Data assimilation has been developed in numerical weather prediction (NWP) and modeling of oceanography. Recently, store and expansion of observations and development of numerical modeling have enabled data assimilation techniques to be applied to aerosol transport models. In this paper, we introduce information about applications (e.g., forecast, inverse modeling, reanalysis, sensitivity analysis) and recent studies about data assimilation with atmospheric aerosol observations and numerical models. We also show a preliminary experiment of ensemble-based data assimilation with global aerosol climate model (SPRINTARS) and Aerosol Optical Thickness (AOT) measured by MODIS/AQUA. In the experiment, the data assimilation improves under-estimates in East Asia, North Pacific Ocean, Central America, Middle East and Central Africa, and over-estimates in oceans over the southern hemisphere. Root mean square difference (RMSD) between SPRINTARS and MODIS AOT is reduced by 21 %, and long-wave aerosol direct forcing at the tropopause increased where dust and carbon aerosol are increased by the data assimilation.