Evaluating Stratospheric Methane in GEOS-Chem with Satellite and Balloon Observations

Global inferences of methane emissions by inversion of satellite observations of atmospheric methane columns require a forward transport model to reliably relate methane emissions to the observed columns. Previous studies have reported large errors in model simulations of stratospheric methane that would propagate to yield biases in the inversions. Here we analyze the stratospheric methane simulation in the GEOS-Chem model, which is used extensively in the inversion of satellite data, using observations made by AirCore balloons and the MIPAS and ACE-FTS satellite instruments. Using a 10-year (2010-2019) full-chemistry simulation in GEOS-Chem we compute a 3-D monthly seasonal cycle of stratospheric methane total loss frequency implied by the losses of methane to oxidation by OH, O(¹D), and Cl atoms. We then use these loss frequencies to drive a multidecade GEOS-Chem methane simulation (1983-2020), with observed surface methane concentrations from the NOAA network as dynamic boundary conditions. The long simulation allows us to properly initialize the stratosphere for comparison to observations, and the use of a surface boundary condition to constrain tropospheric methane avoids propagating emissions-related uncertainties into the stratosphere. We show that the GEOS-Chem simulation at 2^ox2.5^o resolution is successful in reproducing the seasonal and meridional distributions of stratospheric methane. It can be used reliably in global inversions of satellite data, including with reduced vertical resolution in the middle/upper stratosphere and mesosphere. By contrast, a 4^ox5^o simulation shows biases at high latitudes that could plausibly propagate into inverted surface fluxes. Our long-term record of GEOS-Chem stratospheric methane provides unbiased initial conditions for global inversions, accounting for the time lag in stratospheric response to rising methane.