Application of Functional Principal Component Analysis in the Spatiotemporal Land-Use Regression Modeling of PM2.5

Functional data are generally curves indexed over a time domain, and land-use regression (LUR) is promising spatial technique for generating high-resolution estimation of retrospective long-term air pollutants. We developed methodology the novel functional (FLUR) model, which provides temporal estimations Long-term fine particulate matter (PM2.5) in megacity Tehran, Iran, was used as practical example. The hourly measured PM2.5 concentrations were averaged each hour monitoring station. Penalized smoothing employed to construct smooth diurnal curve using 30 stations. principal component analysis (FPCA) extract FPCA scores from pollutant curves, LUR models fitted on scores. mean all had maximum 39.58 µg/m3 at 00:26 a.m. minimum 29.27 3:57 p.m. explained about 99.5% variations observed across city just three components. evaluation spatially predicted every 15 min provided series 96 exposure maps. presented results could benefit future environmental epidemiological studies.