A Functional Data Analysis of Spatiotemporal Trends and Variation in Fine Particulate Matter

Area of Science:

• Environmental Science
• Data Science
• Atmospheric Chemistry

Background:

• Particulate matter (PM2.5) poses significant environmental and health risks.
• Understanding the spatiotemporal variability of PM2.5 components is crucial for effective environmental management.
• Existing methods like Kriging have limitations in capturing complex temporal dynamics.

Purpose of the Study:

• To apply modern functional data analysis (FDA) methods to investigate PM2.5 component variability across the United States.
• To model the dynamic behavior of pollutant annual profiles over time and space.
• To enable prediction of pollutant profiles for unobserved locations and years and facilitate data visualization.

Main Methods:

• Utilized functional data analysis (FDA) to model pollutant annual profiles.
• Applied the method to daily PM2.5 concentrations from monitoring sites across the US (2003-2015).
• Compared the FDA approach with traditional methods like Kriging for spatiotemporal analysis.

Main Results:

• The FDA approach successfully modeled the spatiotemporal variability of PM2.5 components.
• The method allowed for prediction of pollutant profiles in data-scarce areas and years.
• New trends in pollutant changes across seasons and years were identified.

Conclusions:

• Functional data analysis provides a powerful tool for studying complex environmental data like PM2.5.
• The developed method offers enhanced capabilities for prediction, visualization, and trend analysis compared to conventional approaches.
• This research contributes to a deeper understanding of PM2.5 dynamics, aiding in environmental policy and public health initiatives.