Analysis of spatiotemporal variation and relationship to land use - landscape pattern of PM2.5 and O3 in typical arid zone

Land use - landscape pattern act as pollutant sources carries for PM2.5 and ozone, understanding the relationship between them is valuable for the management of atmospheric environment. In this investigation, we examined the spatiotemporal distribution characteristics of PM2.5 and ozone in Xinjiang during 2000 similar to 2022. Meanwhile, we employed the random forest model under optimal model selection to explore the response of the land use - landscape pattern to the pollutants. The conclusions suggested: (1) PM2.5 in 94.82% of the area showed a significant decrease, with high levels concentrated in the Tarim Basin, Turpan-Hami Basin. Ozone demonstrated a remarkable upward trend in 99.30% of the region, showing high levels in the Tarim Basin and the northern Xinjiang. (2) Upon comparison with the univariate and stepwise linear regression models, random forest model holds significant practical value in quantifying the impact of pollutants on land use-landscapes. (3) In the land use - landscape pattern, the complexity and aggregation of cropland and water, the degree of connectivity in barren and impervious, the diversity and complexity of forest, and the degree of aggregation of grasslands all strongly impact PM2.5. For ozone concentrations, the index characterizing the degree of patchy aggregation is overwhelmingly dominant.