A validation of the dBht as a measure of the behavioural and auditory effects of underwater noise

Project: RDCZ/011/0004

J Nedwell, A W H Turnpenny*

J Lovell***Jacobs Babtie Aquatic** Plymouth University

Subacoustech Report No. 534 R 1305

Introduction• Effects of underwater

noise include lethal and physical injury, auditory effects (avoidance and deafness)

• Behavioural effects occur at relatively low level and hence can affect much larger areas than physical injury, can have secondary lethal effects e.g. stranding.

• Behavioural effects related to perceived loudness.

The dBht (Species)

• Frequency weighting technique for determining the level of sound relative to hearing threshold (ht), hence dBht.

• Generalisation of dB(A) used for human behavioural effects and noise induced deafness; can be applied to any species where hearing sensitivity (audiogram) for species known.

• Sound passed through filter that mimics the hearing ability of the species; level of sound is measured after the filter; dBht (Species) proposes effect of the sound for any species primarily related to this perceived level (“loudness”).

Some history….

• Concept of dBht grew out of work on hearing of divers

• Led to concept of audiogram based metric (generalisation of dB(A)

• Now in use for 10 years

Concept of dBht

Project aims and objectives

Aim:

To validate the dBht as an objective metric for measuring the behavioural and auditory effects of sound on marine animals and implement as “Species Sound Level Meter”.

Programme:

• Formation of an open standard for the dBht(Species).

• Report on noise sources and classification into bands of noise pollution.– Nedwell J.R. and Edwards B. A review of measurements of underwater man-made noise carried

out by Subacoustech Ltd, 1993 – 2003. Subacoustech Report ref: 534R010929 September 2004

• Controlled laboratory tests.

• The provision of dBht(Species) sound level meters

• Implementation within industry

Controlled tests

Programme:

1. Re-interpretation of existing data.

2. Level vs. reaction – Calibrated measurements of noise and reaction to establish the dBht levels at which an avoidance response is evoked.

3. Audiograms – Pressure and particle velocity hearing thresholds required for experimental fish species; measured using the Auditory Brainstem Response (ABR) technique.

Re-interpretation of existing data: Doel

Measurements of efficiency

Species Common name Efficiency (%)

Clupea harengis Herring 94.7

Sprattus sprattus Sprat 87.9

Dicentrarchus labrax Bass 75.6

Osmerus eperlanus Smelt 53.5

Perca fluviatilis Perch 51.2

Solea solea Dover sole 46.6

Pomatoschistus sp. Goby 46.1

Abramis bjoerkna Bream 40.1

Platichthys flesus Flounder 37.7

The efficiency (percentage reduction in catch) in the cooling water of the Doel nuclear power plant; measurements on c. 350,000 fish by Leuven University. Signal swept sine from 20 Hz to 500 Hz.

Re-interpretation of existing data: INHS

• Bighead Carp barrier experiments at Illinois Natural History Survey’s Wolff Hatchery in Illinois USA; same signal as Doel

• Number of successful and unsuccessful attempts to pass recorded continually over 6 hour period.

• Found to be 57% effective

• Good quality audiograms available; sound had maximum level of 55.2 dBht (Hypophthalmichthys molitrix)

Re-interpretation of existing data: Marine mammals

Effect

(dB re.1 uPa)

Estimated dBht level Source

Porpoise excluded within 10-15 km of piling

89 dBht (Phocoena phocoena) at 10 km

Tougard et al: Study of piling for offshore windfarm at Horns’ reef

No Ringed Seals observed within 46 m of piling

83 dBht (Phoca hispida) @ 46 m (based on common seal audiogram)

Blackwell: Study of tolerance of seals to pile driving

Tolerance of seismic @ 180-190 dB lin by Ringed Seals

93-103 dBht (Phoca hispida) tolerated (based on common seal audiogram)

Harris et al; Tolerance of seals to seismics in Prudhoe Bay region

TTS in bottlenose dolphin octave band white noise, 30-50 minutes

98 dBht (Tursiops truncatus) Nachtigall et al 2003, 2004

Pool tests on captive dolphins

TTS in bottlenose dolphin, 3 kHz tones, 1 – 8 secs

133 - 135 dBht (Tursiops truncatus)

Finneran and Carder 2005

Pool tests on captive dolphins

Divers “very severe” 500 – 2500 Hz @ 157 dB lin.

