Impact of the noise indicators components correlation [formula omitted] on the uncertainty of the long-term day–evening–night noise indicator [formula omitted

Abstract

The paper studies the actual impact of the noise indicators components on the Lden uncertainty. Various sampling methods in order to correctly determine the Lden uncertainty depending on the measurement sample size were also analysed. The analysis was conducted on the actual measurement data from three continuous noise monitoring stations in Gdańsk, Poland. The statistical inference was based on the results of t-test at significance level γ=0.05. The analysed data were comprehensive and differentiated and covered 3 years. The analysis indicated that the 95% intervals determined with the use of independent draw are narrower than the intervals calculated using dependent sampling. For small samples, n=5, the uncertainty evaluation method using independent draw without correlation has the highest coverage degree of actual value, and for samples n={10,15} it is the method using dependent draw with correlation. This method has also the highest coverage regardless of the acoustic signal variability.