Assessing the Spatiotemporal Characteristics, Factor Importance, and Health Impacts of Air Pollution in Seoul by Integrating Machine Learning into Land-Use Regression Modeling at High Spatiotemporal Resolutions

Abstract

Previous studies have characterized spatial patterns of air pollution with land-use regression (LUR) models. However, the spatiotemporal characteristics of air pollution, the contribution of various factors to them, and the resultant health impacts have yet to be evaluated comprehensively. This study integrates machine learning (random forest) into LUR modeling (LURF) with intensive evaluations to develop high spatiotemporal resolution prediction models to estimate daily and diurnal PM 2.5 and NO 2 in Seoul, South Korea, at the spatial resolution of 500 m for a year (2019) and to then evaluate the contribution of driving factors and quantify the resultant premature mortality. Our results show that incorporating the random forest algorithm into our LUR model improves the model performance. Meteorological conditions have a great influence on daily models, while land-use factors play important roles in diurnal models. Our health assessment using dynamic population data estimates that PM 2.5 and NO 2 pollution, when combined, causes a total of 11,183 (95% CI: 5837−16,354) premature mortalities in Seoul in 2019, of which 64.9% are due to PM 2.5 , while the remaining are attributable to NO 2 . The air pollution-attributable health impacts in Seoul are largely caused by cardiovascular diseases including stroke. This study pinpoints the significant spatiotemporal variations and health impact of PM 2.5 and NO 2 in Seoul, providing essential data for epidemiological research and air quality management.

Statistical Characteristics of Air Quality Index DAQx*-Specific Air Pollutants Differentiated by Types of Air Quality Monitoring Stations: A Case Study of Seoul, Republic of Korea

“…With this monitoring of the general air pollution, the objective of both station types is achieved. The roadside AQMSs are located a few meters from the curbside of main traffic roads in Seoul and are thus affected by adjacent traffic volumes, meaning they record roadside air pollution at approximately 3 m agl but may not reflect impacts from surrounding land use [32]. The specifications of the measuring devices used at the AQMSs in Seoul are listed in Table A4.…”

“…The concentrations of air pollutants required to calculate AQIs depend on emissions, chemical transformation processes in the atmosphere, and atmospheric dispersion and dilution conditions [17][18][19][20][21][22][23][24][25][26][27][28]. They are often approximated by urban form variables in models such as land use regression (LUR) models to predict the spatial distribution of air pollutants [29][30][31][32]. The methodology used to determine AQIs is described in detail in the literature [10,33,34] and is not repeated here.…”

“…Therefore, the objective of this case study is to identify and quantify the variability in the four DAQx* air pollutants within and between AQMS types. Since they are already the subjects of previous investigations, this study does not address the determination of statistical air pollutant concentrations at individual AQMSs [14,32,38,39] or the model-based creation of maps showing spatial distributions of air pollutant characteristics in Seoul [32,[40][41][42].…”

A Machine Learning Approach for High-Resolution Modeling and Apportionment of Black Carbon Concentrations in an Impacted Community

“…Machine learning methods can be used in conjunction with computationally expensive physical models of the atmosphere. , Climate modelers have used statistical downscaling techniques as a computationally efficient way to obtain high-resolution climate predictions from coarse general circulation model output for some time. , Application of downscaling techniques to air quality modeling is a newer development. − Land-use regression models are a data-driven type of model that have historically been used to model long-term average air quality, and which recent studies have combined with machine learning techniques to capture temporal changes in concentrations. , …”

“…20−22 Land-use regression models are a data-driven type of model that have historically been used to model longterm average air quality, 23 and which recent studies have combined with machine learning techniques to capture temporal changes in concentrations. 24,25 The use of machine learning in air pollution studies has grown almost exponentially over the last decade, but these methods are still underutilized compared to their application in fields like biology, chemistry, and medicine. 26 Neural networks are a widely applicable type of machine learning algorithm consisting of layers of interconnected nodes which are able to solve multi-target regression problems, meaning that they are able to predict many continuous (as opposed to categorical) outputs, and that can represent nonlinear phenomena.…”