87-90 dBht (Homo sapiens; submerged)

Fothergill et al. Study of diver aversion to low frequency underwater sound

Re-interpretation of dataReaction vs dBht level; free water

Open water results

• PV results not available, dBht calculated on pressure alone

• Significant size of Doel data set (350,000 fish) yields good reliability of data

• Note difference in results from lab tests – “hide” reaction of flatfish will yield positive result

• Indicates clear increase of reaction with increasing dBht level

• Indicates 90 dBht as “strong reaction” level, confirmed by other experimental evidence

• May be possible to measure audiograms and make use of more results from Doel (Sprat, prawns etc.)

Laboratory tests: Reaction trialsLaboratory tests: Reaction trials

CCTV

Laptop

Amplifier

Oscilloscope

Attenuator

Sound projector

Choicechamber

Camera

Sound projector

Circular tank

CCTV

Laptop

Amplifier

Oscilloscope

Attenuator

Sound projector

Choicechamber

Camera

Sound projector

Circular tank

Block diagram of experimental set-upBlock diagram of experimental set-up

Choice chamber and sound projector set-upChoice chamber and sound projector set-up

Laboratory tests: Reaction trials• ‘Choice-chamber’; 5 fish

per species x 7 sounds; two transducers outside mesh cage drive fish alternately one side to another.

• Avoidance response defined as move into opposite half of cage

• dBht calculated for pressure and particle velocity, pressure dominated in calibrated results

Laboratory tests: Reaction trials

•100 % avoidance (movement away from transducer)

•but only 60% effectiveness

Lab tests

• Reaction tests have been performed on 7 species; goldfish, 5-bearded rockling, dab, flounder, sand-smelt, smelt and pout

• Good-quality pressure and particle velocity audiograms only available for goldfish, dab and pout (hence only this calibrated data presented);

• Other audiograms affected by artefacts due to experimental conditions, also due to sensitivity to handling (esp. sand smelt and smelt). We are in process of retesting with improved handling methods.

A typical reaction for Flounder

Behavioural Reactions of Flounder to Sound Signal Ping 400(Represented as dBht Peak)

Sound Level (dBht Peak)

-20 -10 0 10 20 30 40 50 60

Ave

rage

% s

hift

to n

on-e

nerg

ised

sid

e af

ter

15 s

ec

-0.5

0.0

0.5

1.0

• Flounder instinct is to hide - react to noise by staying motionless on the bottom, hence no avoidance reaction

• The dBht offers a criteria for the likelihood of a behavioural reaction, but requires biological interpretation as to the nature and significance of the reaction

Reaction vs dBht level for swept sine

Difference of open water and lab results

• Significant differences between lab tests and open water results

• May in part result from different way in which reaction is measured in each.

• However, may also indicate that fish reactions in caged tests are suppressed; open water tests or observation are therefore the preferred method.

• Red Funnel piling project 2003 studied caged brown trout, some surprise that there was no apparent reaction even at 25 metres from piling, may be similar effect.

Factors affecting reactions

• Physiological state of fish• Different reaction behaviours, e.g. move away,

move close to bottom etc.)• ‘Follow-my-leader' effect, i.e. they all follow or stick

by one sensitive or insensitive leader (found in other types of fish test,e.g. temperature: Clough & Turnpenny 2002)

• Balance of pressure to move away from source with pressure to stay away from cage walls and other members of shoal

Typical reaction (for Sand Smelt)

Behavioural Reaction of Sand Smelt to Sound Signal Ping 400(Represented as dBht Peak)

Sound Level (dBht Peak)

-20 -10 0 10 20 30 40 50 60

Ave

rage

%sh

ift to

non

-ene

rgis

ed s

ide

afte

r 15

sec

.

-20

-10

0

10

20

30

40

In this case, statistically significant reaction at 15 dBht.

All calibrated results Pout, Goldfish and Dab dBht Peak

Sound Level (dBht Peak)

0 20 40 60 80 100 120

Ave

rag

e %

sh

ift to

no

n-en

erg

ised

sid

e a

fte

r 15

se

c

-40

-20

0

20

40

60

80

PPing 400 vs Col 2 PPing 100 vs Col 4 PNOISE MOD 3 vs Col 6 PNOISE vs Col 8 PFGS vs Col 10 PDAMPED 50 vs Col 12 PDAMPED 300 vs Col 14 PPing 400 vs Col 16 PPing 100 vs Col 18 PNOISE MOD 3 vs Col 20 GNOISE vs Col 22 GFGS vs Col 24 GDAMPED 50 vs Col 26 GDAMPED 300 vs Col 28 D Ping 400 vs Col 30 D Ping 100 vs Col 32 D Noisemod 3 vs Col 34 D Noise vs Col 36 D FGS vs Col 38 D Damped 50 vs Col 40 D Damped 300 vs Col 42 Regression

