As an important tracer and forcing agent of climate change, the variability of dust aerosol abundance is due partly to the emission strength and change of source areas in different climates. This variability feeds back to the simulated meteorology through direct dust-radiation interactions, and indirect dust-cloud interactions, which, in turn, affect the dust emission, forming an emission-meteorology feedback loop. Whether the feedback is positive or negative and how it evolves in response to climate change remain open questions. Crucial factors in untangling these questions include optical properties (i.e., the single scattering albedo) and the emission strength of dust, which show strong spatiotemporal variations. Modeling dust as mineral components in Earth System Models (ESMs), initialized by a soil atlas that contains size-resolved mineral (e.g., iron oxides) fractions, can potentially help reproduce variation of those factors. But existing atlases contain large uncertainties and may not be applicable to a climate that has expended dust sources. In this case, the Earth Surface Mineral Dust Source Investigation (EMIT) could be particularly useful, as it will provide a more accurate characterization of mineralogy not only for soils readily erodible in present day but also for soils exposed within at-risk lands bordering present-day arid and semi-arid sources. For the emission strength, a scheme that represents reasonable sensitivity to changes in factors that affect the emission (i.e., soil moisture) is critical to projecting it in a changing climate. This study tests the performance of a physically-based dust emission scheme in response to climate change, projects the dust-radiation interaction in different climates, and quantifies the emission-meteorology feedback loop in a fully coupled ESM with speciated dust.