A tropospheric chemical-transport model: Development and validation of themodel transport schemes

We describe a 3D off-line chemical-transport model (CTM) for studies of tro pospheric chemistry. The tropospheric CTM is a development of the existing TOMCAT stratospheric CTM to which we have added a parametrized convection s cheme and treatment of vertical diffusion. The CTM diagnoses the occurrence of convection during the model simulations from the forcing analysis field s of winds, temperature and humidity. The spatial distributions of the vertical mass fluxes calculated by these s chemes appear realistic. The convective mass fluxes agree well with values derived from global analyses, having maximum values at low latitudes and at the intertropical convergence zone, although the TOMCAT cloud-top height m ay be too low. We have used the model's convection scheme to diagnose the convective preci pitation when forced by European Centre for Medium-Range Weather Forecasts (ECMWF) initialized analyses. Despite the off-line approach, the CTM-diagno sed convective precipitation is in very good spatial agreement with global observations and output from a general-circulation model (GCM), showing tha t not only does convective venting of tracers occur in the correct location s, but also that the convection scheme can be coupled to wet-deposition sch emes in full chemistry simulations. TOMCAT exhibits a low-latitude bias in cloud amount, due to the model's neglect of mid-level convection, and under estimates convective precipitation and cloud amount, the latter by at least 20% compared with GCM results. This underestimate may be due to excessive stability of the ECMWF analyses used in the off-line model or to difference s in the formulation of convection in the ECMWF forecast model and in TOMCA T. We have evaluated the CTM's transport processes by performing a range of si mulations with a surface-emitted radon (Rn) tracer. The agreement with rado n observations improves with an increase in horizontal resolution from 7.5 degrees x 7.5 degrees to 2.8 degrees x 2.8 degrees and with the inclusion o f vertical diffusion and convection in the model formulation. When run at a horizontal resolution of 2.8 degrees x 2.8 degrees, TOMCAT captures the tr acer transport and seasonal evolution in tracer transport associated with m ajor meteorological features and with convective processes. TOMCAT reproduc es the observed seasonality in radon at surface sites, but overestimates th e surface radon concentration, partly due to insufficient vertical diffusio n in the model. However, transient systems and their tracer transport withi n the boundary layer are simulated realistically. Compared with observation s, the modelled radon profiles are too 'C-shaped' (the modelled profile has a mid-tropospheric minimum which is not present in the observations) which may be caused by the model's neglect of convective downdraughts and/or ent rainment and detrainment rates that are too low in the cloud column, result ing in insufficient mixing within the cloud column. The radon concentration in the upper troposphere is underestimated by TOMCAT; this may be due to a number of factors: the underestimate in convective-cloud amount, the way i n which the base mass flux is calculated, the limitation of organized entra inment to subcloud levels, and a possible static stability of the analyses.