Fróðskaparrit - 01.01.1992, Qupperneq 35

Testing acoustical methods for detection of
odontocete whales
Bertel Møhl
The problem
Stock sizes of most odontocete species are
hard to get. At present, whaling activities
are either banned or existing at such a small
scale that no usable data are generated for
the traditional methods of population esti-
mates. Therefore, sight surveys have now
become the dominating inventory method.
The detection rate of sight surveys, how-
ever, is low (assumed to be 5% in NASS
1989) and not easy to quantify. Sea state,
illumination and visibility are obviously
influencing detection rate; animal density,
diving behaviour and observer experience
are also known to be important. Clearly,
any measures that serve to increase detec-
tion rate will lead to better estimates.
Proposed solution
Since odontocetes are known as vocally
active animals, it has often been proposed
to detect them by acoustic means. Quite a
number of reports describe methods operat-
ing in the low audio frequency range
(Thomas etal., 1986). Such methods, how-
ever, require a silent ship, running at 4 knt
or less. For standard transects as in the
NASS- surveys, such speeds are not attrac-
tive. Further, the ultrasonic sonar pulses of
these animals are not utilised by audio
equipment. These pulses have recently
been found to be very powerful, with
source levels in the 220-230 dB re. 1 pPa
range (Au et al., 1988). As background
noise in the sea is decreasing with frequen-
cy, signal to noise ratio should improve at
ultrasonic frequencies.
To take advantage of ultrasonic pulses
for detection, they have to be converted to
audible frequencies. The only online tech-
nique available is heterodyning, where a
segment of the ultrasonic spectrum is con-
verted down by subtraction from a chosen,
fixed frequency. In essence, the process is a
filter for the spectrum segment in question.
For odontocete whistle signals, this can be
an efficient technique, yielding outputs
with high signal to noise ratios and there-
fore large detection ranges. For the broad
band sonar pulses, heterodyning is ineffi-
cient by filtering out only a small part of the
Fróðskaparrit 40. bók (1992): 39-43