The air pollution in China currently is characterized by high fine particulate matter (PM2.5) and ozone (O3) concentrations. Compared with single high pollution events, such double high pollution (DHP) events (both PM2.5 and O3 are above the National Ambient Air Quality Standards (NAAQS)) pose a greater threat to public health and environment. In 2020, the outbreak of COVID-19 provided a special time window to further understand the cross-correlation between PM2.5 and O3. Based on this background, a novel detrended cross-correlation analysis (DCCA) based on maximum time series of variable time scales (VM-DCCA) method is established in this paper to compare the cross-correlation between high PM2.5 and O3 in Beijing-Tianjin-Heibei (BTH) and Pearl River Delta (PRD). At first, the results show that PM2.5 decreased while O3 increased in most cities due to the effect of COVID-19, and the increase in O3 is more significant in PRD than in BTH. Secondly, through DCCA, the results show that the PM2.5-O3 DCCA exponents α decrease by an average of 4.40% and 2.35% in BTH and PRD respectively during COVID-19 period compared with non-COVID-19 period. Further, through VM-DCCA, the results show that the PM2.5-O3 VM-DCCA exponents α VM in PRD weaken rapidly with the increase of time scales, with decline range of about 23.53% and 22.90% during the non-COVID-19 period and COVID-19 period respectively at 28-h time scale. BTH is completely different. Without significant tendency, its α VM is always higher than that in PRD at different time scales. Finally, we explain the above results with the self-organized criticality (SOC) theory. The impact of meteorological conditions and atmospheric oxidation capacity (AOC) variation during the COVID-19 period on SOC state are further discussed. The results show that the characteristics of cross-correlation between high PM2.5 and O3 are the manifestation of the SOC theory of atmospheric system. Relevant conclusions are important for the establishment of regionally targeted PM2.5-O3 DHP coordinated control strategies.
Keywords: PM2.5-O3 DHP; Time scale; VM-DCCA; Long-term persistence; SOC
The authors Chunqiong Liu and Kai Shi contributed equally to this work.
Despite the pollutant emissions that have been witnessed significant reduction in recent years, there are still heavy fine particulate matter (PM
PM
Due to the temporal variability of meteorology and pollution emissions, the time scale dependence can be observed in the dominant mechanism of the cross-correlation between PM
It is generally believed that the differences in regional pollution emission will interfere with the assessment of the impact of meteorology on air pollution. In 2020, with the outbreak of the COVID-19, the Chinese government adopted a series of nationwide compulsory measures such as city closure, travel restriction and plant shutdown to control the spread of the epidemic. As a result, primary pollutants were greatly reduced, which led to a similar initial atmospheric condition in different regions (Bherwani et al., [
Due to the non-linearity and complexity of evolution of secondary pollutants, existing atmospheric chemistry models are hard to meet the requests of high accuracy for research data and may result in deviations in the simulation of air quality under extreme conditions (Liu et al., [
Taking Beijing-Tianjin-Heibei (BTH) and Pearl River Delta (PRD) as the study areas, this study establishes a new VM-DCCA method based on DCCA method to extract high pollutants, thus to compare time scale characteristics of the cross-correlation between high PM
Both the Beijing-Tianjin-Hebei (BTH) and the Pearl River Delta (PRD) are the major city clusters in China. The BTH (39°28′N-41°05′N, 115°20′E-117°30′E) is located in the north of North China Plain, including two municipalities (Beijing and Tianjin) and 11 prefecture level cities (Baoding, Tangshan, Langfang, Shijiazhuang, Qinhuangdao, Handan, Zhangjiakou, Chengde, Cangzhou, Xingtai and Hengshui). This region belongs to temperate continental monsoon climate, with high temperature and rainy in summer due to the impact of the ocean water vapor and cold and dry in winter due to the impact of the cold inland air (Sun et al., [
According to the National Emergency Response Plan for Public Health Emergencies, control measures were implemented throughout the country on around January 24, 2020, and all work and production activities resumed in early May. In this study, January 24 to May 31, 2020 are set as the COVID-19 period and the same period in 2019 is set as the non-COVID-19 period for comparison. Hourly observations of PM
Graph: Fig. 1The statistics of meteorological factors during the non-COVID-19 period and COVID-19 period in BTH and PRD
Compared with traditional method for correlation analysis, the advantage of DCCA related methods is that the scales characteristic of correlations can be quantified. Piao and Fu ([
Firstly, the series
Graph
where
Secondly, the above series
Thirdly, the trend signal is subtracted from the cumulative signal to obtain the residual signal, and the covariance of each residual signal is calculated as follows:
2
Graph
The covariance of the whole time series is
3
Graph
Finally, repeat the above steps for each time scale
For a specific time scale,
Due to the adverse effects of extreme pollution (Kumar et al., [
This paper applies the method proposed by Muchnik et al. ([
Graph: Fig. 2Maximum sequences of PM 2.5 and O 3 obtained from original sequence with interval length of 4 h
Self-organized criticality (SOC) theory was proposed by Bak et al. ([
The "sandpile model" is used to illustrate the physical significance of SOC theory. Individual grains are dropped and gradually form a small pile, and then simulate and calculate how many grains will slide when a grain falls. In the initial stage, the falling grains just add to the growing pile, but as the grains accumulate to a certain extent, it stops growing and reaches a critical slope in a statistically stationary state. At this moment, the result of adding grains is unpredictable. The addition of grains may cause either a small perturbation or even trigger a large avalanche. Statistical analysis suggests that the frequency of avalanches is a power-law function with the size of the pile in the long-term grain adding process.
