Abstract. This study is a follow up of Sow et al. (2009) who had used the gradient method to determine the size-resolved emission flux of 3 different erosion events monitored in natural conditions at the Banizoumbou (Niger) supersite of the African Monsoon Multidisciplinary Analysis (AMMA) experiment. Our aim is to compare these measured fluxes with the predictions of the Dust Production Model (DPM) of Alfaro and Gomes (2001), which was derived from wind tunnel experiments. For each event, the model is run using the soil aggregate dry size-distribution, soil roughness length, and wind friction velocities derived from the field measurements as input parameters. We find that the mass emission flux is correctly predicted by the model if the binding energies of the 3 populations of fine particles released by sandblasting are reduced by a factor varying between 2.5 – for the intense convective event – and 5 – for the two less energetic events of the monsoon type. We explain this need to reduce the binding energies by an underestimation of the wind velocity due to the averaging over periods of 15' required by the calculation of the wind friction velocity. In all the studied cases the emission flux can, as already assumed in the DPM, be considered as a mixture of 3 fine particle populations the proportions of which depend on the intensity of the event. However, if the geometric mean diameter (2.0 μm) of the finest population compares well to the one used in the model (1.7 μm), the values of the intermediate and coarse modes (5.0 and 10.3 μm, respectively) are smaller than previously assumed (6.7 and 14.2 μm). Finally, the amplitude of the emission does not increase with wind speed for the coarsest mode, contrary what the DPM predicts. This suggests that the scheme describing the rate at which the relative proportions of the 3 populations evolve should be revised in the model.
Homogeneized modeling of mineral dust emissions over Europe and Africa using the CHIMERE model