Modeling of dust emission for a crusted surface

Dust storms are frequent phenomena in the southwestern United Sates. Dust source areas in the region are often (partly) crusted. A critical prerequisite in dust aerosol modeling is an accurate representation of dust emission. While several dust emission schemes have been developed over the last decades, their applicability for crusted surfaces is not well tested. In this study, we use and test the applicability of the dust emission scheme of Shao (2004) (S04), which estimates dust emission based on the soil volume removed by saltation particle impacts, to model dust emission from a crusted surface in New Mexico, USA, for three dust events in spring 2016. Detailed field data are available for these events which are used as scheme input (surface crust and vegetation fraction, friction velocity, minimally- and fully-dispersed particle-size distributions) and for evaluation (saltation flux and dust emission flux). Results show that the saltation flux modeled with the scheme of White (1979) was overestimated by three orders of magnitude. This is expected as the supply of particles available for saltation is limited at the site. As our focus is on dust emission, a constant scaling factor was applied to match modeled and observed saltation fluxes. Parameters that describe the efficiency of saltator impacts to emit dust and the degree of dispersion during erosion need to be adapted in the S04 scheme to represent the soil surface setting at the study site. Our results show that changing those parameters has little effect on the modeled dust emission and dust emission is generally underestimated when PSDs of the top 1 cm soil layer are used as it is common. The reason for this is that the crust at the site is relatively thin and the soil overall sandy, which results in only a small difference between the two PSDs. If, however, the minimally- and fully-dispersed PSDs are replaced with the PSDs of, respectively, loose erodible material and crust, then the difference increases and the scheme parameters can be well adapted to give accurate estimates of dust emission. Our tests show that the scheme is able to model dust emission from a crusted surface provided that sufficient input information is provided. Further tests are required in the future to investigate the parameter variability for different crusted surfaces.