Global Modeling of Fluorescent Aerosol Particles with Machine Learning Reveals Potential Regional Anthropogenic Impacts

Primary biological aerosol particles (PBAPs) significantly affect human health and aerosol-cloud-climate interactions. Fluorescent aerosol particles (FAPs), detected using light/laser-induced fluorescence (LIF) instruments, serve as a crucial proxy for understanding the concentration and size distribution of PBAPs and the factors that influence their variability in the atmosphere. This study systematically evaluates FAPs collected from field measurements worldwide and simulates their concentrations on a global scale using machine learning algorithms, incorporating comprehensive global weather, climate and emissions data. The simulated global concentration reveals spatial variations in size and concentration, with heightened annual mean concentration predominantly observed in the tropics and the Asia region. The post-hoc Shapley Additive Explanation (SHAP) method indicates that the spatio-temporal patterns of FAPs concentrations and size distributions are primarily driven by anthropogenic emissions in urban regions, while weather factors are more closely linked to variations in oceanic and rural areas. Notably, certain non-biological emissions (e.g., dust and black carbon) exhibit strong correlations with FAPs, particularly in densely populated areas and the Arctic region. Overall, this study underscores the significant role of anthropogenic emissions in shaping simulated FAP concentrations on a global scale and provides guidance for future investigations into FAP concentrations in unexplored regions.