@article {Wilson:2025:0736-2935:849,
title = "A perspective from two decades of analyzing the error bars on outdoor sound propagation predictions",
journal = "INTER-NOISE and NOISE-CON Congress and Conference Proceedings",
parent_itemid = "infobike://ince/incecp",
publishercode ="ince",
year = "2025",
volume = "271",
number = "2",
publication date ="2025-07-25T00:00:00",
pages = "849-858",
itemtype = "ARTICLE",
issn = "0736-2935",
url = "https://ince.publisher.ingentaconnect.com/content/ince/incecp/2025/00000271/00000002/art00087",
doi = "doi:10.3397/NC_2025_0151",
author = "Wilson, Keith Keith",
abstract = "The complexity of outdoor sound propagation raises many predictive challenges. Errors can be conceptualized in two categories, namely aleatory (due to random chance) and epistemic (due to incomplete knowledge). Examples of aleatory uncertainties include scattering of sound by vegetation
and atmospheric turbulence. Often, we can formulate reasonable statistical models for the scattering, e.g., the Rayleigh fading model. Epistemic uncertainties, which can be more difficult to assess, include imperfect knowledge of the acoustical properties of buildings, roadways, ground surfaces,
and the atmosphere. With respect to the atmospheric effects, by employing high-resolution turbulence simulations, it was shown that, even with accurate weather data from a location and time very close to the propagation path, sound level predictions have inherent random errors of about 6-8
dB (Wilson, D. K., Lewis, M. S., Weatherly, J. W. and Andreas, E. L. Noise Control Eng. J. 56(6), 465-477 (2008)). Fortunately, the errors do diminish when averaging is performed in time or over frequency bands. Monte Carlo simulations are particularly useful for characterizing the impacts
of uncertainties. The idea is to create a large ensemble of random realizations of the environment, make an acoustic prediction for each, and then find statistics for the ensemble.",
}