Modelling daily PM2.5 concentrations at high spatio-temporal resolution across Switzerland

摘要

AbstractSpatiotemporal resolved models were developed predicting daily fine particulate matter (PM2.5) concentrations across Switzerland from 2003 to 2013. Relatively sparse PM2.5monitoring data was supplemented by imputing PM2.5concentrations at PM10sites, using PM2.5/PM10ratios at co-located sites. Daily PM2.5concentrations were first estimated at a 1 × 1km resolution across Switzerland, using Multiangle Implementation of Atmospheric Correction (MAIAC) spectral aerosol optical depth (AOD) data in combination with spatiotemporal predictor data in a four stage approach. Mixed effect models (1) were used to predict PM2.5in cells with AOD but without PM2.5measurements (2). A generalized additive mixed model with spatial smoothing was applied to generate grid cell predictions for those grid cells where AOD was missing (3). Finally, local PM2.5predictions were estimated at each monitoring site by regressing the residuals from the 1 × 1km estimate against local spatial and temporal variables using machine learning techniques (4) and adding them to the stage 3 global estimates. The global (1 km) and local (100 m) models explained on average 73% of the total,71% of the spatial and 75% of the temporal variation (all cross validated) globally and on average 89% (total) 95% (spatial) and 88% (temporal) of the variation locally in measured PM2.5concentrations.Graphical abstractDisplay OmittedHighlights•High resolution daily PM2.5maps are needed to facilitate epidemiological research.•Spatial-temporal statistical approach with AOD and ground measurements.•PM10monitoring sites used to supplement relative sparse number of PM2.5sites.•Daily PM2.5concentrations estimated for Switzerland (2003–2013) at 100 m resolution.•We achieve reliable estimates across complex topography and climatic conditions.We present high spatiotemporal models for daily PM2.5exposure (2003–2013) across Switzerland at 100 × 100 m, using satellite AOD and supplementing sparse monitoring data with the PM10network.