A comprehensive inter-comparison of 5 atmospheric chemistry transport models (TM5, TM4, TM3, IMAGES, and LMDZ) has been performed. The main objective was to analyze differences in model transport, in particular vertical mixing (boundary layer and convective transport), synoptic variations, and large scale global circulation (including inter-hemispheric exchange and stratospheric tropospheric exchange (STE)). For this purpose simulations of various tracers with very different atmospheric lifetimes τ have been carried out: 222Rn (τ = 3.8 days), SF6 (τ = ~3000 years), and CH4 (τ = ∼9 years), using prescribed boundary conditions for all models. Furthermore, OH fields from various model simulations with full chemistry have been compared.
222Rn simulations show significant differences in vertical transport between models, leading to differences of simulated 222Rn concentrations near the surface of up to a factor of ~3. The TM5 and TM4 model have generally the highest 222Rn concentrations near the surface, while the other models tend to stronger vertical mixing. Comparison with in-situ measurements at 9 surface monitoring sites show that synoptic variations are simulated relatively well by all models which use (re)analyzed meteorological fields (i.e. all models except IMAGES, which is using monthly mean climatological fields). Comparison of TM5 and TM4 simulations (which have the same parameterization of atmospheric transport) illustrate that increasing horizontal model resolution significantly improves agreement with observations.
Simulations of SF6 show significant differences in inter-hemispheric transport between the applied models, ranging between 6 and 12 months. This range is consistent with previous model inter-comparisons, e. g. within TransCom2 [Denning et al., 1999]. STE is weaker and probably more realistic (15–16 months) in TM5, TM4, and LMDZ than in TM3 and IMAGES (7-8 months). The difference in STE between TM3 vs. TM5/TM4 is probably largely due to the different vertical resolution of the applied model versions.
CH4 tracer simulations with prescribed OH fields were performed for TM5, TM4 and IMAGES. Consistent with the 222Rn simulations, TM5 and TM4 show higher CH4 mixing ratios near the surface over CH4 source regions compared to IMAGES. Both TM5 and TM4 simulate synoptic variations very well at most surface monitoring sites. Similar as for the 222Rn experiments agreement with CH4 surface observations is improving with increasing horizontal model resolution. The large difference in STE between TM5/4 and IMAGES is also clearly reflected in the CH4 simulations.
Furthermore, OH distributions have been compared from model simulations with full chemistry. For these simulations the applied models (TM5, TM4, IMAGES) used different emission inventories (representing typical standard configurations of the corresponding models). Simulated OH fields show significant differences near the surface, probably largely due to the applied different emission inventories (CO, NMHC, NOx). In the free troposphere, however, the spatial OH distribution are relatively similar. In addition, also the seasonal OH variation is very consistent for all model runs. Global CH4 + OH lifetimes in the range of 8.3 - 11.4 years have been calculated for the different OH fields. All models suggest 20-40% higher CH4 lifetime in the SH, compared to NH.
P Bergamaschi, JF Meirink, et al.. Model inter-comparison on transport and chemistry – report on model inter-comparison performed within European Commission FP5 project EVERGREEN (Global satellite observation of greenhouse gas emissions)