PM10 (suspended particles with diameter below 10 £gm) and O3 (ozone) are the dominant air pollutants in Kao-Ping airshed, in which ozone is a secondary pollutant produced from its precursors of NOx (= NO + NO2) and HC (hydrocarbons) via complex photochemical reactions in sunlight. This study first statistically summarized the trends of ozone concentrations using box plots over recent five to six years from four and three air-quality monitoring stations in Kaohsiung City and Ping-Tung County, respectively. Then, the long-term variations of ozone concentrations were analyzed using trend formula proposed by Holland et al. (1999). Finally, multi-variable factor analysis was applied to study the relationships among the ozone concentrations with other air pollutants and meteorological parameters. Results reveal that the highest peak of ozone concentration appears in October and the second peak appears in March, while the lowest one appears in summer. Except being moderate relationships in Tzyo-Yin station, trend results show strong relationships in all other stations. Results also show that the percentage annual increase in ozone concentration in Kaohsiung City is higher than those in Kaohsiung and Ping-Tung Counties. The factor analyses reveal that the concentration of ozone is positively correlated with air temperature, wind speed and period of sunshine, while negatively correlated with concentrations of NO2, CO, NO, and NOx in the seasons of spring, autumn and winter; but negatively correlated with relative humidity in autumn. Notably, the percentage increases of ozone events in recent years should be also related to the rises of air temperature and period of sunshine, which should be watched continuously.

PM10 and ozone are the dominant air pollutants in the Urban Kaohsiung. Ozone is a secondary pollutant generated in the troposphere from the precursors nitrogen dioxide and non-methane hydrocarbons. The trends of ozone concentrations first statistically are summarized utilizing the monitoring data during the period 1998¡Ð2007. All data are collected from four fixed-site air quality monitoring stations in Kaohsiung City. The results show that ozone concentration in Kaohsiung has one perennial peak concentration, occurring in October and March. The highest values occur in October and the secondary high value in March. The lowest values occur in the summer. The monitor data possess timeliness of data and the non-linear dynamic tendency. Artificial Neural Network ¡]ANN¡^, a system recognition, self-study function and ability of the solution to non-linearity dynamic system problem, was used as a tool to analyze these monitor data. This work utilizing neural networks develops a model to predict the trend of ozone situations in the Urban Kaohsiung. The network was trained using meteorological factor and air quality data when the ozone concentrations are the highest. The optimum set value of five parameters including date partition, hidden layer neurons, training function, leraning rate , and momentum coefficient were obtained based on trial and error methods. The simulated results of ozone concentration have a correlation coefficient within the range 0.865¡Ð0.899 and IOA within the range 0.927¡Ð0.934. The trend results of ozone concentration reflect strong relationships in all stations. The results of this study indicate that the artificial neural network (ANN) is a promising method for air pollution modeling.