基于EEMD的盐城大丰区降水量时间序列多尺度分析
Multiscale Analysis of the Precipitation Time Series in Dafeng District of Yancheng City Based on EEMD
陈超 1李鹏飞 2陈尚云 1陈凯 1陈则煌2
作者信息
- 1. 江苏省盐城市大丰区气象局,江苏盐城224100
- 2. 中国气象局气溶胶与云降水重点开放实验室,江苏南京210044;南京信息工程大学大气物理学院,江苏南京210044
- 折叠
摘要
[目的]研究盐城大丰地区年降水量时间序列规律.[方法]基于集合经验模态分解(EEMD)的数据处理方法,通过Hilbert变换对降水量数据进行多尺度特征研究,并对研究结果进行显著性和误差检验.[结果]近55 a盐城大丰地区年降水量序列可以分解成4个IMF分量和1个趋势分量;主要存在0～0.05和0.45 ～0.50 Hz2个频率区间,其间主周期为2.12a年际周期和27.50 a年代际周期,二者均通过95％显著性检验.EEMD可以有效地重构原始降水量数据变化曲线,与原始数据对比误差百分比始终控制在±0.6％范围内,由白噪声带来的干扰误差几乎可以完全忽略.[结论]该研究可为合理开发与充分利用水文资源、防汛抗旱等提供科学依据.
Abstract
[Objective] To research the precipitation time series in Dafeng District of Yaucheng City.[Method] Based on the data treatment method of ensemble empirical mode decomposition (EEMD),multi-scale characteristics research of precipitation data were researched by Hilbert transform.Significance and error tests on researchresults were carried out.[Result] The sequence of annual precipitation of Dafeng District in recent 55 years could be broken down into four IMF (Intrinsic Mode Function) components and a trend component.There were mainly two frequency regions,which were 0-0.05 Hz and 0.45-0.50 Hz.The annual cycle was 2.12 years and the decadal cycle was 27.50years during the main period.Both of them passed the 95％ significance test.EEMD precipitation could effectively reconstruct the curve of original data.Compared with the original data,error percentage was always controlled within ± 0.6％.Then the interference error by the white noise could be ahnost completely ignored.[Conclusion] This research provides scientific basis for the rational development and fully utilization of hydrologic resources,flood control and drought relief.
关键词
年降水量/时间序列/EEMD/Hilbert变换/显著性检验/盐城大丰地区Key words
Annual precipitation/Time series/EEMD/Hilbert transform/Significance test/Dafeng District in Yancheng City引用本文复制引用
基金项目
国家自然科学基金(41175003)
江苏高校优势学科建设工程资助项目()
出版年
2016
国家科技期刊平台
NSTL
维普
万方数据