Chemical transport models (CTM) are used extensively in air quality and carbon cycle studies to simulate abundances of chemical species in the atmosphere. CTMs are driven by assimilated meteorological fields (off-line transport) from general circulation models (on-line transport) which significantly reduces the computational time. Metfields are further degraded to run the model at different grids with different horizontal and vertical resolution. Manipulations with off-line meteorology as well as the use of off-line versus on-line transport reduced accuracy of simulation and introduces systematic biases in the model. One such example is reduced vertical transport and excessive diffusion at low horizontal resolution. These biases have significant negative implications when CTMs are used to infer sources and sinks of atmospheric species.
We use GEOS-Chem CTM and evaluate the impact of model biases on the global methane simulation. We compare it to GOSAT XCH4 retrieval and find that at low (4x5 degrees) resolution a significant portion of discrepancies between GEOS-Chem and GOSAT is explained by biases in transport rather than in surface emissions. Later, we investigate the sensitivity of GOSAT XCH4 retrievals to different types of model biases and show that they can be diagnosed and partly mitigated using Weak Constraint 4D-Var data assimilation approach.