Characterizing model errors in chemical transport modeling of methane: Impact of model resolution in versions v9-02 of GEOS-Chem and v35j of its adjoint model

Abstract

The GEOS-Chem simulation of atmospheric CH4 was evaluated against observations from the Thermal and Near Infrared Sensor for Carbon Observations Fourier Transform Spectrometer (TANSO-FTS) on the Greenhouse Gases Observing Satellite (GOSAT), the Atmospheric Chemistry Experiment Fourier Transform Spectrometer (ACE-FTS), and the Total Carbon Column Observing Network (TCCON). We focused on the model simulations at the 4° × 5° and 2° ×2:5° horizontal resolutions for the period of February- May 2010. Compared to the GOSAT, TCCON, and ACEFTS data, we found that the 2°×2:5° model produced a better simulation of CH4, with smaller biases and a higher correlation to the independent data. We found large resolutiondependent differences such as a latitude-dependent XCH4 bias, with higher column abundances of CH4 at high latitudes and lower abundances at low latitudes at the 4° ×5° resolution than at 2° ×2:5°.We also found large differences in CH4 column abundances between the two resolutions over major source regions such as China. These differences resulted in up to 30% differences in inferred regional CH4 emission estimates from the two model resolutions. We performed several experiments using 222Rn, 7Be, and CH4 to determine the origins of the resolution-dependent errors. The results suggested that the major source of the latitude-dependent errors is excessive mixing in the upper troposphere and lower stratosphere, including mixing at the edge of the polar vortex, which is pronounced at the 4° ×5° resolution. At the coarser resolution, there is weakened vertical transport in the troposphere at midlatitudes to high latitudes due to the loss of sub-grid tracer eddy mass flux in the storm track regions. The vertical air mass fluxes are calculated in the model from the degraded coarse-resolution wind fields and the model does not conserve the air mass flux between model resolutions; as a result, the low resolution does not fully capture the vertical transport. This produces significant localized discrepancies, such as much greater CH4 abundances in the lower troposphere over China at 4° ×5° than at 2° ×2:5°. Although we found that the CH4 simulation is significantly better at 2°×2:5° than at 4°×5°, biases may still be present at 2°×2:5° resolution. Their importance, particularly in regards to inverse modeling of CH4 emissions, should be evaluated in future studies using online transport in the native general circulation model as a benchmark simulation.