Nonlinear time series analysis of ground-level ozone dynamics in Southern Taiwan

The ozone pollution at ground level in rural and urban areas has been a long-standing problem in the world. This paper focuses on estimating self-affined nature of nonlinearity of ground-level peak ozone time series, which is analyzed by two nonlinear fractal statistical methods, including R/S analysis and BDS test. To explore the underlying structure of ozone observations at ground level and improve the forecasting capacity in urban region, practical implementation was assessed by a case study via collecting and analyzing the monitoring data at Chaojhou and Zenwu in the Kaohsiung metropolitan region, Taiwan. Based on R/S analysis, the time series can be identified as persistent and long-memory processes with Hurst exponents of both about 0.75. In addition, the V statistics specifies possible fluctuation cycle lengths of 32, 170, and 420 day simultaneously. Such results are consistent with the regional meteorological conditions leading to help characterize the regional scale ozone behavioral trend. Furthermore, the BDS test results confirm a strong nonlinearity of both time series associated with these two cities. Yet in both cases, nonlinearity implies chaos. The R/S analysis and BDS test provide strong evidence for nonlinearity and fractality of ozone time series due to noisy chaos, and we could not rule out the possibility of deterministic chaos in tropospheric ozone system.