Related research and method

After realizing the importance of ozone to the survival and development of human beings,scholars in many countries have also carried out relevant research on the forecasting method of total amount of ozone.

The meteorological data often have obvious non-linear and non-steady state characteristics.Most of the research is to establish analysis of time series forecasting model of the ozone by using the monitoring data of ozone total amount.In 1990,Robeson et al.(1990)used the stochastic time series model(ARIMA)to study the ozone solubility prediction based on the ozone monitoring data from 1978 to 1986 in the Columbia Sand River Valley.In order to explore the long-term trend of global atmospheric ozone more accurately,two more advanced trend analysis methods are proposed,Singular Spectrum Analysis(SSA)and Ensemble Empirical Mode Decomposition(EEMD).These two methods have high self-adaptive characteristics.There is also the Dynamic Harmonic Regression(DHR)method proposed by Young et al.(1999),which simulates future changes based primarily on the characteristics on frequency domain of the original time series.