Interpretation of Ground-Based Measurements from the Surface Particulate Matter Network to Understand the Global Distribution of Fine Particulate Matter

Abstract

Exposure to ambient fine particulate matter (PM2.5¬) is increasingly recognized as the leading environmental risk factor for global burden of disease. This thesis develops the Surface PARTiculate mAtter Network (SPARTAN) to provide long-term measurements of PM2.5 mass and chemical composition, collocated with existing aerosol optical depth (AOD) observations in highly populated, globally diverse regions. Three projects are presented that interpret SPARTAN measurements to provide insight into the spatial variation in ground-based PM2.5 chemical composition, into the sources contributing to PM2.5, and into the relationship between AOD and PM2.5 used in satellite-based estimates of PM2.5. Analysis of SPARTAN filter samples collected across multiple continents for PM2.5 chemical composition show that absolute concentrations of several major components vary by more than an order of magnitude across sites, and exhibit consistency with available, collocated studies. Elevated Zn:Al ratios reveal an enhanced anthropogenic dust fraction relative to natural sources, signifying the need to include this PM2.5 source in global models and emission inventories. The developed compositional dataset provides much needed long-term chemical data for investigation of sources leading to the spatial variation of PM2.5 mass and chemical composition. Evaluation of the GEOS-Chem model, constrained by satellite-based estimates of PM2.5 and informed by SPARTAN compositional measurements, shows significant spatial consistency for major chemical components. Measured PM2.5 composition corroborate source attribution from sensitivity simulations, providing confidence in utilizing sensitivity simulations to explore the influence of source categories to global population-weighted PM2.5. This approach of coupling observational datasets with modelling at the global scale allows for insight into the main sources determining PM2.5 global variation, but also identification of modelled processes that require development to represent the wide range of PM2.5 and composition observed globally. An initial comparison between empirical and simulated relationships of PM2.5 and columnar AOD ( ) was conducted using the GEOS-Chem global chemical transport model. This comparison is the first to develop empirical, ground-based and provide an evaluation of modelled values widely used in satellite-based estimates. Collocated, modelled values generally fall within a factor of two of measured values and have a mean fractional bias that is an order of magnitude lower than for either PM2.5 or AOD alone. This lower bias in indicates that satellite-derived PM2.5 inferred using is likely to have lower bias than purely simulated PM2.5¬.