Central Asia’s worsening dust storms, driven by three major expanding desert source regions, could nearly double by 2100 under high-emission scenarios. Our analysis shows that these storms cool the upper atmosphere while enhancing low-level atmospheric heating, locally reducing surface sunlight by up to 20%—enough to threaten crop productivity. In spring, dust storms near Kashgar heat the lower atmosphere up to 30 times faster than at less dust-affected sites such as Lake Issyk-Kul.

This paper provides substantial consistent updates to the atmospheric boundary layer schemes of the AROME model, yet they can be used for both forecasting and climate modelling. The study employs a single-column model versus large eddy simulations comparison and uses a machine learning tool to calibrate parameterizations. The model's ability to simulate shallow clouds has been enhanced, especially for shallow precipitating cumulus and stratocumulus clouds.