Estimating regional nitrous oxide emissions using isotopic ratio observations and a Bayesian inverse framework

Abstract

Atmospheric nitrous oxide (N₂O) significantly impacts Earth's climate due to its dual role as an inert potent greenhouse gas in the troposphere and as a reactive source of ozone-destroying nitrogen oxides in the stratosphere. Global atmospheric concentrations of N₂O, produced by natural and anthropogenic processes, continue to rise due to increases in emissions linked to human activity. The understanding of the impact of this gas is incomplete as there remain significant uncertainties in its global budget. The experiment described in this thesis, in which a global chemical transport model (MOZART-4), a fine-scale regional Lagrangian model (NAME), and new high-frequency atmospheric observations are combined, shows that uncertainty in N₂O emissions estimates can be reduced in areas with continuous monitoring of N₂O mole fraction and site-specific isotopic ratios. Due to unique heavy-atom (15N and 18O) isotopic substitutions made by different N₂O sources, the measurement of N₂O isotopic ratios in ambient air can help identify the distribution and magnitude of distinct sources. The new Stheno-TILDAS continuous wave laser spectroscopy instrument developed at MIT, recently installed at the Mace Head Atmospheric Research Station in western Ireland, can produce high-frequency timelines of atmospheric N₂O isotopic ratios that can be compared to contemporaneous trends in correlative trace gas mole fractions and NAME-based statistical distributions of the origin of air sampled at the station. This combination leads to apportionment of the relative contribution from five major N₂O sectors in the European region (agriculture, oceans, natural soils, industry, and biomass burning) plus well-mixed air transported from long distances to the atmospheric N₂O measured at Mace Head. Bayesian inverse modeling methods that compare N₂O mole fraction and isotopic ratio observations at Mace Head and at Diibendorf, Switzerland to simulated conditions produced using NAME and MOZART-4 lead to an optimized set of source-specific N₂O emissions estimates in the NAME Europe domain. Notably, this inverse modeling experiment leads to a significant decrease in uncertainty in summertime emissions for the four largest sectors in Europe, and shows that industrial and agricultural N₂O emissions in Europe are underestimated in inventories such as EDGAR v4.3.2. This experiment sets up future work that will be able to help constrain global estimates of N₂O emissions once additional isotopic observations are made in other global locations and integrated into the NAME-MOZART inverse modeling framework described in this thesis.