Li et al. (2026) Grey Wolf optimization enhanced adaptive decomposition for trend periodic analysis of nonstationary and nonlinear hyrologic series

Identification

• Journal: Scientific Reports
• Year: 2026
• Date: 2026-01-07
• Authors: Jinbei Li, Wei Ding, Hao Wang
• DOI: 10.1038/s41598-026-35076-6

Research Groups

• School of Infrastructure Engineering, Dalian University of Technology, Dalian 116024, China
• Department of Water Resources Research, China Institute of Water Resources and Hydro-power Research, Beijing 100038, China

Short Summary

This study proposes and validates a Grey Wolf Optimization-enhanced adaptive decomposition method (GITPA) for integrated trend-periodic analysis of nonstationary and nonlinear hydrological series. The method demonstrates superior accuracy and robustness compared to traditional approaches, revealing complex hydro-meteorological trends and multi-timescale periodicities in the Yangtze River Basin and enabling more accurate runoff extreme forecasting.

Objective

• To develop and validate a novel Grey Wolf Optimization-enhanced Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (GITPA) method for robust trend-periodic analysis of non-stationary and nonlinear hydrological series, overcoming limitations of traditional methods.
• To apply the GITPA method to the Yangtze River Basin to systematically reveal the dynamic characteristics of hydro-meteorological variables and provide a reliable basis for predicting future extreme events.

Study Configuration

• Spatial Scale: Yangtze River Basin, China, including its mainstem and nine major tributaries (Jinsha River, Yalong River, Min River, Jialing River, Wujiang River, Dongting Lake, Han River, Poyang Lake).
• Temporal Scale: 44-year time series dataset from 1979 to 2022, with forecasts extending to 2030.

Methodology and Data

• Models used:
  • Proposed: Grey Wolf Optimization-enhanced Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (GITPA), comprising GWO-ICEEMDAN decomposition, GWO-ICEEMDAN trend analysis (GITA), and GWO-ICEEMDAN periodic analysis (GIPA).
  • Comparative: Mann-Kendall (MK) method, Innovative Trend Analysis (ITA), Continuous Wavelet Transform (CWT), Wavelet Packet Decomposition (WPD).
  • Techniques: t-test, block bootstrap method, Fast Fourier Transform (FFT), Thiessen polygon method, Spearman correlation coefficient.
• Data sources:
  • Synthetic datasets (for method validation).
  • Natural runoff data: Changjiang Water Resources Commission (from nine control hydrological stations in the Yangtze River Basin).
  • ENSO and IOBW data: National Climate Center of the China Meteorological Administration.
  • Sunspot number data: World Data Center for the Sunspot Index and Long-term Solar Observations (WDC-SILSO).
  • Meteorological data: China Meteorological Science Data Sharing Service Network (annual precipitation and mean annual temperature from 147 stations within the basin).

Main Results

• Method Validation:
  • GITPA achieved trend detection accuracy consistently exceeding 85%, outperforming ITPA (above 80%), MK (over 70%), and ITA (55-80% depending on PDF).
  • GITPA demonstrated high robustness to different probability density functions (PDFs) and autocorrelation structures, unlike ITA (PDF-sensitive) and MK (autocorrelation-sensitive).
  • For periodic analysis, GITPA accurately identified all periodic components in synthetic sequences, while ITPA showed limitations in shorter sequences and CWT was effective only for periods less than 100.
• Application in Yangtze River Basin (1979-2022):
  • Temperature: All sub-basins exhibited a consistent upward trend in annual mean temperature, with a warming rate of approximately 0.03 °C per year.
  • Precipitation: Spatially heterogeneous changes, with increases in the northwest and east (e.g., Yalong and Min rivers) and decreases in central and southwestern regions (e.g., Jialing, Han rivers, Dongting Lake basin).
  • Runoff: Overall declining trends in the mainstem and most tributaries, with the Dongting Lake basin experiencing the most significant reduction (22,730 cubic metres per year).
  • Periodicities: Identified short-term (2–3 years, linked to ENSO and IOBM), medium-term (approximately 11 years, linked to Schwabe solar cycle), and long-term (22–44 years, linked to Hale solar cycle) periodic characteristics in runoff.
• Runoff Extreme Prediction:
  • GITPA achieved a prediction accuracy of 79.31% for wet and dry runoff conditions, significantly higher than the Wavelet Packet Decomposition (WPD) method (58.64%).
  • GITPA's performance metrics were superior: RMSE of 755.60 x 10^6 cubic metres, MAE of 582.45 x 10^6 cubic metres, and R² of 0.52, compared to WPD's RMSE of 1169.28 x 10^6 cubic metres, MAE of 1129.38 x 10^6 cubic metres, and R² of 0.33.
  • Forecasts indicate a high probability of low-flow events in the Yangtze River Basin between 2026 and 2028, suggesting a sustained high frequency or intensity of severe droughts over the coming decade.

Contributions

• Proposes a novel and robust Grey Wolf Optimization-enhanced adaptive decomposition method (GITPA) for integrated trend-periodic analysis of non-stationary and nonlinear hydrological series.
• Demonstrates superior accuracy and adaptability of GITPA in trend detection (over 85% accuracy) and periodic identification compared to traditional methods (MK, ITA, CWT, WPD), especially for complex sequences with heterogeneous PDFs and autocorrelation.
• Provides a comprehensive analysis of hydro-meteorological variability in the Yangtze River Basin, revealing specific warming trends, spatially heterogeneous precipitation changes, declining runoff, and multi-timescale periodicities linked to climate oscillations and solar activity.
• Offers a highly accurate (79.31%) and reliable method for forecasting runoff extreme events, with practical implications for water resource management and drought mitigation strategies.

Funding

• National Natural Science Foundation of China (no. U2240204)

Citation

@article{Li2026Grey,
  author = {Li, Jinbei and Ding, Wei and Wang, Hao},
  title = {Grey Wolf optimization enhanced adaptive decomposition for trend periodic analysis of nonstationary and nonlinear hyrologic series},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-026-35076-6},
  url = {https://doi.org/10.1038/s41598-026-35076-6}
}