Observations and cloud-resolving modeling of haboob dust storms over the Arabian Peninsula

Strong mesoscale haboob dust storms in April 2007 in the central Arabian Peninsula were studied using the cloud-resolving Weather Research and Forecasting-Chemistry (WRF-Chem) modeling system and observations collected during an intensive atmospheric field campaign. The field campaign provided the valuable aircraft and Doppler weather radar measurements. Active convection persisted for several days during the study period. Dust generation was caused by both strong large-scale winds and locally produced density currents. Because of insufficient spatial resolution, the event was not resolved accurately by the conventional reanalyses. However, the WRF-Chem model did successfully capture the primary features of the convection, its location, and precipitation patterns. Although the amount of rainfall in the model was slightly underestimated compared to the satellite measurements, it was approximately double the rainfall in the reanalysis. The convection-associated dust outbreaks were simulated well, with the aerosols optical depth magnitude and the temporal variability being in good agreement with both the ground-based and satellite aerosol retrievals. The model captured the major dust generation patterns, transport pathways, and several of the largest haboobs identified from the satellite observations. About 25 Tg of dust was emitted in the Arabian Peninsula during the 10-day period. Approximately 40% of the locally deposited dust was subject to wet removal processes. During periods of high local dust production, the WRF-Chem model underestimated the PM10 mass concentration (associated mostly with dust particles larger than 3 mu m in diameter) by nearly a factor of 2. This suggests that the current dust parameterizations, which prescribe the size distribution of the emitted dust, underestimate the number of large particles that increases at strong wind conditions. Plain Language Summary In this study, we use a sophisticated numerical model of atmospheric circulation with an aerosol component to simulate a series of local-scale haboob dust storms that occurred in the central Arabian Peninsula during April 2007. This type of dust storm is associated with a cold air outflow from thunderstorms and is specific to the wet spring season in this region. We compare the model results with the data obtained from aircraft and meteorological radar during a field campaign conducted at that time. The high-resolution model produces good results that capture the atmospheric convection, rainfall features, and major dust outbreaks recorded at a ground station in Riyadh. We demonstrate that it performs better than other global reanalysis products. However, our results show that the aerosol component of the numerical model needs to be improved, as the model underestimates the concentration of coarse dust particles. This is caused by the uncertainties introduced when prescribing the physical properties of dust aerosols generated from the surface.