Analysis of the Uncertainty of High-Speed Rail Noise Predictions

The effects of noise from future railway schemes are assessed using validated prediction methods that rely on multiple input parameters and assumptions. Generally, single values representing reasonable worst-case assumptions are selected for each input parameter. Variation in factors such as local meteorological conditions, receiver location and source characteristics means that there is uncertainty in the model inputs that, when propagated through the model, result in uncertainty in the model predictions. This paper presents an analysis to characterise and quantify the uncertainty of high-speed railway noise predictions via Monte Carlo simulation and global sensitivity analysis. The HS2 airborne noise prediction method is used as the basis of the study, along with elements of the CONCAWE method to take meteorological effects into account. Forty-eight generic scenarios representing different combinations of meteorological conditions, time of day, screening and receiver distances have been analysed for the L-Aeq and L-Amax metrics. Output distributions representing the prediction uncertainty have been calculated for each scenario and are presented as boxplots. The results of the global sensitivity indicate that meteorological effects have the largest contribution to the prediction uncertainty. The aerodynamic source term parameters were found to be the most sensitive parameters related to the source