Fractal Feature And Risk Evaluation Of PM_2.5 Evolution During Typical Haze Periods

Heavy air pollution in recent years is an important and popular topic in China as concerns have been raised about the health impacts, ecological destruction and the influence on sustainable socio-economic development by Fine particulate matter(PM_2.5 )pollution. PM_2.5 mass concentration provides a referential parameter describing air pollution levels, which is the primary pollutant of air pollution, especially during a serious haze pollution. In view of the severe air pollution situation, it is necessary that understanding of the complex dynamic characteristics of PM_2.5 can contribute to developing advanced techniques for air pollution forecasting.PM_2.5 pollution is a complex atmospheric phenomenon, the evolution of PM_2.5 has obvious complexity characteristics such as highly nonlinear, non-stationary and non-deterministic. In this paper, PM_2.5 pollution were considered as the result of the complex behavior of the open and dissipative atmosphere systems under anthropogenic pollution pressure. The author introduced the fractal theory to study the complex behavior of PM_2.5 evolution; analyzed the relationship between their complexity characteristics and air pollution; based on the theory of self-organized criticality, the evolution mechanism of the nonlinear characteristics of PM_2.5 were discussed; from the perspective of “the return intervals of extreme event”, we further discussed the distribution law of the return intervals of PM_2.5 , and the mathematical model of the prediction of high concentration PM_2.5 is initially established. The main works and results are shown as follows:1. The long-rang correlation characteristic of PM_2.5 evolution at the six air automatic monitoring stations of Chengdu were explored by using detrended fluctuation analysis method（DFA）, during a typical haze episode（from January 25 to February 2,2013）. The results show that the evolution of PM_2.5 exhibit stable long-rang correlation in spatial domain, scaling exponents a were all above 1.038. Statistical analysis shows that, the scaling exponents which quantitatively characterize the dynamic characteristics of PM_2.5 evolution have no significant difference in different stations, which reflect the same inherent dynamic nature of PM_2.5 evolution.2. The long-rang correlation characteristic of PM_2.5 evolution were also explored by DFA method, during another typical haze episode in Chengdu, China（from March16 to 23, 2014）. The results show that, both in haze and non-haze days, the evolution of PM_2.5 exhibit stable long-rang correlation in temporal domain, scaling exponents a were all above 0.881. Statistical analysis shows that, both in spatial and temporal scaling, the scaling exponents of PM_2.5 evolution have no significant difference. It is stable long-range correlation characteristics that generate high PM_2.5 concentrations and result in PM_2.5 evolution sustaining for such a long time.3. Box counting technique has been used to investigate the clustering properties of PM_2.5 time series. The existence of scale-invariant or self-similar behavior of PM_2.5 time series in temporal scaling can be characterized by a box-dimension（DB）, which also shows the distribution characteristics of PM_2.5 concentration on the time axis. The results also suggest that box counting technique is an efficient tool for the characterization and comparison of the spatial and temporal variation of PM_2.5 .4. There is a certain critical mass concentration in the dynamic evolution process of PM_2.5 . Long-rang correlation and scale-invariant fractal feature are only exist under the level of critical mass concentrations, if the PM_2.5 content level is over the critical concentration, there will be a disappearance of long- rang correlation property, as well as the scale-invariant structure. The above mentioned two different nonlinear physical features can be said entailing a nice dynamic coincidence, indicating that the relevant factors involving in PM_2.5 pollution might obey the same evolution rules. We believe it is related to the self-organized criticality of atmospheric pollution.5. Self-organized criticality（SOC） was believed as one of the main mechanisms inducing PM_2.5 pollution and its evolution. A comparative investigation of the dynamic features of PM_2.5 pollution during haze days and the criteria of SOC shows that PM_2.5 evolution has the basic characteristics of self-organized criticality. The power-law,long-rang correlation and fractal properties are the extrinsic manifests of self-organized criticality of PM_2.5 pollution. The scaling exponents a and box-dimension DBdepict the macro numerical properties of self-organized critical behavior of PM_2.5 pollution, and reflect the impact of atmospheric environmental characteristics on the complex properties of PM_2.5 pollution.6. An empirical exploration of the distribution of the return intervals of PM_2.5 were present and find that it obeys a stretched exponential form, and with the pronounced clustering of extreme events on the time axis. We believe that the stretched exponential distribution function can be used as an efficient risk evaluation of atmospheric environment.