A Bayesian framework for deriving sector-based methane emissions from top-down fluxes

Atmospheric methane observations are used to test methane emission inventories as the sum of emissions must be balanced by the change of observed methane concentrations. Typically, concentrations are inversely projected to a net flux through an atmospheric chemistry-transport model. However attribution of flux to the underlying sector-based emissions is challenged by prior assumption about the emission distribution, spatial resolution and correlated uncertainties between fluxes, and model chemistry and transport error. Here we show a Bayesian optimal estimation method that projects these methane fluxes to sector-based emissions on an arbitrarily sized grid, while accounting for these inverse estimate characteristics. We apply this method to 2010-2015 satellite-derived fluxes over the U.S. and show how we can explicitly characterize a posterior sector-based emission budget based on fluxes. We also apply this approach to two different emissions estimates over the Permian Basin, between 2010-2015 to 2018-2019, and can robustly conclude that increased gas emissions drive the observed changes in fluxes, consistent with activity data. This approach provides robust comparisons between top-down estimates of emissions from different observing systems, critical for assessing the efficacy of policies or global agreements intended to reduce emissions using the ever increasing number of satellite observations.