Contribution to reducing uncertainties regarding the dry deposition velocities of fine aerosols: Study case of a prairie (description of experimental methods and comparison with models)

In the context of environmental protection, the transfer of aerosols to vegetative canopies must be properly estimated, since aerosols constitute a preferential transport pathway for atmospheric pollution (bacteriological, chemical or radioactive). In order to properly assess the transport of pollutants, it is essential to accurately determine their dry deposition fluxes. Previous studies conducted around a nuclear site showed that the deposition prediction models do not yield an accurate representation of measured radionuclide activity (mainly 106Ru) in grasses, with observed discrepancies probably attributable to an underestimate of the dry deposition velocities. In addition, the literature mentions uncertainties in the order of two decades for the deposition velocities of particles larger than 1 urn, and little experimental data is available for submicronic particles. In order to reduce these uncertainties and choose a suitable model, a series of experiments to measure the dry deposition velocities of aerosols has been scheduled for 2005-2007. This document presents the method used and the results obtained for a fine aerosol (0.24 μm) dispersed over a prairie under neutral or unstable meteorological conditions. The technique used consists of generating a monodispersed fluorescein aerosol. The fluorescein concentration is measured in the grass and air by spectrofluorimetry and used to obtain the dry deposition velocities. A second approach using natural radioactivity (214Pb) and gamma spectrophotometry measurements has enabled the confirmation of the results obtained. The mean deposition velocities measured in this first series of experiments for a prairie-type substrate vary between 4 10-4 and 2 10-3 m.s-1. The comparison with the Slinn, Zhang and Wesely models shows that the first of these underestimates the deposition velocities by a factor of 5 to 20 and the other two more accurately represent the experimental measurements, with model/measurement discrepancies of generally less than 2.