Spatiotemporal distribution of PM2.5 concentrations in Shaanxi Province, China, and its responses to land use changes and meteorological factors

Abstract

Understanding the spatiotemporal patterns and factors influencing PM_2.5 concentrations is crucial for implementing effective pollution control measures. In this study, we used a gridded dataset of annal PM_2.5 concentrations, together with meteorological and land cover data from 2000 to 2020, to analyze the spatial‒temporal patterns of PM_2.5 concentrations and their responses to land use changes and meteorological factors in Shaanxi Province, China. Trend analysis was employed to identify overall temporal patterns, a random forest (RF) was used to evaluate the importance of influencing factors, and the geographically weighted regression (GWR) method was applied to assess spatial heterogeneity and local effects. The annual PM_2.5 concentration decreased by 43.52 % from 2000 to 2020, with higher concentrations in the central region and lower concentrations in the southern and northern areas. The PM_2.5 concentration was negatively correlated with the interconversion of forests and grasslands and positively correlated with conversions among croplands, impervious surfaces, and water bodies. The RF regression results indicated that croplands, impervious surfaces, and their mutual interconversions exerted a greater impact on PM_2.5 concentrations than did the other land use types. The GWR analysis results revealed that the factors influencing PM_2.5 concentration, in descending order of importance, were as follows: wind speed, precipitation, relative humidity, temperature, sunshine duration, atmospheric pressure, conversion of impervious surfaces to cropland, conversion of cropland to impervious surfaces, unconverted cropland, and unconverted impervious surfaces.