AbstractWe developed a simulation code named WT, which calculates tephra fallout from eruption plumes bent by wind. This proposed model assumes a series of radial particle sources arraying along the theoretically predicted trajectory of the plume center. To validate WT, we reconstructed the tephra dispersal during the 2011 sub‐Plinian eruption of Shinmoedake in Japan. This eruption was ideal for the validation because it was observed through various approaches; however, high‐resolution data of the fluctuating plume height made it difficult to determine a representative eruption time and plume height for running the simulations. Also, the amount of particle segregation along the plume could not be determined a priori. We thus implemented inversion calculations to study the optimum particle segregation pattern for possible ranges of time and the plume height, and the misfit between the observed and calculated mass loadings on the ground was evaluated. After this process and the following analysis, the optimum eruption time was determined for one of three major explosions (18:00 Japan Standard Time on 26 January). The optimum plume height was estimated to be 4 km above sea level, slightly lower than the estimation derived by the weather satellite (5 km). When wind shear exists, the WT model has a significant advantage in reconstructing tephra dispersal over the classical code named Tephra2, the prototype of WT. The WT inversion implies a simple segregation model from a well‐mixed plume, and further studies on particle segregation patterns will help to improve the accuracy of forward WT simulations.