Naomi Zimmerman

Mitigating household air pollution in low- and middle-income countries (LMICs) is critical for sustainable development and saving lives. This thesis explores strategies to reduce black carbon (BC) exposure from burning solid biomass fuels (SBFs) in rural Indian households.In Chapters 1 and 2, an extensive literature survey explores the challenges and opportunities in BC monitoring. Emphasis is placed on how advanced techniques can enhance actionable research, particularly in developing engineering solutions and policies for economically disadvantaged populations. Key research questions and gaps are highlighted to steer future studies.Chapter 3 focuses on the assessment of a BC monitoring device. A portable battery-operated microaethalometer was evaluated for BC mass concentration monitoring with a high degree of precision compared to a reference aethalometer (slope range 0.73-1.01, R2 = 0.9). The chapter also explores established aethalometer correction protocols and suggests methods for estimating source contributions.Chapter 4 employs an integrated data collection approach using surveys and BC data from rural northern India. With 81% of households using SBF, intra-day BC fluctuations occurred both indoors (4.2 to 37.6 µgm−3) and outdoors (5.7 to 33.3 µgm−3). Rapid BC changes are mainly due to atmospheric ventilation post-morning cooking. Surveys and BC data helped form hypotheses on fuel-use related to atmospheric conditions, helping to inform interventions: e.g., hypothesized optimal SBF cooking times.Chapter 5 presents the nested mass balance-box (NMB2) model, parameterized with field data and scholarly sources. This model employs the Monte Carlo method to simulate hourly concentrations of indoor-outdoor BC concentrations under various household interventions, such as transitioning fuels and enhancing ventilation, to assess the efficacy of ways to mitigate BC exposure. Transitioning fully to LPG can decrease BC levels during cooking by 99% in poorly ventilated homes. An interesting finding was noted for homes with complete fuel transition to LPG andhigh ventilation: infiltration of outdoor air led to a slight 9% increase in BC concentration.Chapter 6 summarizes key findings and lessons from the thesis, offering evidence-based decision guidance for reducing BC exposure through the study’s holistic approach of field campaigns, advanced instruments, and modeling.

View record

The rapid expansion of cannabis cultivation and processing facilities (CCFs/CPFs) following legalization has outpaced the understanding of their environmental and occupational impacts, particularly on air quality. This thesis addresses several stages of an air quality impact assessment, including method development, emission estimation, and consequences of emissions for occupational and ambient air.In Chapter 2, I present a comprehensive literature review on the air quality impacts of cannabis cultivation, identifying 16 key research gaps that informed the methodology of subsequent chapters. Chapter 3 examines the temporal variation of 22 terpenes across eight rooms in a CCF and six of a CPF using automated gas chromatography. Emissions increased with plant maturity, were affected by environmental conditions (e.g., lights-on periods raised terpene concentrations by 20–90%), and peaked during specific CPF activities. Omitting the chemical and temporal variability of emissions risks underestimating their impact, especially in odour modeling.In Chapter 4, I assess occupational concentrations of ultrafine particles (UFPs), an unregulated but health-relevant pollutant. Key findings include: (i) an artifact in the FMPS when terpene concentrations exceed 600 ppb, (ii) greater UFP concentrations in CPF rooms due to source-sink dynamics, and (iii) lung-deposited surface area (LDSA) concentrations in CPF environments comparable in magnitude to those in traffic and welding contexts.Chapter 5 evaluates downwind ambient air impacts of a > 440,000 m2 CCF using mobile monitoring and dispersion modeling. Results show: (i) a 3 ppb increase in ozone in urban downwind areas, (ii) a 150–200% decrease in UFPs downwind, and (iii) elevated sub-20 nm particles when urban NOx plumes intersected the facility under strong sunlight. Modeled terpene enhancements aligned with odour reports, defining a 3 km odour impact zone.Chapter 6 synthesizes findings and revisits the identified research gaps in light of five years of research and engagement with stakeholders across academia, industry, regulation, and the community.

View record

Recent advancements in low-cost sensor (LCS) technology have presented a new and affordable opportunity to understand and subsequently improve air quality. This thesis assessed the different stages of adoption and application of LCS technology, including calibrating the sensors, using sensors to build spatiotemporal pollutant maps, and using these maps to identify inequities in air pollution exposures.In Chapter 3, a general calibration method for commercially available low-cost PM₂.₅ sensors (PurpleAir/Plantower) was explored, such that the calibration models can be transferable to large geographical areas, especially in areas with limited monitoring. Inter-city models (e.g., trained in California and tested in India) built for regional concentrations were found to be effective in reducing errors by 30% in measurements. Chapter 4 used data from a network of 50 LCS deployed in Pittsburgh (Pennsylvania, USA) to build daily average land-use regression (LUR) and random-forests (LURF) spatiotemporal models for PM₂.₅, NO₂, and CO. The LURF models outperformed traditional regression techniques, with an increase in average externally cross-validated R² of 0.10-0.19. Models built after separating local contributions from the regional signal improved the R² by 0.14. In Chapter 5, the LURF models for PM₂.₅ were then used to build static (population spends 24 hours/day in a fixed residential area) and dynamic models (population moves between residential and commercial areas) and used to estimate variations in residents’ exposures to PM₂.₅ due to movement. The exposure estimates were consistently about 10% higher when the population spends more time in commercially-dense locations (dynamic model) vs residentially-dense locations (static model). Weekend concentrations were also 10% higher than weekday concentrations. Chapter 6 describes the deployment and analysis of data from a network of 11 LCS deployed in an environmental injustice neighborhood in Vancouver (British Columbia, Canada). PM₂.₅, NO₂, and O₃ concentrations were used to calculate cumulative hazard indices (CHIs) to identify hotspots within the neighborhood and to address the inequities in air pollution when compared to the Greater Vancouver region. Lastly, Chapter 7 summarizes the lessons learned from this thesis and provides insight into key design deployment considerations.

View record