Inverse modeling of Asian dust was performed using MODIS coarse-mode aerosol optical thickness (AOT) and a regional adjoint dust model. The a posteriori AOT shows better agreement with the MODIS AOT and OMI Aerosol index (AI). Independent validation with the NIES Lidar network shows that the inversion leads to a significant improvement and reduces root mean squares of differences by 29-32% at Seoul, Matsue, and Toyama. The inversion results show considerably greater dust emissions over the Gobi Desert and Mongolia. Especially, optimized emissions are estimated as 8.1 Tg (352% of the a priori) for 29 March and 19.5 Tg (325%) for 30 March. Comparison with inversion results using NIES Lidar (Yumimoto et al. 2008) shows consistency with our a posteriori dust emissions which underscores the feasibility of dust inversion with satellite optical observations, and encourages combined inversion that assimilates the Lidar extinction coefficient (vertical profiles) and MODIS AOT (horizontal distribution) concurrently. However, the a posteriori results might reflect effects of air pollutants from highly industrialized regions, which might contaminate the assimilated results. Further quality control of the input for the data assimilation is necessary in future studies.