Non-parametric methods of estimation of type A uncertainty of the environmental noise hazard indices

Abstract

A control of environmental noise hazards requires estimation of uncertainty of noise indices LDEN, LN. Assessment of the type A standard uncertainty in measurement results - expressed as the standard deviation of the mean, calculated the most often at the assumption of a normal distribution - is significant for the process. Such assumption - in relation to the noise measurement results - is of a relatively low likelihood. Thus, there is a need of looking for non-standard procedures of the standard deviation estimation of the mean of results, without any information of belonging to a certain class of distribution. The aim of the hereby paper is an indication of the possibility of using non-parametric estimators of a density function in the calculation process of the type A standard uncertainty of environmental noise hazard indices. An attention was directed towards kernel estimators. The origin of their application, advantages and the method of constructing was described on the basis of a continuous monitoring of a traffic noise recorded on one of the main arteries of Kraków in 2004 and 2005. Usefulness of three forms of estimators, it means: kernel, unbiased and of maximum likelihood, was analysed. Keywords:acoustic monitoring of environment, analysis of the results, type A standard uncertainty in measurements, kernel estimator.