Intrinsic variability due to micrometeorological effects and/or ground effects, measurement uncertainty and model uncertainty are the main sources of spreading of the parameters influencing outdoor sound propagation. Thus spreading associated to outdoor SPL is a complex combination of deterministic, stochastic and epistemic uncertainties, and can be quantified thanks to a probabilistic process. This statistical process is presented in this paper and is called Calibration Under Uncertainty (CUU). Quantitative uncertainty assessment involves a pre-existing physical system to be studied, input data which can be measured or derived from measurements, and a sufficient amount of available (experimental and/or numerical) data with an eventual human expertise. CUU couples information from experimental and modelled data taking into account their own uncertainties (measurements errors, lack of knowledge on physical behavior, etc.)under specific assumptions. Quantify the global uncertainty on SPL, rank or apportion the contributions of influent parameters to a given output quantity of interest, compare experimental and effective parameters, and more generally understand the whole input-output structure are the main tasks of such a statistical method. CUU process has been applied to more or less complex cases using a large experimental set of data (Lannemezan 2005 (F)). An application to near ground sound propagation has been first led to understand the relative influence of ground parameters. A more complex case considering large distances and including micrometeorological effects has also been fulfilled with promising results which are presented in this paper.