Li et al. (2026) Ecological flow guarantee rate along the Xijiang River mainstream at different scales based on multiple probability distributions

Identification

• Journal: Scientific Reports
• Year: 2026
• Date: 2026-03-10
• Authors: J. W. Li, Xinyi Deng, Ju Liu, Zhanhong Cheng, Shilang Deng, Xiaoli Song
• DOI: 10.1038/s41598-026-43793-1

Research Groups

• School of Water Conservancy and Hydropower Engineering, North China Electric Power University, Beijing, China
• Guizhou Wujiang Hydropower Development Limited Liability Company, Guiyang, China
• School of Foreign Languages, North China Electric Power University, Beijing, China

Short Summary

This study quantifies ecological flow guarantee rates along the Xijiang River mainstream by reconstructing quasi-natural runoff using a random forest model and applying a probabilistic framework with multiple distribution functions. It finds that ecological flow guarantee rates decreased during the change period, particularly in the upper reaches and during the flood season (July to October), identifying these as priority areas and sensitive periods for ecological flow management.

Objective

• To quantify the ecological flow guarantee rate at multiple temporal and spatial scales along the Xijiang River mainstream.
• To address runoff nonstationarity by reconstructing quasi-natural runoff using a random forest model trained on natural period data.
• To select the optimal monthly probability distribution for ecological flow estimation using multiple goodness-of-fit tests and explicit decision criteria.
• To evaluate the temporal evolution and longitudinal spatial variability of ecological flow guarantee rates and identify ecologically sensitive periods.

Study Configuration

• Spatial Scale: Xijiang River mainstream, Pearl River Basin, China. Four representative hydrological stations: Tian'e, Qianjiang, Wuxuan, and Wuzhou.
• Temporal Scale:
  • Daily runoff records: Tian'e and Wuzhou (1961–2020), Qianjiang and Wuxuan (1961–2019).
  • Meteorological data: 1961–2020.
  • Natural periods: 1961–1986 (Tian'e, Qianjiang), 1961–1992 (Wuxuan), 1961–1983 (Wuzhou).
  • Change periods: 1987–2020 (Tian'e), 1987–2019 (Qianjiang), 1993–2019 (Wuxuan), 1984–2020 (Wuzhou).
  • Monthly scale for ecological flow estimation and guarantee rate analysis.

Methodology and Data

• Models used:
  • Random Forest model (for quasi-natural runoff reconstruction).
  • Six change point detection methods: cumulative anomaly, sliding t-test, Pettitt test, hierarchical clustering, Mann-Kendall test, and moving rank sum test.
  • Five probability distributions: Normal, Gamma, Generalized Extreme Value (GEV), P-III, and Logistic (for monthly runoff fitting).
  • Maximum Likelihood Estimation (MLE) for distribution parameter estimation.
  • Three goodness-of-fit tests: Kolmogorov-Smirnov (K-S) test, Anderson-Darling (A-D) test, and Chi-Square (C-S) test.
  • Month-by-month frequency method (to calculate ecological flow at 90% guarantee rate).
  • Tennant method (to evaluate the rationality of derived ecological flows).
  • Hierarchical clustering (to identify sensitive periods).
• Data sources:
  • Hydrological observations: Daily runoff records from four hydrological stations (Tian'e, Qianjiang, Wuxuan, Wuzhou).
  • Meteorological data: Daily precipitation, evaporation, temperature, wind speed, and sunshine duration from 13 meteorological stations (China Meteorological Network).
  • Geographic data: Provincial boundaries (Natural Earth), Pearl River Basin boundary and river network (HydroSHEDS products), hydrological station coordinates (published hydrological yearbooks).

Main Results

• Runoff change points were identified at Tian'e (1986), Qianjiang (1986), Wuxuan (1992), and Wuzhou (1983), delineating natural and change periods.
• The random forest model effectively reconstructed quasi-natural runoff, achieving Nash-Sutcliffe Efficiency (NSE) and coefficient of determination (R²) values above 0.9 during calibration and above 0.6 during verification.
• The Generalized Extreme Value (GEV) distribution (54.17%) and P-III distribution (20.83%) were most frequently identified as the optimal distributions for monthly runoff across the four stations.
• Monthly ecological flows, calculated at a 90% guarantee rate, were consistently lower than monthly average flows in both natural and change periods, with peak ecological water needs occurring during the high-flow season (June to August).
• Tennant method evaluation classified ecological flows at Tian'e and Qianjiang as "Excellent" during normal water use and "Optimum range" during spawning/rearing periods. Wuxuan and Wuzhou were rated "Good" during normal water use and "Optimum range" during spawning/rearing periods, indicating reasonable and reliable ecological flow estimates.
• Ecological flow guarantee rates showed a fluctuating decline across all stations in the early 1990s, followed by a gradual recovery around 1995.
• During the change period, annual ecological flow guarantee rates decreased at Tian'e, Qianjiang, and Wuzhou compared to the natural period, with more pronounced declines at Tian'e and Qianjiang (upper reaches). All stations maintained guarantee rates above 83%.
• The most significant reductions in ecological flow guarantee rates during the change period were concentrated from July to October, which was identified as the ecologically sensitive period.

Contributions

• Developed an integrated framework that combines change point detection, random forest-based quasi-natural runoff reconstruction, and a multi-test decision rule for selecting optimal month-specific probability distributions.
• Addressed runoff nonstationarity in ecological flow assessment by reconstructing quasi-natural runoff, providing a more consistent baseline for analysis.
• Quantified the long-term temporal evolution and longitudinal spatial variability of ecological flow guarantee rates along the Xijiang River mainstream.
• Identified priority river reaches (upper reaches: Tian'e and Qianjiang) and ecologically sensitive months (July to October) for targeted ecological flow protection and adaptive reservoir operation.
• Provided a transferable framework for assessing ecological flows in regulated river systems to support sustainable freshwater ecosystem protection and water resources management.

Funding

• National Natural Science Foundation of China (U2243224, 52579009)
• Special Foundation of the National Key Laboratory (2024nkzd01, 2025SDG03)

Citation

@article{Li2026Ecological,
  author = {Li, J. W. and Deng, Xinyi and Liu, Ju and Cheng, Zhanhong and Deng, Shilang and Song, Xiaoli},
  title = {Ecological flow guarantee rate along the Xijiang River mainstream at different scales based on multiple probability distributions},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-026-43793-1},
  url = {https://doi.org/10.1038/s41598-026-43793-1}
}