基于ICEEMDAN-MFDFA的广东日降水序列的多重分形特征  

ICEEMDAN-MFDFA-based multifractal characteristics of daily precipitation series in Guangdong

作　　者：余锐 孙丽颖 王敏 张兰 侯灵 YU Rui;SUN Liying;WANG Min;ZHANG Lan;HOU Ling(Sanshui Meteorological Service,Foshan 528100,Guangdong,China;Zengcheng Meteorological Service,Guangzhou511300,Guangdong,China;Shaoguan Meteorological Service,Shaoguan 512026,Guangdong,China;Guangzhou Meteorological Observatory,Guangzhou 511430,Guangdong,China;Guangdong Meteorological Data Center,Guangzhou 510640,Guangdong,China)

机构地区：[1]佛山市三水区气象局,广东佛山528100 [2]广州市增城区气象局,广东广州511300 [3]韶关市气象局,广东韶关512026 [4]广州市气象台,广东广州511430 [5]广东省气象数据中心,广东广州510640

出　　处：《水利水电技术（中英文）》2023年第2期85-95,共11页Water Resources and Hydropower Engineering

基　　金：国家自然科学基金项目(U2142213);广东省气象局科学技术研究项目(GRMC2021M11,GRMC2021XQ16);广东省气象局科技创新团队(GRMCTD202103);广州市气象学会科学技术研究项目(Q202201)。

摘　　要：【目的】降水时间序列是典型的非平稳、非线性时间序列,在全球变暖的背景下呈现出明显的复杂波动特征。通过全面考察广东日降水序列的分形特征,为有效应对区域水旱灾害风险提供参考。【方法】引入改进的自适应噪声完全集合经验模态分解(ICEEMDAN)方法改善传统多重分形去趋势波动分析(MFDFA)方法在去趋势过程中的不足,利用ICEEMDAN-MFDFA方法得到1973—2021年广东日降水序列的多重分形结果。【结果】结果显示:广东日降水序列具有复杂的多重分形特征,多重分形谱表现为右偏、左勾状,Hurst指数小于0.5;各项多重分形谱参数呈现阶段性上下波动特征,谱宽、不对称指数、谱端高差和Holder指数的极值分别达到1.00、-0.72、1.01和0.12;降水序列的多重分形谱宽与谱端高差、不对称指数与Holder指数两两之间有着密切的联系。【结论】结果表明:降水序列内部演变的波动行为具有长程相关性,表现为反持续性;日降水对小幅度的局部气候波动较敏感,且有增大的趋势;当降水复杂性越强时对应着日降水增大的态势越明显,当多重分形谱的右偏程度越弱时其精细结构越丰富;与传统方法相比,ICEEMDAN-MFDFA方法可以有效提高降水序列多重分形结果的可靠性和稳定性。[Objective] The precipitation series is a typical nonstationary and nonlinear time series with obvious complicated fluctuation characteristics under the background of global warming. The fractal characteristics of daily precipitation series in Guangdong are comprehensively examined to provide a reference for floods and droughts prevention work. [Methods] The improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN) method is introduced to address the shortcomings of the traditional multifractal detrended fluctuation analysis(MFDFA), and the ICEEMDAN-MFDFA method is used to obtain the multifractal analysis results of daily precipitation series in Guangdong from 1973 to 2021. [Results] The daily precipitation series exhibits complicated multifractal characteristics in Guangdong, with the multifractal spectrum being right-hand deviated and left-hook like, and the Hurst exponent less than 0.5. The multifractal spectral width, asymmetry index, spread of singularity spectrum and Holder exponent fluctuate up and down periodically, with extreme values reaching 1.00, 0.72, 1.01 and 0.12 respectively. Close relations are there between both the multifractal spectral width and the spread of singularity spectrum as well as the asymmetry index and Holder exponent. [Conclusion] The fluctuation behavior of the internal evolution of precipitation series is long-range correlated, which is manifested as anti-persistence. The daily precipitation is sensitive to small local climate fluctuations with an increasing trend. The stronger the precipitation complexity is, the more obvious the increase of the corresponding daily precipitation is to be, while the weaker the right deviation of multifractal spectrum is, the richer its fine structure is to be. Compared with traditional methods, the ICEEMDAN-MFDFA method can effectively improve the reliability and stability of multifractal analysis results of precipitation series.

关 键 词：多重分形 日降水序列 气候波动 ICEEMDAN MFDFA 降水 气候变化 全球变暖 

分 类 号：P426[天文地球—大气科学及气象学] TV125[水利工程—水文学及水资源]