Urban greenhouse gas emissions as observed using a high-density sensor network
More than 70% of anthropogenic carbon dioxide emissions originate from cities, and that fraction is expected to grow with the increasing urbanization of the world population. Monitoring, reporting, and verification (MRV) of emission mitigation initiatives are complicated in urban areas by the combination of spatio-temporally heterogeneous landscapes of CO2 sources and poorly constrained turbulent fluid dynamics near the rough and irregular surfaces of urban topography. In this dissertation, I present a novel approach to MRV of urban emissions using a distributed network of near-surface ambient CO2 monitors stationed at 2 km intervals across an urban dome. The study area is the San Francisco Bay Area, a location characterized by a high density of disparate emission sources, irregular topography, as well as strong municipal and state level commitments to emission reductions. First, I describe the design, implementation, and evaluation of this unprecedented monitoring approach, hereafter the BErkeley Atmospheric CO2 Observation Network (BEACO2N). I demonstrate how the use of lower cost sensor technologies enables a high volume of monitors to be deployed at a cost competitive with conventional, non-spatially resolved monitoring approaches and find the lower cost monitors to be of sufficient accuracy and precision to capture typical urban CO2 phenomena. Second, I leverage this validated framework to provide the first high-resolution characterization of the neighborhood-to-neighborhood variability in local CO2 concentrations across the San Francisco Bay Area. I find significant differences even between nearby pairs of monitoring sites, leading to the derivation of a relatively short spatial correlation length scale for the study area and demonstrating a high degree of sensitivity to the unique emission sources local to each site. I determine these sensitivities to be capable of detecting changes in emission processes of policy-relevant magnitudes, providing a viable new constraint on regulations concerning, for example, fuel efficiency and vehicle electrification. Finally, I compare the CO2 concentrations measured at nine BEACO2N sites to those predicted by an ensemble of atmospheric transport models. I find large disagreements between the observations and the simulations, only a small portion of which can be explained by known measurement errors or the spread in plausible transport and emissions scenarios, highlighting the utility of high-density observations for the validation and improvement of conventional modeling frameworks. Taken together, this body of work demonstrates the promise of atmospheric observation networks based on moderate quality sensor technologies as a practicable approach to the challenge of monitoring urban greenhouse gas emission patterns and trends.