**4. Discussion**

#### *4.1. Temporal and Spatial Variations of PM10 and PM2.5*

The mean annual concentrations of PM10 and PM2.5 decreased significantly from 2013 to 2017, while concentrations during 2015–2017 did not exhibit any significant variations. Studies have shown that improvements in air quality in megacities are significant in the early stages of governance and then stalled. The improvement of environmental quality rarely accompanies urbanization and industrialization [21]. The inter-annual variation may be attributed to human activities, including auto emissions and steel productions. Additionally, the demand for housing intensified by the expansion of urban land could increase construction dust emissions. The observed variations in the concentrations of PM2.5 and PM10 may thus be a result of specific weather conditions, combined with changes in land use and the increased development of urban infrastructure [2,24].

Regarding the seasonal variations in PM10 and PM2.5, both the highest concentrations occurred in winter and the lowest in summer, the second-highest in spring and autumn. This phenomenon might be attributable to frequent sandstorms originating in northwestern China with bare land cover and strong wind, which produced anomalously high concentrations of mineral dust, resulting in a prominent rise of particulate matter in spring [6]. In autumn, straw burning after agricultural harvests results in the rise in particulate matter concentrations. In addition, meteorological factors also play an important role in these processes. Cloudy weather and low wind speed could contribute to the elevated particulate matter concentrations [25]. Many cities seek to alter meteorological conditions to some extent by changing the urban landscape [26]. Wind and turbulence within the urban canopy can play an important role in local measurements of meteorological conditions. This is especially true if pollutants are frequently capped by a statically stable atmospheric layer [2,14]. Such stable conditions may contribute to the lack of spatial variations observed in the downtown area. Our results showed nonsignificant difference for the concentrations of PM10 and PM2.5 among the 13 stations. Zhang et al. [27] analyzed the particulate matter at six sites in Xi'an, and also found nonsignificant spatial variations. However, the concentrations of particulate matter were lower in the stations that locate near places with higher tree coverage rates and large water bodies. Meanwhile, the concentrations were higher in the stations surrounded by high energy consumption and bare land. Based on these observations, we believe that the spatial homogeneity of the air quality is also affected by human activities (i.e., pollutant emissions and land use).
