Based on the ozone monitoring data from 2014 to 2018, we presented the variation of ozone concentration in Xi'an and revealed the effects of ozone concentration by meteorological factors based on the generalized additive model (GAM). The results showed that ① with increasing ozone concentration year by year, the assessment standard of ozone pollution was overtaken by three consecutive years since 2016. However, the rising trend was slowed down since 2017 as a result of the strengthened pollution control during summer. ② The monthly curve of ozone concentration was presented as a reversed "V" model with a rising trend accompanying the rising temperature from January to July and a decreasing one during the rest of the year, peaking in July in terms of average monthly ozone concentration. However, this model would turn into an "M" in years with high precipitation when the valley witnessed the highest precipitation in a month. ③ The ozone pollution increased from the year 2014 to 2018 with a stretch-forward ozone polluted time. Furthermore, the rates of ozone non-attainment increased from 1.9% in 2014 to 14% in 2018. In addition, the time ozone pollution emerged advanced from July to May. ④ Based on the GAM model, ozone concentration was non-colinearly related to temperature, air pressure, sunshine duration, and relative humidity. However, the curves of these factors varied considerably, with a positive influence of temperature and sunshine duration and a negative influence of air pressure and relative humidity. The influence of precipitation was mainly witnessed in summers, while no influence of wind was observed. Furthermore, ozone pollution can be easily triggered under the following conditions:temperature>24℃, air pressure <962 hPa, sunshine duration>9 h, and a relative humidity 36%-65% with no rain.
Keywords: Xi'an; generalized additive model (GAM); influence factors; ozone concentration; variation features.