The sound propagation model supports acoustic noise pollution studies. Since most of the noise sources are located outdoors, an acoustical-meteorological coupled simulation is required. However, meteorological parameters are always uncertain and can affect model output. In order to ensure that proper conclusions will be obtained from the simulation, an uncertainty analysis should be carried out. However, such an analysis is still rarely applied in the research field. Therefore, this thesis addresses the question of which approach is suitable for analyzing the uncertainty and sensitivity of a particular acoustic-meteorological coupled model. In this study, the uncertainty of the Lagrangian-based acoustical-meteorological coupled model is analyzed.

This study focuses on the methods of Monte Carlo and the generalized polynomial chaos expansion, both of which make it possible to obtain the first statistical moment and the Sobol' sensitivity indices for uncertainty and sensitivity analysis, respectively. These methods are applied to acoustical-meteorological simulations with several hill topographies and also with several variations of meteorological parameters. The results are compared and it is observed that both methods can be applied. The generalized polynomial chaos expansion method produces similar results to the Monte Carlo method but with lower computational costs. The results also show the same pattern as other studies on the meteorological effects on sound propagation for specific test cases.