A statistical description of ground-to-ground sound propagation

Sound propagation over moderate distances is significantly influenced by the conditions of the land surface and the prevailing weather. As is well known, sound velocity varies with weather conditions, primarily air temperature and wind velocity. Most research on sound propagation has focused on the prediction of individual event levels based on the conditions in effect at that time. Few have studied the distribution of events over long periods. Rather, they have concentrated on case studies and single-event predictive methods. This paper examines the statistical distributions of received sound data as functions of distance. The source was pink noise in the one-third octave bands ranging from 40 Hz to 10 kHz. One source height, two opposing directions, and several receiving distances ranging from 46 to 800 m are considered. The results show that the distributions of received sounds form a regular family of curves and that each can be closely bounded by a simple normal distribution, by fitting a straight line to the higher level portion of the distribution data when plotted on normal probability paper. The straight lines representing the higher level, when drawn on normal probability paper, are such that the point at which they cross 50 percent decays regularly with distance, and the slope to these lines regularly increase with distance and frequency. This paper also compares the statistics for 8-months of propagation west to east (somewhat the prevailing wind direction) with the same 8-months of propagation east to west (more often upwind). Also considered are wind conditions (strong downwind, downwind, crosswind,