| Abstrakt: |
As a band of noise propagates through the atmosphere, atmospheric absorption attenuates the higher frequencies in the band more rapidly than the lower frequencies. For large propagation distances, atmospheric absorption shapes the spectrum such that most of the acoustic energy is in the lower part of the band. When this occurs, the effective atmospheric absorption coefficient for this band of noise is less than when the energy is evenly distributed over the band. When a real filter is used to examine the received spectrum, the problem becomes even more complex since the filter does not cut off completely at the nominal lower frequency limit for the band. The equations governing the attenuation and spectrum analysis of noise under these conditions have been derived and a numerical integration process has been used to obtain quantitative results. It is shown that predicting observed values of atmospheric absorption loss for bands of noise involves two correction factors which are added to single-frequency absorption loss to define the effective attenuation for the band of noise. The first correction factor ΔI accounts for the difference between the attenuation predicted over a given path in a homogeneous atmosphere for a pure tone at the geometric center frequency of an ideal filter band, and the actual attenuation over the same path for the same ideal band of noise. The second correction factor ΔF accounts for the difference between the band level measured at one point with a practical filter with finite transmission outside the nominal filter pass band and the true band level measured at the same point with an ideal (rectangular) filter. Representative calculations show that the sum of these corrections become substantial when the product of distance, in km, and the square of frequency, in kHz, exceeds about 4 to 40 for octave and one-third octave band filters, respectively. Although the net effect of such corrections will generally be small for overall frequency weighted-noise levels (e.g., perceived noise level), the errors in diagnosing spectral detail in noise signatures of distant sources, such as aircraft, can be important. [ABSTRACT FROM AUTHOR] |
