Chaotic Prediction Model Of Ground-level Ozone Time Series In Shanghai

Ozone is one of the typical secondary pollutants formed in the ground-level atmosphere through photochemical reactions involving nitrogen oxides (NO_X) and volatile organic compounds (VOCs) in the presence of sunlight. It is a strong oxidant and can do harm to the society as well as human health. Many mega-cities especially those industrial ones have suffered from the effects of ozone pollution. In recent years, with the development of industry and the growth of vehicle number, there is a quick increase in NO_X and hydrocarbons emissions by human being. Consequently, ground-level ozone pollution becomes increasingly serious. Therefore, to carry out the research on ozone concentration prediction is of vital importance for public awareness to reduce the loss caused by ozone pollution.The hourly ozone concentration time series of Dianshanhu, Putuo and Nanhui monitoring stations are provided by Shanghai Environmental Monitoring Centre. By analyzing the ozone concentration time series with different methods, the chaotic characteristics were found. Based on that, an ozone concentration time series chaotic prediction model was constructed, and the prediction results were acceptable. The main points can be generalized as follows:Firstly, on the basis of phase space reconstruction theory, the largest Lyapunov exponent of each station was calculated using the Rosenstein small data sets method. The result that all the largest Lyapunov exponents were positive indicated that the ozone concentration fluctuations were chaotic. Besides, multi-time-scale features of the ozone concentration time series were analyzed with wavelet transform method. The isolines obviously showed that the ozone concentration fluctuation periods varied in different time scales, and self-similar structures and bifurcation structures were found in different time scales. The characteristics differed from those of stochastic series, which proved the chaotic features of ozone fluctuation again. And that was the basis of the ozone chaotic prediction model.Secondly, to better understand the complexity of the ozone concentration time series, Lemper-Ziv complexities were calculated for each ozone time series. The results showed that the complexities of ozone time series in Putuo were the largest, the complexities of ozone in Nanhui were the smallest. Besides, complexities of different seasons were also different. Complexities in spring and summer were the largest, those in winter were the smallest.Thirdly, the methods of chaotic time series prediction were systematicallysummarized. Since the ground-level ozone concentration time series fluctuations had chaotic characteristics, the prediction model was constructed based on the chaotic time series prediction method. This model adopted direct prediction method, so error accumulation effect was avoided. The training data set was real-time updated so that the model was always the newest. Additionally, the input parameters were very simple, so it was easy for application.Finally, prediction experiments were conducted with the hourly ozone concentration observations from Putuo, Nanhui and Dianshanhu sites in order to test its prediction performance. It turned out that 1-8 hour ahead predictions were satisfying. The 24 hour ahead predictions were not as good as 1-8 hour ahead predictions, but the prediction results were also acceptable.