Title of Thesis: Linking Climate and Air Quality: Insights from Machine Learning, Modeling, and Field Observations
Abstract of Work: 
Climate change and air pollution are two of the most pressing environmental challenges of the 21st century, both contributing significantly to adverse human health outcomes. Their complex interactions—driven by meteorology, atmospheric chemistry, aerosol-radiation interactions—make understanding their linkages critical for public health and policymaking. This dissertation investigates key pathways of these interactions using machine learning models, chemical transport modeling, and field measurements.
First, we examined the relationship between summertime PM2.5 and O3concentrations and ambient temperature across the contiguous United States, using high-resolution machine learning estimates and ground-based observations. We found widespread positive temperature sensitivities, which have weakened in the eastern U.S. due to emission reductions under the Clean Air Act, but have intensified in the western U.S., driven by increasing wildfire activity. 
It is important for models to capture this relationship and its long-term pattern for future air quality and climate projection. Therefore, in the second part, we further evaluated and improved the GEOS-Chem chemical transport model, identifying and correcting sources of bias that previously limited its ability to reproduce observed PM2.5–temperature relationships. The improved model showed better agreement with observations and allowed us to quantify the contributions of temperature-sensitive processes. In the eastern U.S., chemical production was the dominant driver of both the magnitude and long-term decline in PM2.5 sensitivity, while transport explained interannual variability. In the west, wildfire emissions emerged as the primary contributor to the recent increase in PM2.5–temperature sensitivity, underscoring the need for stronger fire management policies. 
Finally, we investigated the mixing state of black carbon (BC)—a key light-absorbing aerosol—using in situ measurements from an urban site (Atlanta, GA) and a rural background site (Boone, NC). BC particles in Atlanta exhibited relatively uniform relative coatings due to dominant local sources, while those at the rural site showed greater heterogeneity, reflecting the influence of distinct air masses. The absorption enhancement factor calculated based on realistic mixing states deviated significantly from assumptions of fully internal mixing—by over 40% at the rural site—highlighting the importance of resolving aerosol mixing state in radiative transfer simulations.
Defense Date: July 28
Defense Time: 11 am – 2 pm
Defense Location: L1114
Committee: Pengfei Liu (Advisor), Greg Huey, Rodney Weber, Sally Ng, Da Pan