High-resolution quantification of methane emissions from satellites

Abstract

Global high-resolution observations of methane concentrations from satellites can improve our understanding of methane emissions through inverse analyses, but require understanding the information content of observations that are often heterogeneous in time and space. This work develops and applies tools to use satellite observations to quantify continent-scale methane emissions and the associated information content at high resolution. Chapter 1 addresses the computational challenge. Analytical solution of the inverse problem provides closed-form characterization of the error statistics and information content associated with the optimized emissions but is computationally expensive due to the need to construct the Jacobian matrix that relates emissions to atmospheric concentrations. We propose two methods to reduce this cost. The reduced-dimension method generates a multiscale grid that preserves high resolution where the satellite provides information content and goes to coarser resolution elsewhere. The reduced-rank method constructs the Jacobian matrix along the dominant directions of information content so that the inversion optimizes emissions where the satellite provides a constraint and defaults to the initial emission estimate elsewhere. We apply these methods to an inversion of Greenhouse Gases Observing Satellite (GOSAT) methane data with augmented information content over North America in July 2009, demonstrating their ability to reproduce the standard solution at a fraction of the computational cost. Chapter 2 applies the reduced-rank Jacobian method to an inversion of observations from the Tropospheric Monitoring Instrument (TROPOMI) to infer methane emissions at 0.25° × 0.3125° (≈25 × 25 km2) resolution over the contiguous U.S. (CONUS) for 2019. Our optimal (posterior) estimate of anthropogenic emissions in CONUS is 30.9 (30.0 - 31.8) Tg a−1, where the values in parentheses give the spread of an eight-member inversion ensemble. This is a 13% increase from the 2023 GHGI estimate for CONUS of 27.3 (24.6 - 30.0) Tg a−1 for 2019, where the values in parentheses give the 95% confidence interval. Relative to the GHGI, we find the largest increase 51% for landfills. We find a large median 77% increase in landfill methane emission estimates reported by 73 facilities to the EPA’s Greenhouse Gas Reporting Program (GHGRP), a key data source for the GHGI, which we attribute to overestimated recovery efficiencies at landfill gas recovery facilities and to underestimated emissions from operational changes and leaks. We also quantify emissions in the 48 states in CONUS, which we compare to the GHGI’s new state-level inventories. Our posterior emissions are on average 34% larger than the 2022 GHGI in the largest 10 methane-producing states, with the biggest upward adjustments in states with large oil and gas emissions. We finally calculate emissions for 95 geographically diverse urban areas in CONUS, where we find posterior emissions of 6.0 (5.4 - 6.7) Tg a−1, equivalent to a fifth of CONUS anthropogenic emissions. Urban area emissions increase on average by 39 (27 - 52) % compared to a spatially allocated version of the 2023 GHGI. We attribute the discrepancy to underestimated landfill and gas distribution emissions. The large upward corrections to the GHGI at all scales found here may present challenges for climate policies and goals, many of which target methane emission reductions. More generally, this work demonstrates the potential to quantify high resolution greenhouse gas fluxes on continent and global scales, improving our ability to mitigate emissions.