To this date, estimates of the climate response to mineral dust remain largely uncertain because of our limited capability to quantify dust distribution in the atmosphere. Focusing on the Central and East Asian dust source regions, this thesis aims to develop a coupled regional dust modeling system to provide an improved modeling capability of atmospheric dust as well as to aid the integration of ground-based and satellite observations. The objectives of this study are as follows: 1) evaluate the capabilities of the available data to detect and quantify mineral dust in the atmosphere; 2) develop and test a coupled regional dust modeling system able to simulate size resolved dust concentrations accounting for the regional specifics of Central and East Asia; and 3) outline a methodology for data and modeling integration. The capabilities of ground-based and satellite data to characterize dust in the atmosphere are examined in great details. Based on analysis of MODIS data reflectance and radiances, we found evidence for regional signature of dust in near-IR and proposed a new probabilistic dust-cloud mask that explicitly takes into account the spatial variability characteristics of dust aerosols. We developed a coupled regional dust modeling system (WRF-DuMo) by incorporating a dust emission module (DuMo) into the NCAR WRF model. The WRF-DuMo unique capabilities include explicit treatment of land surface properties in Central and East Asia, a suite of dust emission schemes with different levels of complexity, multiple options for dust injection in the atmosphere and flexible parameters of the initial size distribution of emitted dust. Two representative dust events that originated in East Asia in the springs of 2001 and 2007 have been modeled with WRF-DuMo. Simulations with different initial size distribution of dust, injection and emission parameterizations have been performed to investigate their relative role on the modeled dust fields. We performed an integrated analysis of modeled dust fields and satellite observations by introducing an ensemble model dust index, which used in conjunction with satellite dust retrievals improves the capability to characterize dust fields. Finally, we provide recommendations for the development of an integrated observational and modeling dust framework.