Modeling and Analysis of Black Carbon in the Third Pole: From Emissions to Snowmelt

Abstract

High Mountain Asia (HMA), often referred to as the Third Pole and Asia's water tower, sustains the livelihood of over 30% of our global population. However, increasing anthropogenic activities within developing economies in the vicinity of HMA's glaciers threaten the sustainability of this water tower. Deposition of light-absorbing particles from these human activities (LAPs, particularly black carbon (BC) and dust) and their interactions with regional climate have been found to contribute to accelerated snowmelt and glacier retreat in the last few decades. There are significant uncertainties in our understanding of how LAPs interact with the cryospheric interface, their sources, transport/removal across HMA, and their representation in Earth system models, precluding accurate predictions and assessments of their impacts. This dissertation addresses some of these gaps by integrating across available observations, a fully coupled novel regional reanalysis, inverse modeling, and machine learning diagnostics to quantify the impacts of LAPs, particularly BC, on HMA’s cryosphere and their key emission sources. First, this dissertation introduces and evaluates MATCHA, a high-resolution (12 km), fully coupled hydroclimate-chemical reanalysis for Asia spanning 17 years between 2003 to 2019. MATCHA provides a first-of-its-kind regional dataset that accounts for aerosol-snowpack interactions, tags BC emissions across major regions and sectors, and assimilates satellite data for chemical species. Second, tagged estimates of BC from MATCHA are leveraged within a hierarchical Bayesian inversion framework constrained by 91 observation sites across HMA, to quantify regional and sectoral BC source contributions and assess model biases, revealing significant underestimations in prior emission inventories, particularly for anthropogenic sources in remote locations and biomass burning. Subsequently, this work investigates the impact of LAPs on HMA's cryosphere, utilizing statistical and machine learning methods with network theory to help unravel the non-linear interactions between aerosols and meteorological processes impacting snowmelt. My findings demonstrate that these coupled aerosol-meteorology interactions are statistically significant and an underrepresented contributor to snow cover variability, especially during the late snowmelt season, where BC and large-scale circulation emerge as dominant factors. The results also reveal inconsistencies in how feedbacks between dust, temperature, and circulation are represented across state-of-the-art reanalyses, highlighting the need for improved representation of these feedbacks in the development of Earth system models. Overall, this dissertation provides a quantitative, observationally-constrained assessment of BC, its sources, and its complex interactions with HMA's cryosphere. By investigating biases in emissions and modeled aerosol-meteorology-snow processes, the results emphasize the need to refine emission inventories, develop region-specific model-coupled parameterizations, and implement targeted policies to mitigate LAPs-related pollution. Furthermore, it introduces a novel diagnostic framework and allied techniques that can assist in evaluating structural biases in current models, informing targeted model developments and observational monitoring strategies to improve Earth system predictability and future projections of freshwater availability in this climate-vulnerable region.