Ishac N., Bullen R.

Noise propagation is significantly affected by prevailing meteorological conditions, leading to variations in received noise levels. Several standard modelling methods rely on measured meteorological data and estimation techniques. Rather than accept the uncertainty of modelling methods, we decided to obtain realistic and actual noise level data including the effect of atmospheric conditions by conducting a year long experiment on sound propagation. Loud speakers were placed at a central location on a site, to be used as an artificial sound source. A constant sound signal of a set of pure tones with varying sound intensity levels between each frequency is triggered every hour, for one minute, twenty four hours per day and for a year. The primary frequencies in the source signal were chosen to adequately simulate the main frequency range of typical mechanical plan. The transmitter consists of a CD player with a CD containing the source noise, a timer to trigger playback, a power amplifier and loud speakers. The arrangement is powered by solar panels and housed in a wire mesh compound. The sound bursts were recorded by loggers at distant off-site locations, as well as at an intermediate position near the speakers. Each logger contains a calibrated measurement microphone and a Digital Audio Tape (DAT) recorder. The loggers are pre programmed to record the received noise levels. Meteorological data is continuously collected by a weather station. Each monitor’s DAT was then analysed using narrow band spectral analysis to filter the discrete pure tones from the ambient noise recorded. In the first instance the fluctuation of sound at each monitoring site is quantified. The meteorological and noise data is correlated and analysed to quantify the effects of weather on noise propagation. The results are encouraging with significant differences in noise levels being recorded, attributed to weather influences.