Study of the Relationship between Urban Expansion and PM10 Concentration Using Multi-Temporal Spatial Datasets and the Machine Learning Technique: Case Study for Daegu, South Korea

Abstract

To protect the population from respiratory diseases and to prevent the damages due to air pollution, the main cause of air pollution should be identified. This research assessed the relationship between the airborne particulate concentrations (PM10) and the urban expansion in Daegu City in South Korea from 2007 to 2017 using multi-temporal spatial datasets (Landsat images, measured PM10 data) and the machine learning technique in the following steps. First, the expanded urban areas were detected from the multiple Landsat images using support vector machine (SVM), a widely used machine learning technique. Next, the annual PM10 concentrations were calculated using the long-term measured PM10 data. Finally, the degrees of increase of the expanded urban areas and of the PM10 concentrations in Daegu from 2007 to 2017 were calculated by counting the pixels representing the expanded urban areas and computing variation of the annual PM10 concentrations, respectively. The experiment results showed that there is a minimal or even no relationship at all between the urban expansion and the PM10 concentrations because the urban areas expanded by 55.27 km2 but the annual PM10 concentrations decreased by 17.37 μg/m³ in Daegu from 2007 to 2017.