Expected mean in an environmental noise measurement and its related uncertainty

Abstract

In the context of the implementation of the Environmental Noise Directive 2002/49/EC a study on noise measurement uncertainty was developed. In performing any noise measurement average, there might be assumptions over the distribution and independency of the samples. In a concrete example dealing with environmental noise, this might be the case of a series of measurements of a constant noise source like an industrial plant or a fluctuating noise source like road traffic. Using a series of 1 day Leq in the first case, or a series of 15 minutes samples in the second case, the average of these values is usually considered as the expected mean, however, the error caused by the specific selection of the samples is not evaluated. Statistically speaking, before establishing an average value, at least the lognormal distribution of the samples and the effect of adding-up several uncertainties should be evaluated. This article will focus on the formulas to be used and will discuss differences in assessing the expected mean for normally distributed values, or for lognormally distributed, and will suggest an approach to properly add-up all uncertainties related to an environmental noise measurement campaign.