Meteorology-driven variability of air pollution (PM1) revealed with explainable machine learning

Abstract

Air pollution, in particular high concentrations of particulate matter smaller than 1 span classCombining double low lineinline-formulam in diameter (PMspan classCombining double low lineinline-formula1), continues to be a major health problem, and meteorology is known to substantially influence atmospheric PM concentrations. However, the scientific understanding of the ways in which complex interactions of meteorological factors lead to high-pollution episodes is inconclusive. In this study, a novel, data-driven approach based on empirical relationships is used to characterize and better understand the meteorology-driven component of PMspan classCombining double low lineinline-formula1 variability. A tree-based machine learning model is set up to reproduce concentrations of speciated PMspan classCombining double low lineinline-formula1 at a suburban site southwest of Paris, France, using meteorological variables as input features. The model is able to capture the majority of occurring variance of mean afternoon total PMspan classCombining double low lineinline-formula1 concentrations (coefficient of determination (span classCombining double low lineinline-formulaiR/i2) of 0.58), with model performance depending on the individual PMspan classCombining double low lineinline-formula1 species predicted. Based on the models, an isolation and quantification of individual, season-specific meteorological influences for process understanding at the measurement site is achieved using SHapley Additive exPlanation (SHAP) regression values. Model results suggest that winter pollution episodes are often driven by a combination of shallow mixed layer heights (MLHs), low temperatures, low wind speeds, or inflow from northeastern wind directions. Contributions of MLHs to the winter pollution episodes are quantified to be on average span classCombining double low lineinline-formulag1/45 span classCombining double low lineinline-formulag/mspan classCombining double low lineinline-formula3 for MLHs below span classCombining double low lineinline-formula<500 m a.g.l. Temperatures below freezing initiate formation processes and increase local emissions related to residential heating, amounting to a contribution to predicted PMspan classCombining double low lineinline-formula1 concentrations of as much as span classCombining double low lineinline-formulag1/49 span classCombining double low lineinline-formulag/mspan classCombining double low lineinline-formula3. Northeasterly winds are found to contribute span classCombining double low lineinline-formulag1/45 span classCombining double low lineinline-formulag/mspan classCombining double low lineinline-formula3 to predicted PMspan classCombining double low lineinline-formula1 concentrations (combined effects of span classCombining double low lineinline-formulaiu/i-and span classCombining double low lineinline-formulaiv/i-wind components), by advecting particles from source regions, e.g. central Europe or the Paris region. Meteorological drivers of unusually high PMspan classCombining double low lineinline-formula1 concentrations in summer are temperatures above span classCombining double low lineinline-formulag1/425 span classCombining double low lineinline-formulagC (contributions of up to span classCombining double low lineinline-formulag1/42.5 span classCombining double low lineinline-formulag/mspan classCombining double low lineinline-formula3), dry spells of several days (maximum contributions of span classCombining double low lineinline-formulag1/41.5 span classCombining double low lineinline-formulag/mspan classCombining double low lineinline-formula3), and wind speeds below span classCombining double low lineinline-formulag1/42 m/s (maximum contributions of span classCombining double low lineinline-formulag1/43 span classCombining double low lineinline-formulag/mspan classCombining double low lineinline-formula3), which cause a lack of dispersion. High-resolution case studies are conducted showing a large variability of processes that can lead to high-pollution episodes. The identification of these meteorological conditions that increase air pollution could help policy makers to adapt policy measures, issue warnings to the public, or assess the effectiveness of air pollution measures./p.