Spatially heterogeneous relationships of PM 2.5 concentrations with natural and land use factors in the Niger River Watershed, West Africa

Identifying the main factors influencing PM 2.5 concentrations is critical for the control and management of elevated atmospheric pollution. Although the impact of natural and land use factors on PM 2.5 concentration has previously been quantified, in this paper we show that multiple environmental factors exhibit significant spatial heterogeneity at a range of scales. A more comprehensive analysis of PM 2.5 variations require the potential effects of seasonal variations in meteorological conditions, interactions between natural and land use factors, and possible multi-scale spatial effects to be accounted for. Based on the global remote sensing inversion dataset, the aim of this study is to elucidate the spatially heterogeneous relationships of PM 2.5 concentrations with natural and land use factors, and their multi–scale effects. This is achieved through combining geographical detector (GD), random forest (RF) and multiscale geographically weighted regression modelling (MGWR) to reveal the underlying mechanisms of spatial variation of PM 2.5 concentrations in the Niger River Watershed (NRW). The results show that natural factors, represented by meteorological and topography variables, are most prominent. Human activities, represented by land use, have significant indirect effects through interactions with meteorological factors. Elevation (ELE) and annual minimum temperature (TMIN) emerge as the most important interaction centers among natural factors. In addition, the interaction between natural and land use factors are crucial in considering PM 2.5 transmission and dispersion in the NRW. The topography and meteorological factors underlie the basic spatial pattern PM 2.5 concentrations in the NRW, among which seasonal meteorological parameters have a more prominent impact on PM 2.5 concentration. Both natural and land use factors exhibit strong non–stationarity in respect of spatial variations in PM 2.5, and the effect of environmental factors in general on PM 2.5 is scale–dependent. Transport and dispersion are important constraints in terms of aerosol loading regionally. The study suggests that coupling processes between meteorological conditions, land use and PM 2.5 concentrations, based on the scale effects of each factor, should be accounted for in developing aerosol pollution regulations and cooperative management strategies in the region.