Convective distribution of dust over the Arabian Peninsula: the impact of model resolution
Abstract. Along the coasts of the Arabian Peninsula, convective dust storms are a considerable source of mineral dust to the atmosphere. Reliable predictions of convective dust events are necessary to determine their effects on air quality, visibility, and the radiation budget. In this study, the Weather Research and Forecasting Model coupled with Chemistry (WRF-Chem) is used to simulate a 2016 summertime dust event over the Arabian Peninsula and examine the variability in dust fields and associated vertical transport due to the choice of convective parameterization and convection-permitting versus parameterized convection. Simulations are run at 45 and 15 km grid spacing with multiple cumulus parameterizations, and are compared to a 3 km simulation that permits explicit dry and moist convective processes. Five separate cumulus parameterizations at 15 km grid spacing were tested to quantify the spread across different parameterizations. Finally, the impact these variations have on radiation, specifically aerosol heating rates is also investigated. On average, in these simulations the convection-permitting case produces higher quantities of dust than the parameterized cases in terms of dust uplift potential, vertical dust concentrations, and vertical dust fluxes. Major drivers of this discrepancy between the simulations stem from the convection-permitting case exhibiting higher surface wind speeds during convective activity; lower dust emission wind threshold velocities due to drier soil; and more frequent, stronger vertical velocities which transport dust aloft and increase the atmospheric lifetime of these particles. For aerosol heating rates in the lowest levels, the shortwave effect prevails in the convection-permitting case with a net cooling effect, whereas a longwave net warming effect is present in the parameterized cases. The spread in dust concentrations across cumulus parameterizations at the same grid resolution (15 km) is an order of magnitude lower than the impact of moving from parameterized towards explicit convection. We conclude that tuning dust emissions in coarse-resolution simulations can only improve the results to first-order and cannot fully rectify the discrepancies originating from disparities in the representation of convective dust transport.