Determining the source regions for surface to stratosphere transport: A retro-transport approach

The transport of `very-short lived' species (i.e., chlorinated and brominated hydrocarbons, VSLS hereafter) from the planetary surface to the stratosphere is well-known to have a significant impact on stratospheric ozone. Several studies [e.g. Levine et al. 2007, Pisso et al. 2010] have shown that the amount of chlorine and bromine reaching the stratosphere is strongly dependent upon the geographical location of the source. To investigate this geographical dependence carefully involves simultaneously determining the sensitivity of a single quantity, such as the flux of a species into the stratosphere, upon emissions from many possible source locations. It therefore makes sense to adopt an adjoint approach. Such an approach allows the following question to be addressed: can the average efficiency of surface-to-stratosphere transport for all surface locations be meaningfully compared in an objective manner? Here we present results, in the form of sensitivity maps, for transport of finite lifetime species from the atmospheric boundary layer to the stratosphere, to answer the above question. The sensitivity maps are calculated using a newly-developed retro-transport model [Hourdin and Talagrand 2006]. The new model serves as a highly-accurate adjoint to a standard chemistry transport model with full convective and boundary layer parametrizations. It is based on a numerical formulation that is extremely close to that of the forward model, with attendant advantages in terms of ease-of-use and stability of model solutions. Results are validated by comparison with those obtained via repeated solution of the forward model, with the adjoint calculation requiring only a small fraction of the computational expense. The accompanying map shows the sensitivity of tracer mass within a receptor region located close to the extratropical tropopause (6.5-9.5 km in height, demarcated by the white rectangle) over a ten day period in January 2005, to surface sources across the northern hemisphere. To demonstrate the accuracy of the retro-transport results, the map has been calculated in two ways. The left middle panel shows a blow-up of the results from a single integration of the retro-transport model. The left lower panel shows the same results from 429 separate integrations of the forward model, each of which has an isolated source at a different location. The spatial pattern of the absolute and relative error (max. 2x 10^-7) is shown on the right. The seasonal and inter-annual variations of the sensitivity maps are explored, as well as their dependence on the lifetime of the species under investigation. We also examine the effect of convection and turbulent mixing (both of which are easily incorporated within the computational framework) upon the details of the transport. While the application here is to surface to stratosphere transport, it is to be emphasized that the computational techniques demonstrated are applicable to a wide range of problems in atmospheric transport, particularly where it is desirable to compare the impact of a large number of possible source sites. References: Hourdin, F. and Talagrand, O. 2006: Q.J.R. Meteorol. Soc., 132, 567–583, doi:10.1256/qj.03.198.A Levine, J. G., Braesicke, P., Harris, N. R. P., Savage, N. H., and Pyle, J.A. 2007: J. Geophys. Res., 112, D04308, doi:10.1029/2005JD006940 Pisso, I., Haynes, P. H., and Law, K. S. 2010: Atmos. Chem. Phys., 10, 12025-12036, doi:10.5194/acp-10-12025-2010