Simulation of a Dust‐And‐Rain Event Across the Red Sea Using WRF‐Chem

Abstract

We present simulations of a series of dust‐and‐rain events in July/August 2020 that caused significant loss of life and property over northeast Africa and the Red Sea coast using convection‐permitting Weather Research and Forecasting‐Chem simulations at 1.5 km horizontal grid spacing coupled with prognostic dust and aerosol scheme together with a double‐moment cloud microphysical scheme. The model generally captures the optical properties and vertical profiles of dust but has difficulty reproducing the seasonal dust plumes that originate over northeast Africa and reach the western Arabian Peninsula through the Tokar gap. Overlapping of synoptic events with local, diurnal‐scale, convective features caused continuous rainfall over the study domain in July/August 2020. The model well simulates the synoptically‐driven clouds over the ocean (∼4–8 km) and daytime deep convective clouds over the land but cannot reproduce cumulus clouds over the ocean advected from land, which leads to some biases in simulating peak rainfall rates. Although our simulations do not explicitly consider aerosol particles as ice nuclei, the model reproduces the cirrus clouds found at ∼10–15 km height during the study period. Simulations with a planetary boundary layer (PBL) scheme better represented the diurnal profile of surface temperature and winds, as well as the rainfall distribution, than without using a PBL scheme. Dust generally enhanced rainfall during the synoptic rain event through indirect effects but suppressed rainfall during sea‐breeze‐induced convective events through dust direct effects by weakening the sea‐breeze circulation.

Recent and projected changes in climate patterns in the Middle East and North Africa (MENA) region

“…5 b) indicating the CMIP6 MMEM does not capture well the timing and spatial extent of these events, most likely because of an incorrect representation of the mid-latitude baroclinic systems in the colder months, as noted before and seen in the inter-model spread (Supplementary Fig. 5b), and the failure to replicate the necessary ingredients for summertime convection and its spatial coverage 137 , 151 , 160 . The normalized error variance is mostly in the range 0.8–1, Fig.…”