Simulation of the transport of Rn222 using on-line and off-line global models at different horizontal resolutions:: a detailed comparison with measurements

The short-lived radionuclide Rn-222 is emitted at a fairly constant rate from the continents and is a good surrogate for studying the transport of "air pollution" from polluted continental areas to clean, remote regions. The large concentration gradients of 2-3 orders of magnitude which exist between the continents and the remote atmosphere present a major challenge to the modelling of horizontal and vertical atmospheric transport. We use the global off-line tracer transport model TM3 at 3 different resolutions. Input to the model consists of meteorological data for the year 1993 obtained from the European Centre for Medium Range Weather Forecasts (ECMWF). The same meteorological data is used to constrain the climate model ECHAM4-T42-L19. Using these meteorological data, Rn-222 simulations are used to evaluate and document model performance and associated uncertainties. High time-resolution measurements made at 2 continental stations, 2 stations under continental influence and 4 remote sites, and aircraft measurements obtained during the NARE aircraft campaign are used for a detailed comparison. Although in specific regions there are inter-model differences of up to a factor of 2 in the calculated boundary layer concentrations, these differences are not translated into a better performance of either model for the stations used for comparison. We generally obtain high correlations of model results and measurements; these range from r = 0.6-0.8 for the continental and coastal stations and 0.5-0.6 for the remote sites. Calculated mean concentrations and corresponding standard deviations generally agree favourably with observations, lending credibility to the usefulness of our models for evaluating transport of air pollutants from continental sources to remote regions. The main cause of model deviations is probably related to uncertainties in the meteorological input data set provided by the ECMWF model and to a lesser extent by our knowledge of the spatial distribution of Rn-222 emissions and uncertainties involving sub-grid scale parameterisation of vertical transport, e.g., diffusion and convection.