Analysis of stochastic acoustical hazards in environment

Abstract

Acoustical assessment of environment, realized on the basis of changes observation held as part of national environment monitoring for the need of long term noise protection policy in environment, demands long term average noise level $L_{\textrm{DEN}}$ and $L_{N}$ estimation. The estimation is based on noise measurement results from all the days in the year, taking into account day, evening and night times. The basis for correct statistical estimation when the data is incomplete (time limited monitoring of acoustic environment) is a choice of time sample so that results gained reflect the mechanism of noise hazard changes that works throughout the year. It should provide identification of particular event occurring frequency and time relations between events, e.g. sound extreme value appearance or recognition of periodical changes of important parameters in uncertainty assessment. Such process demands assumption of particular statistical techniques related to the assumed model of noise level changes where the random factor is important and always present. That problem is part of the article. The paper presents basic analysis of monitored phenomenon by usage of models framing its stochastic volatility. Important elements of modeling are described, illustrated by examples of yearly noise data analysis gathered at one of main streets in Kraków. Attention is paid to the analysis problem with the process periodicity, measurement uncertainty related to the measurement conditions volatility or time uniformity of the monitored noise processes in view of its internal causality relations. They are loaded with random factor whether we assign randomness to noise source volatility, whole year measurement conduct inability or distortions related to proper sampling of given field of noise hazards.