To assess the radiological dose to the public resulting from the Fukushima Daiichi Nuclear Power Station (FDNPS) accident in Japan, especially for the early phase of the accident when no measured data are available for that purpose, the spatial and temporal distributions of radioactive materials in the environment need to be reconstructed through computer simulations using the atmospheric transport, dispersion, and deposition model (ATDM). For the ATDM simulation, the source term of radioactive materials discharged into the atmosphere is essential and has been estimated in many studies. In the present study, we further refined the source term estimated in our previous study and improved the ATDM simulation with an optimization method based on Bayesian inference, which used various measurements such as air concentration, surface deposition, fallout, and newly released hourly air concentrations of 137Cs derived by analyzing suspended particulate matter (SPM) collected at air pollution monitoring stations. This optimization improved not only the source term but also the wind field in meteorological calculation, which led to the reduction of discrepancies in plume passage time at monitoring points to less than 3 h between calculations and measurements, by feeding back comparison results between the dispersion calculations and measurements of radionuclides. As a result, the total amounts of 137Cs and 131I by the present study became 1.0 × 1016 and 1.2 × 1017 Bq, respectively, and decreased by 29% and 20%, respectively, in comparison with those by previous study. The ATDM simulation successfully reproduced both the air concentrations at SPM monitoring points and surface depositions by airborne monitoring. FA10 for total samples of air concentrations of 137Cs at SPM monitoring points increased from 35.9% by the previous study to 47.3%. The deposition amount on the land decreased from 3.7 × 1015 Bq by the previous study to 2.1 × 1015 Bq, which was close to the measured amount of 2.4 × 1015 Bq. We also constructed the spatiotemporal distribution of some major radionuclides in the air and on the surface (optimized dispersion database) by using the optimized release rates and ATDM simulations. The optimized dispersion database can be used for comprehensive dose assessment in tandem with behavioral patterns of evacuees from the FDNPS accident by collaborating research group in the Japanese dose assessment project. The improvements in the present study lead to the refinement of the dose estimation.