Estimating the model-specific uncertainty of aircraft noise calculations

Aircraft noise contours are estimated with model calculations. Due to their impact, e.g., on land use planning, calculations need to be highly accurate, but their uncertainty usually remains unaccounted for. The objective of this study was therefore to quantify the uncertainty of calculated average equivalent continuous sound levels (L-Aeq) of complex scenarios such as yearly air operations, and to establish uncertainty maps. The methodology was developed for the simulation program FLULA2. In a first step, the partial uncertainties of modelling the aircraft as a sound source and of modelling sound propagation were quantified as a function of aircraft type and distance between aircraft and receiver. Then, these uncertainties were combined for individual flights to obtain the uncertainty of the single event level (L-AE) at a specified receiver grid. The average L-Aeq of a scenario results from the combination of the L-AE of many single flights, each of which has its individual uncertainties. In a last step, the uncertainties of all L-AE were therefore combined to the uncertainty of the L-Aeq, accounting also for uncertainties of the number of movements and of prognoses. Uncertainty estimations of FLULA2 calculations for Zurich and Geneva airports revealed that the standard uncertainty of the L-Aeq ranges from 0.5 dB (day) to 1.0 dB (night) for past-time scenarios when using radar data as input, and from 1.0 dB (day) to 1.3 dB (night) for future scenarios, in areas where L-Aeq >= 53 dB (day) and LAeq >= 43 dB (night), respectively. Different uncertainty values may result for other models and/or airports, depending on the model sophistication, traffic input data, available sound source data, and airport peculiarities such as the specific aircraft fleet or prevailing departure and arrival procedures. The methodology, while established for FLULA2 on Zurich and Geneva airports, may be applied to other models and/or airports, but the partial uncertainties have to be specifically re-established to account for individual models and underlying sound source data. (C) 2014 Elsevier Ltd. All rights reserved.