Pollutants concentrations are usually spatiotemporal heterogeneity due to the emission source locations, urban underlying surface conditions, climate differences and inter regional transmissions, etc. Understanding spatial distribution of air pollution is a necessary prerequisite for environmental, ecological and epidemiological studies, as well as pollution control and management.  

A research team led by Prof. Hu Yuanman from the Institute of Applied Ecology of the Chinese Academy of Sciences has identified spatial pattern of PM_2.5 at urban scales using limited monitoring resources in Liaoning central urban agglomeration (LCUA). 

Land use regression (LUR) model builds the relationship between sample locations and environmental variables to predict pollutant concentrations at unknown sites. At urban or even smaller scales, the approach has shown stronger performance when compared to dispersion modelling, remote sensing techniques and spatial interpolation.  

Sufficient numbers of samples are required to obtain high-quality LUR models. However, there are no more than twenty national fixed monitoring stations in urban area in China, and only eleven monitoring sites exist in cities of LUCA at most, which makes LUR modelling difficult at each urban scale. 

In the study, forty-one fixed monitoring data from seven cities in LUCA were combined to construct the heating, non-heating and annual average LUR models. Then models extended with regression kriging were used to predict the pollution surfaces of LCUA cities. The fitted models explained 52–61% of the variation in the PM_2.5 concentrations. The realization of the method relied on building three-dimensional morphology variables.  

The study entitled “Land use regression modelling of PM_2.5 spatial variations in different seasons in urban areas” was available online in Science of the Total Environment, and will be published on November 15. 

The research was supported by the National Natural Science Foundation of China.   

Fig 1. Heating-season (A), non-heating-season, (B) and annual average, and (C) PM_2.5 pollution surfaces of LCUA cities generated using regression kriging (Image by SHI Tuo).