Analysis and Ranking of the Acoustic Disturbance Potential of ...

Report No. 6945
BBN Systems and Technologies Corporatiop
Critical ratios tend to increase with increasing frequency. In the
bottlenose dolphin, a pure tone signal at 6 kHz must exceed spectrum level
noise by 22 dB to be detected, whereas a 70 kHz tone must exceed spectrum level
noise by about 40 dB (Fig. 2.28). Critical ratios for the bottlenose dolphin
have not been measured below 5 kHz. Burdin et al. (1973a) obtained some evidence
that, at 1-10 kHz, critical ratios of dolphins are lower (better) than those of a
human. Below 1 kHz though, the frequency discrimination abilities of the dolphin
deteriorate rapidly (Thompson and Herman 1975), and bottlenose dolphin critical
ratios mdy not closely resemble those of humans at low frequencies.
The critical ratios of the northern fur seal range from a low of 19 dB at 4
kHz to 27 dB at 32 kHz (Moore and Schusterman 1987). These values are a few
decibels lower than the critical ratios of the bottlenose dolphin at
corresponding frequencies (Fig. 2.28). In contrast, the ringed seal has critical
ratios about 10 dB higher than those of the fur seal and several dB above the
dolphin through the same frequency range (Terhune and Ronald 1975b; Fig. 2.28).
However, Moore and Schusterman (1987) suggest that the ringed seal values are
suspiciously high, and may be artefactual.
Critical ratios are not greatly different for underwater and aerial hearing,
or across a wide range of vertebrates (Fig. 2.28, Moore and Schusterman 1987).
The dolphin, fur seal and ringed seal data quoted above all represent underwater
hearing. In-air critical ratios have been determined for the harp seal (Terhune
and Ronald 1971) and the harbor seal (Renouf 1980). The validity of the harp seal
data, at least for frequencies up to 8 kHz, has been questioned (Moore and
Schusterman 1987). The in-air critical ratios for the harbor seal are generally
consistent with the underwater values for the fur seal and bottlenose dolphin
(Fig. 2.28).
Masking Bands. A pure tone is masked almost exclusively by noise at
frequencies near the frequency of the tone. Noise at frequencies outside of this
masking band has little influence on detection of the signal. The determination
of the width of the masking band has been the subject of much effort. Fletcher
(1940) proposed one method, based on the assumption that signal power must equal
total noise power in the masking band in order to be audible. Since the spectrum
level intensities of masking noise [dB re ( 1 ~P~)~/HZ] and the intensities of
tones (dB re 1 pPa) are not compatible units, the spectrum level of the masking
noise must be converted to a band level. The white noise often used in masking
experiments has a flat spectrum, and therefore the energy in a masking band of
no.ise is proportional to the masking bandwidth in Hz. Band level is computed
from spectrum level by the formula
BL = SL + 10 log BW (1
where BL represents band level, SL represents spectrum level, and BW equals the
bandwidth in Hz (Urick 1983). If it is assumed that signal power must equal or
exceed noise power in the masking band in order to be detectable (Fletcher 1940),
then the masking bandwidth is
BW = antilog CR/10 (2)