Development of atmospheric dispersion simulation system based on the efficient use of LES model output

Large-eddy simulation (LES) models can evaluate complex flow fields and atmospheric dispersion process in urban areas with higher accuracy than conventional RANS based models. However, simulation in urban areas requires high spatio-temporal resolution, which causes the heavy computational load. In this study, we aim to build an atmospheric dispersion evaluation system which can utilize the LES model in low computational cost. The proposed system consists of pre-calculated LES output database and atmospheric dispersion model. To build the system, we discuss the method to use the LES database efficiently, including the reduced-order modelling based on the mode decomposition method. We also discuss the relationship between large-scale environment (e.g., wind speed, wind direction and atmospheric stability) and flow field in target area simulated by LES. The reduced-order model can simulate the flow field from certain number of dominant modes and large-scale environment, that makes the computational load much lower than full simulation. In addition, we investigate the accuracy of atmospheric dispersion models, Eularian and Lagrangian models, in case of using the LES database as input data. Combining the reduced-order model and atmospheric dispersion model, we propose the atmospheric dispersion evaluation framework which aimed to use widely and easily even in small computational resources.