Errors and improvements in the use of archived meteorological data for chemical...

Yu, K., C. A. Keller, D. J. Jacob, A. M. Molod, S. D. Eastham, and M. S. Long (2018), Errors and improvements in the use of archived meteorological data for chemical transport modeling: an analysis using GEOS-Chem v11-01 driven by GEOS-5 meteorology, Geosci. Model. Dev., 11, 305-319, doi:10.5194/gmd-11-305-2018.
Abstract: 

Global simulations of atmospheric chemistry are commonly conducted with off-line chemical transport models (CTMs) driven by archived meteorological data from general circulation models (GCMs). The off-line approach has the advantages of simplicity and expediency, but it incurs errors due to temporal averaging in the meteorological archive and the inability to reproduce the GCM transport algorithms exactly. The CTM simulation is also often conducted at coarser grid resolution than the parent GCM. Here we investigate this cascade of CTM errors by using 222 Rn–210 Pb–7 Be chemical tracer simulations off-line in the GEOS-Chem CTM at rectilinear 0.25◦ × 0.3125◦ (≈ 25 km) and 2◦ × 2.5◦ (≈ 200 km) resolutions and online in the parent GEOS-5 GCM at cubed-sphere c360 (≈ 25 km) and c48 (≈ 200 km) horizontal resolutions. The c360 GEOS-5 GCM meteorological archive, updated every 3 h and remapped to 0.25◦ × 0.3125◦ , is the standard operational product generated by the NASA Global Modeling and Assimilation Office (GMAO) and used as input by GEOS-Chem. We find that the GEOS-Chem 222 Rn simulation at native 0.25◦ × 0.3125◦ resolution is affected by vertical transport errors of up to 20 % relative to the GEOS-5 c360 online simulation, in part due to loss of transient organized vertical motions in the GCM (resolved convection) that are temporally averaged out in the 3 h meteorological archive. There is also significant error caused by operational remapping of the meteorological archive from a cubed-sphere to a rectilinear grid. Decreasing the GEOS-Chem resolution from 0.25◦ × 0.3125◦ to 2◦ × 2.5◦ induces further weakening of vertical transport as transient vertical motions are averaged out spatially and temporally. The resulting 222 Rn concentrations simulated by the coarse-resolution GEOS-Chem are overestimated by up to 40 % in surface air relative to the online c360 simulations and underestimated by up to 40 % in the upper troposphere, while the tropospheric lifetimes of 210 Pb and 7 Be against aerosol deposition are affected by 5–10 %. The lost vertical transport in the coarse-resolution GEOS-Chem simulation can be partly restored by recomputing the convective mass fluxes at the appropriate resolution to replace the archived convective mass fluxes and by correcting for bias in the spatial averaging of boundary layer mixing depths.

PDF of Publication: 
Download from publisher's website.
Research Program: 
Modeling Analysis and Prediction Program (MAP)