Methodological Advancements for Source Attribution and Optical Characterization of Carbonaceous Aerosols Using Advanced Statistical and Optimization Techniques

Carbonaceous aerosols critically impact both air quality and climate, necessitating a precise understanding of their sources and optical properties for effective mitigation. Current source apportionment methods often rely on oversimplified assumptions, leading to significant uncertainties in understanding these complex atmospheric components. This dissertation addresses these limitations by presenting methodological advancements for the source attribution and optical characterization of carbonaceous and PM2.5 components using advanced statistical and optimization techniques. Specifically, to improve the source apportionment of light-absorbing carbonaceous aerosols (LAC), a three-stage variable component (3SVC) model was developed using a generalized reduced gradient (GRG) optimization method. This model accurately estimated contributions of black carbon (BC) and brown carbon (BrC) without the restrictive reliance on fixed absorption Ångström exponent (AAE) assumptions, outperforming existing approaches in error reduction and tracer correlation. Furthermore, a Technique for order preference by similarity to ideal solution (TOPSIS) based multi-criteria decision-making (MCDM) framework was employed to derive source-specific AAE values from short-term Aethalometer data. This framework provided biomass burning and fossil fuel BC estimates that were highly consistent with established temporal patterns and molecular marker trends. All methods were rigorously applied to high-resolution wintertime datasets collected over four intensive campaigns in Interior Alaska. This region, characterized by extreme cold, shallow mixing heights, and persistent PM2.5 pollution episodes, provided a challenging and invaluable testbed for evaluating the efficacy of these advanced source attribution methods.