Sound production and associated behaviours of benthic

Sound production and associated behaviours of benthic invertebrates from a
coastal habitat in the Northeast Atlantic
Laura Coquereau a,*, Jacques Grall b, Laurent Chauvaud a,b, Cédric Gervaise c,d, Jacques Clavier a,
Aurélie Jolivet a,e, Lucia Di Iorio c
a
Université de Bretagne Occidentale, Institut Universitaire Européen de la Mer, Laboratoire des
Sciences de l’Environnement Marin, UMR 6539, Rue Dumont D’Urville, 29280 Plouzané, France
Observatoire Marin, UMS 3113, Institut Universitaire Européen de la Mer, Rue Dumont D’Urville,
29280 Plouzané, France
b
c
Chaire CHORUS, Fondation Grenoble INP, 46 avenue Felix Viallet, 38031 Grenoble, France
d
GIPSA-LAB, 11 rue des Mathématiques, Grenoble Campus, 38402 Saint Martin d'Hères, France
e
TBM environnement/Somme, 115 rue Claude Chappe, Technopole Brest Iroise, 29280 Plouzané,
France
[email protected]
Estimation of peak frequency errors induced by tank reverberation
Numerical simulations were performed to estimate the peak frequency errors resulting from
reverberations in the tank. Tank dimensions as well as minimal and maximal distances of the
sound source to the hydrophone were used to simulate the experimental settings of the tank
experiments.
The test signal was a short tonal pulse of duration equivalent to the ones of the data. The
frequency of the pulse was varied from 2 kHz to 50 kHz to span the entire peak frequency (fp)
intervals estimated from our data (fs = sampling frequency = 156250 Hz or 192000 Hz; Fig.
1). Simulations were performed for different sample lengths, the same used for feature
estimations in the tank recordings (32, 64, 128, 256 samples). Since the number of samples is
positively correlated with the transmission path, peak frequency errors were calculated for all
possible reverberations comprised between the minimal reverberation distance (20 cm, mean
value of the distance between the animal and the hydrophone) and the maximal transmission
path length for the given signal (e.g., 2.5 m for a 256-sample signal and a fs=156250 Hz). For
each reverberation within the total transmission path, we considered a transmission loss of
1/R, with R being the reverberation distance. Frequency errors were calculated as the
difference between the frequencies of the original signal and the signal after accounting for
reverberation (Fig. 1) :
𝑒(𝑡) = cos⁡(2𝜋𝑓0 𝑡) × 𝑤𝑇 (𝑡) emitted signal with frequency f0 , wT time window of duration T,
1
𝑟
𝑠(𝑡) = ∑0.2<𝑟<2.5 𝑟 𝑒(𝑡 − 𝑐 ) received signal with fp obtained by looking for the peak value of
the fft of s(t)
Resulting mean and maximal peak frequency errors are reported in Table 1. The mean fp error
is of the order of 1.5 kHz and the maximum fp error does not exceed 6 kHz. In all tested cases,
mean and maximum errors are smaller than the frequency resolution of the Fourier transform,
suggesting that reverberation does not significantly affect fp estimation.
Fig. 1 Example of simulation results. Top left: frequencies of signal without reverberations;
top right: map showing frequencies of signal with reverberation as a function of the distance
of the reverberating surface. Discontinuities indicate reverberation artefacts of within the
signal’s frequency band. Bottom left: Map of reverberation error (top right – top left). The
dark blue area represents the frequency with maximal error (error = 937.5 Hz). Parameters
used: fft = 128, sound source distance to hydrophone = 30.8 cm, sampling frequency =
192000
Table 1 Summary of simulation results of peak frequency (fp) errors (mean ± standard
deviation and maximum) for minimum and maximum distances between the sound source and
the hydrophone
min distance/ cm
fs / Hz
frequency resolution
/ Hz
fft
snapping
12
192000
6000
32
1466.2 ± 531
3937.5
32.22
633.8 ± 350
1875.0
moving
14
192000
6000
32
1674.4 ± 533
3937.5
33.68
616.7 ± 344
2062.5
jumping
swimming
swimming
moving
feeding
moving
feeding
moving
feeding
12
12
12
13
13
13
13
13
13
2441
6000
6000
6000
6000
6000
6000
6000
6000
64
32
32
32
32
32
32
32
32
32.22
32.22
32.22
32.94
32.94
32.94
32.94
32.94
32.94
10
6000 / 4883
32
3937.5 / 1678.5
30.81
type 3 (feeding)
type 2
10
10
750
1500
256
128
172.9 ± 94.9
1466.2 ± 531
1466.2 ± 531
1574.9 ± 53
1574.9 ± 53
1574.9 ± 53
1574.9 ± 53
1574.9 ± 53
1574.9 ± 53
1239.4 ± 530 /
697.4 ± 264
40.5 ± 39
102.3 ± 90
457.8
3937.5
3937.5
3937.5
3937.5
3937.5
3937.5
3937.5
3937.5
type 1
156250
192000
192000
192000
192000
192000
192000
192000
192000
192000/
156250
192000
192000
187.5
375.0
30.81
30.81
Species
behaviour
Athanas nitescens
Crepidula fornicata
Mimachlamys varia
Pecten maximus
Echinus esculentus
Paracentrotus lividus
Psammechinus miliaris
Maja brachydactyla
mean fp error ±
max fp error /Hz max distance/ cm mean fp error/ Hz max fp error/ Hz
SD / Hz
409.1 ± 211
1220.7
633.8 ± 350
1875.0
633.8 ± 350
1875.0
625.9 ± 347
1875.0
625.9 ± 347
1875.0
625.9 ± 347
1875.0
625.9 ± 347
1875.0
625.9 ± 347
1875.0
625.9 ± 347
1875.0
650.9 ± 355 / 681.7
1875.0 ou 1373.3
± 364
55.9 ± 53
375.0
158.1 ± 122
937.5