Qian et al. (2025) Comparative analysis of flood characteristic changes under Meiyu and typhoon rainfall in Zhejiang Province, China (1964–2021)

Identification

• Journal: Scientific Reports
• Year: 2025
• Date: 2025-12-11
• Authors: Jinglin Qian, Wu Yunxin, Zhu-Sheng Qian
• DOI: 10.1038/s41598-025-30736-5

Research Groups

• Zhejiang University of Water Resources and Electric Power, Hangzhou, China
• China Three Gorges University, Yichang City, Hubei Province, China

Short Summary

This study comparatively analyzed flood characteristic changes in Meiyu-dominated (XF) and typhoon-dominated (HJT) watersheds in Zhejiang Province (1964–2021), revealing stable flood trends in the XF watershed but a decoupled response of decreased frequency and amplified intensity in the HJT watershed, highlighting the importance of rainfall mechanisms under climate change.

Objective

• To comparatively analyze flood characteristic changes under Meiyu and typhoon rainfall in Zhejiang Province, China (1964–2021), focusing on the differences in extreme flood responses to climate change under different rainfall mechanisms, particularly before and after 1990.

Study Configuration

• Spatial Scale: Two watersheds in Zhejiang Province, China: XF watershed (Meiyu-dominated) and HJT watershed (typhoon-dominated), located within the geographic grid 119°E–122°E and 28°N-31°N.
• Temporal Scale: 58 years, from 1964 to 2021. The analysis was segmented into two periods: Pre-1990 and Post-1990.

Methodology and Data

• Models used:
  • Flood Event Sampling: Annual Maximum Flood (AMF) method, Peaks-Over-Threshold (POT) method.
  • Change-Point Detection: Pettitt test, Standard Normal Homogeneity Test (SNHT), Buishand Range Test.
  • Extreme Value Distributions: Pearson Type III (P-III) distribution, Generalized Pareto Distribution (GPD), Gumbel distribution, Generalized Extreme Value (GEV) distribution.
  • Uncertainty Quantification: Parametric Bootstrap method (2000 iterations).
  • Model Selection Criteria: Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC).
  • Goodness-of-Fit Tests: Anderson-Darling (AD) Test, Kolmogorov-Smirnov (KS) Test.
• Data sources:
  • Hydrological data: Hourly precipitation and flow data from 1964 to 2021, collected by the Zhejiang Provincial Hydrology Management Center. This includes data from 5 rain gauge stations and 1 streamflow station in the XF watershed, and 4 rain gauge stations and 1 streamflow station in the HJT watershed.
  • Land use data: High-precision Zenodo China Land Cover Dataset (CLCD) with a 30-meter spatial resolution, covering the period from 1985 to 2022.

Main Results

• XF Watershed (Meiyu-dominated):
  • Flood characteristics show overall stability around 1990.
  • The 100-year design flood decreased by 2.3% (AMF method, from 334.5 m³/s to 326.7 m³/s) and increased by 3.9% (POT method, from 317.5 m³/s to 329.9 m³/s).
  • The GPD shape parameter (ξ) decreased from 0.0345 (Pre-1990) to -0.0827 (Post-1990), indicating a shift in flood distribution from a fat-tail to a bounded-tail.
  • The GPD scale parameter (σ) increased by 19.1% (from 48.70 to 58.02).
  • The annual exceedance rate decreased by 9.9% (from 2.333 to 2.103 events per year).
  • Uncertainty in design flood estimates (relative confidence intervals) reduced in the Post-1990 period (e.g., Q100 AMF: 71.2% to 63.9%; Q100 POT: 47.3% to 41.4%).
• HJT Watershed (Typhoon-dominated):
  • Exhibits a decoupled response of "decreased frequency - amplified intensity".
  • The 100-year design flood increased by 64.2% (AMF method, from 1754.3 m³/s to 2880.2 m³/s) and by 7.0% (POT method, from 1948.6 m³/s to 2084.9 m³/s).
  • The annual exceedance rate decreased by 4.1% (from 1.885 to 1.808 events per year).
  • The GPD scale parameter (σ) increased significantly by 42.2% (from 208.33 to 296.23), indicating substantial intensity amplification.
  • The GPD shape parameter (ξ) increased from 0.1771 (Pre-1990) to 0.2170 (Post-1990), reflecting a strengthening of heavy-tailed characteristics.
  • High statistical instability and uncertainty were observed in the Pre-1990 period (Q100 POT relative width: 152.0%), which significantly decreased in the Post-1990 period (58.7%).
• Methodological Discrepancy: A substantial discrepancy in Q100 estimates between the AMF (+64.2%) and POT (+7.0%) methods in the HJT watershed highlights the sensitivity of block maximum models (AMF) to extreme outliers in highly variable systems, while POT provides a more robust, decomposed view of frequency and intensity changes.
• Statistical Significance: All identified changes in design floods were not statistically significant at the 95% confidence level, primarily due to sample size limitations and inherent natural variability.

Contributions

• Provides a systematic comparative analysis of flood characteristic changes under distinct Meiyu and typhoon rainfall mechanisms, addressing a gap in existing literature.
• Identifies differentiated flood response patterns to climate change (stability for Meiyu-dominated, decoupled frequency-intensity for typhoon-dominated systems), offering new insights into regional hydrological impacts.
• Demonstrates the complementary value of dual-method (AMF and POT) analysis in flood frequency studies, showing how POT can reveal intrinsic physical mechanisms (frequency and intensity changes) in highly variable systems more robustly than AMF.
• Offers practical recommendations for regional flood control planning and risk management, advocating for a conservative approach in typhoon-dominated areas despite statistical insignificance, by considering the amplified intensity signals.

Funding

• Joint Fund of Zhejiang Provincial Natural Science Foundation of China (Grant No. LZJWZ24E090003).
• "Unveiling the List and Taking Command" Project (No. JBGS2025006) of Zhejiang University of Water Resources and Electric Power.

Citation

@article{Qian2025Comparative,
  author = {Qian, Jinglin and Yunxin, Wu and Qian, Zhu-Sheng},
  title = {Comparative analysis of flood characteristic changes under Meiyu and typhoon rainfall in Zhejiang Province, China (1964–2021)},
  journal = {Scientific Reports},
  year = {2025},
  doi = {10.1038/s41598-025-30736-5},
  url = {https://doi.org/10.1038/s41598-025-30736-5}
}