Figure 3 shows the variations of PM
Graph: Fig. 3Variations of PM 2.5 and O 3 in each cities of BTH and PRD during the COVID-19 period compared with the non COVID-19 period
Figure 4 shows the comparison of variances above of PM
Graph: Fig. 4Averaged variations of PM 2.5 and O 3 in BTH and PRD during the COVID-19 period compared with the non-COVID-19 period
In order to explore the long-term cross-correlation between PM
Graph: Fig. 5DCCA plot of PM 2.5 and O 3 in Beijing and Guangzhou during the COVID-19 period
In order to verify whether DCCA method truly reflects the long-term persistence characteristics of the cross-correlation between non-stationary series, this paper applies the same way to analyze the random shuffled series. According to Markov process theory, if the time series
Based on the above analyses, DCCA calculation of PM
Table 1 The main statistical values of DCCA exponents
City clusters Mean value ( Standard deviation Skewness Kurtosis Shapiro–Wilk test BTH Non-COVID-19 period 0.91 0.06 −0.12 1.53 Obey normal distribution Exist significant differences COVID-19 period 0.87 0.02 0.12 1.91 Obey normal distribution PRD Non-COVID-19 period 0.85 0.02 0.74 2.51 Obey normal distribution Exist significant differences COVID-19 period 0.83 0.03 −0.68 2.24 Obey normal distribution
The Shapiro–Wilk test is performed to test the normality of
Graph: Fig. 6Comparison of DCCA exponents α of PM 2.5 and O 3 between COVID-19 period and non-COVID-19 period in each cities of BTH and PRD
The spatiotemporal distribution of PM
Graph: Fig. 7Days of high pollution in BTH and PRD during the non-COVID-19 period and COVID-19 period
Further, taking 1 h (original concentration series), 4 h, 8 h, 12 h, 16 h, 20 h, 24 h and 28 h as interval length, the cross-correlation between high PM
Graph: Fig. 8Variations of average VM-DCCA exponents αVM with interval length in BTH and PRD during the non-COVID-19 period and COVID-19 period
It can be seen from Fig. 8 that
The SOC theory is used to explain the source of long-term persistence in the evolution of complex systems from a perspective of macro integrity. In the previous studies, Shi and Liu ([
The evolution of the cross-correlation between high PM
In Fig. 7,
The greater the long-term persistence, the longer the cross-correlation between two pollutants. In this way, PM
Therefore, the long-term persistence in PM
Based on the COVID-19 outbreak, this paper establishes a novel VM-DCCA model to study the long-term persistence of the cross-correlation between the high PM
The results show that the PM
The above results showed that the spatiotemporal evolution of cross-correlation between high PM
Bingyi Bao and Kai Shi analyzed the results and wrote the main manuscript text. Youping Li prepared Figs. 1, 2, and 3, and Chunqiong Liu and Ye Wen prepared Figs. 4, 5, 6, 7, and 8. All authors reviewed the manuscript.
This work is financially funded by the National Natural Science Foundation of China (No. 52160024), Natural Science Foundation of Hunan Province, China (No. 2022JJ30475), and Innovation Foundation for Postgraduate of Jishou University, China (No. JGY2022074). We also thank the anonymous referees and the editor-in-chief.
The data used to support the findings of this research are available from the corresponding author upon reasonable request.
All authors have read, understood, and have complied as applicable with the statement on "Ethical responsibilities of Authors" as found in the Instructions for Authors and are aware that with minor exceptions, no changes can be made to authorship once the paper is submitted.
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
By Bingyi Bao; Youping Li; Chunqiong Liu; Ye Wen and Kai Shi
Reported by Author; Author; Author; Author; Author