Increased air pollution is driven by anthropogenic pollution emissions and climate change, which pose great challenges to environmental governance. Strengthening the monitoring of regional air quality levels and analyzing the causes of regional pollution is conducive to the management and sustainable development of the regional atmosphere. Functional data obtained on a wavelet basis were used in the fitting of air quality data of Fujian Province, and wavelet decomposition was performed to obtain low-frequency and high-frequency information. While the Fourier basis cannot adaptively adjust the time–frequency window, resulting in the loss of location information of special frequencies, the wavelet basis solves this problem. Functional analysis of variance was utilized for analyzing spatial differences in air pollution characteristics. Furthermore, the study established a multivariate functional linear regression model to explore the impact of meteorological factors and ozone precursor factors. The results indicated that the overall air quality was gradually improving in Fujian Province, but the concentration of ozone was progressively increasing. Air pollution in coastal areas was higher than that in inland areas. The p-values of the functional analysis of variance for energy values and crest values were less than 0.05. Moreover, the energy entropy and kurtosis values were greater than 0.05. There were significant differences of AQI in the fluctuation amplitude and variation characteristics of different cities. The total squared multiple correlation of regression model was above 50% on average. Ozone is currently the most serious pollution factor, mainly affected by wind speed, temperature, NO2, and CO. In summer, it was principally influenced by VOCs. The findings of this study could act as a reference in exploring the time–frequency characteristics of air quality data and support of air pollution control.
