30%).
• A robust northwest-southeast gradient in seasonal streamflow intensity is observed: upstream source regions (northwest) show low variability (sensitivity index < 5.2) due to permafrost and snow buffering, while downstream forested canyon regions (southeast) exhibit high variability (sensitivity index > 5.6) due to concentrated monsoon rainfall, rapid lateral flow, and enhanced evapotranspiration. Forest cover is positively correlated with seasonal streamflow intensity (R = 0.80–0.86), while grassland shows a negative correlation (R = 0.75–0.83).
• Significant hydrological regime shifts are projected for the mid-21st century (mainly 2040s–2050s), with earlier shifts under higher emission scenarios.
• Extreme hydrological events are expected to become more frequent (24–42 events per scenario) and clustered. Low-flow events have longer average durations (1.6–2.1 months) than high-flow events (1.2–1.7 months) and can reach intensities as low as 0.17–0.20 times the mean discharge. Under SSP5-8.5, extreme high-flow event intensity can double (up to 2.70 times the mean discharge), and prolonged low-flow periods are projected (e.g., mean duration of 18.0 months in Yalong).

Contributions

• Provides a comprehensive, multidimensional analytical framework integrating time-frequency coherence, spatial sensitivity patterns, and extreme event evolution for complex mountainous basins.
• Unravels the nonlinear and spatially heterogeneous evolution of streamflow across multiple time-frequency scales, which was previously insufficiently understood.
• Offers a process-based understanding of how complex mountainous basins amplify or buffer climatic signals across space and time.
• Identifies a critical "upstream buffering-downstream sensitivity" pattern, emphasizing the need for spatially differentiated water resources management strategies under nonstationary climate conditions.

Funding

National Key Research and Development Program of China, grant number 2022YFC3202402.

Citation

@article{Gu2026Nonstationary,
  author = {Gu, Kaiye and AO, YANHUI and Li, Yong},
  title = {Nonstationary Runoff Evolution and Structural Regime Shifts in Cold-Region Plateau Rivers Under Climate Change},
  journal = {Water},
  year = {2026},
  doi = {10.3390/w18070816},
  url = {https://doi.org/10.3390/w18070816}
}