成都冬季PM2.5演化的EEMD分解及自组织临界态

Multiscale EEMD Decomposition of PM2.5 in Chengdu during Winter and Self-organized Criticality of Air Pollution

为探讨冬季期间大气PM2.5演化的多时间尺度效应,并阐释重度灰霾发生及演化的动力机制,该研究应用集合经验模态分解(EEMD)方法与自组织临界(SOC)理论,对成都市20171201-20180228冬季期间4个国控监测站点(即大石西路、金泉两河、君平街、三瓦窑) PM2.5浓度时空演化规律进行实证研究。通过EEMD分解,获得了不同时间尺度上具有良好平稳性特征的固有模态函数(IMF)。大气PM2.5演化的主要模态存在准4 h、准8 h和准24 h的平均周期,这些典型周期对应模态的累积贡献率基本上达到95%以上。研究表明,主要模态的准周期变化与各类型生产活动、交通排放紧密关联,这反映了大气系统中人为污染源的周期性输入作用。同时,研究发现,PM2.5小时平均质量浓度波动函数服从幂律分布结构,具有标度不变性特征。进一步基于SOC理论探讨了大气PM2.5浓度时空演化的内在动力规律,结合典型区域气象特征,阐明了冬季期间严重大气污染产生的宏观涌现机制。结果表明,EEMD方法所获得的不同IMF分量可以揭示大气PM2.5时空演化的多尺度特征,但不同时间尺度上的IMF分量之间互不独立,各IMF分量的形成既受到准周期大气污染排放的作用,也受到大气系统非线性SOC动力机制的控制。

To explore the multiple time scale effects of PM2.5 evolution during winter, and to explain the dynamic mechanism of a typical haze episode occurrence and evolution, the cumulative distributions statistics and temporal scaling properties of PM2.5 concentration during winter(from 1 December, 2017 to 28 February, 2018) at four state-controlled monitoring stations of Chengdu, southwestern China, are explored by using ensemple empirical mode decompositiom(EEMD) and self-organized criticality theory(SOC) method. The good stability characteristics of the intrinsic mode function(IMF) on different time scales are obtained by using the EEMD decomposition method. It shows that the main modes of the PM2.5 evolution exist quasi4 h, 8 h, 24 h on average cycles, and the cumulative variance contribution of these typical cycles to the corresponding modes is basically over 95%. This study shows that quasiperiodic changes in major modes are closely related to various types of production activities and traffic emissions, it reflects periodicity input of anthropogenic sources of pollution in the atmospheric system. At this moment, the probability distribution functions of PM2.5 hourly average concentrations are characterized by power-law distribution, and the temporal evolution of PM2.5 exhibit scale invariant behavior. It has further discussion of the inherent dynamic mechanisms of PM2.5 evolution based on the SOC theory, the macroscopic emergence mechanism of severe air pollution is interpreted by using typical regional meteorological characteristics during winter. The results show that the different IMF components obtained by the EEMD method can reveal the multiple time scales characteristics of PM2.5 evolution, but the IMF components on different time scales are not independent, the formation of each IMF component is subject to emissions of quasiperiodic air pollution, and is also controlled by the nonlinear SOC dynamic mechanism of the atmospheric system.