Goldfish, Dab and Pout dBht RMS 1s av

Sound Level (dBht RMS 1s av)

-20 0 20 40 60 80 100

Ave

rage

% s

hift

to n

on-e

nerg

ised

sid

e af

ter

15

sec

-40

-20

0

20

40

60

80

GPing 400 vs Col 2 GPing 100 vs Col 4 GNoiseMod 3 vs Col 6 GNOISE vs Col 8 GFGS vs Col 10 GDamped 50 vs Col 12 GDamped 300 vs Col 14 DPing 400 vs Col 16 DPing 100 vs Col 18 DNOISEMOD 3 vs Col 20 DNOISE vs Col 22 DFGS vs Col 24 DDamped 50 vs Col 26 DDamped 300 vs Col 28 PPing 400 vs Col 30 PPing 100 vs Col 32 PNoisemod 3 vs Col 34 PNoise vs Col 36 PFGS vs Col 38 PDamped 50 vs Col 40 PDamped 300 vs Col 42

Regression

RegressionsBehavioural Reactions ofGoldfish (solid), Dab (dashed), and Pout (dotted) dBht Peak

Sound Signal (dBht Peak)

-20 0 20 40 60 80 100

Ave

rage

% s

hift

to

non-

ener

gise

d si

de a

fter

15

sec

-40

-20

0

20

40

60

80

Damped 300Damped 50FGSNoiseNoisemod 3Ping 100Ping 400

No significant differences between different sound types (damped 300 Hz and 50 Hz simusoid, swept sine, modulated and unmodulated noise and 100 Hz and 400 Hz “sonar pings”

Fit to reaction data

Results

• Results generally indicate region of no reaction for < 40 dBht

• > 40 dBht increased reaction with increasing level

• Peak dBht level collapses the data better than 1 second average level (perhaps not surprisingly)

• No significant difference in reaction to the different sounds

• Difficult to assign a unique fit to data

Data averaged over all signals in 5 dB bins

Behavioural Reactions of Goldfish, Dab and Pout to sound averaged at 5dB steps

Sound Level (dBht Peak)

0 20 40 60 80 100 120

Ave

rage

% s

hift

to n

on-e

nerg

ised

sid

e

-20

-10

0

10

20

30

40

50

G. dBht Peak Goldfish RegressionD. dBht Peak vs S. Shift DabRegressionP dBht Peak vs Av shift Pout Regression

• Some indication that Goldfish and Pout are reacting differently in lab tests

• May be specific to caged tests, no equivalent behaviour in open water results

Further deliverables

Standard for dBht. Have standard approach to constructing dBht filters using FIR approach (details already published in I of A conference).

• Implementation required by industry• Main issue is lack of suitable audiograms. In particular, they

typically do not span a sufficient frequency range and dynamic range. Those chosen are suitable for a limited range of applications.

Species Sound Level Meter. Have constructed a SSLM based on an FIR filter – can select species; can be made available as executable for implementation on modest laptop; user has to provide A/D card and hydrophone (c. £2500) Six sets available for partners.

• Method of facilitating takeup by industry/regulators?

Summary (1)Experimental results indicate:• In all cases, increasing reaction with perceived dBht(Species)

level. • But significant difference between lab scale and open water

tests.Free water reactions:• Good fit indicating response (avoidance) proportional to

dBht(Species) level• Indicates 90 dBht as “strong avoidance” (~100%) limitLab tests:• Peak dBht level best to collapse the data • No significant difference in reaction to the different sounds• No reaction for < 40 dBht

• > 40 dBht increased reaction with increasing level

Summary (2)

Results indicate:• Better audiograms required, including repeated audiogram

tests under different conditions/labs/methods to provide confidence in result

Benefits of metric:• A simple metric relating noise to its effect in an identical

manner to airborne noise criteria.

• Enables simple criteria (“the sound shall not exceed 90 dBht at the stated range for the specified species….”).

• Allow non-expert personnel (e.g. MMOs) to measure and interpret noise (Species Sound Level Meter)

Current status

Technical issues: • Dynamic range” - important or not? • No information on habituation - shortcoming of

validation

Political issues• Would like to see metric adopted as best practice,

usually gives “right” results, easy to understand and use, only validated metric, 10 years of use, vast body of data

• Has run into powerful political lobby - well heeled team in US, Subacoustech sidelined