Abstract :
Sixty four noise level recordings were carried out at building facades in the residential areas of Hong Kong in the present study. These records were then categorized into groups of different statistical structures using the Pearson frequency curve system. The effects of these statistical structures on the accuracy of predicting the statistical parameters of a record formed by the addition between any two noise records of similar statistical structures were examined. Results suggest that the currently adopted logarithmic addition tends to over-estimate the noise climates for bell-shaped noise level distributions, but it may also produce under-estimation for J-shaped ones. The worst deficiency of the logarithmic procedure appears for Pearson Type I(M) noise records which are commonly found in urban areas. Results also manifest the importance of noise level statistical structures in the noise impact assessments.
