Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - i - December 2014
EXECUTIVE SUMMARY
The Roberts Bank Terminal 2 Project (RBT2 or the Project) is a proposed new three-berth marine
terminal at Roberts Bank in Delta, B.C. The Project is part of PMV’s Container Capacity Improvement
Program, a long-term strategy to deliver projects to meet anticipated growth in demand for container
capacity to 2030. Studies described in this technical data report contribute to an understanding of the
environmental effects of the Project. As part of the proposed Project, underwater noise studies were
undertaken in the vicinity of Roberts Bank and in other areas of the Salish Sea. To measure transmission
loss (TL) and estimate source levels (SL) of vessels, acoustic measurements were collected for the Ship
Sound Signature Analysis Study in four locations: one at Roberts Bank near RBT2, two in Haro Strait,
and one in Juan de Fuca Strait south of Victoria. The objectives of the TL measurements were to: 1)
improve sound propagation models at multiple noise modelling locations; and 2) allow accurate estimates
of container ship and other ship SLs collected over multiple years by the Whale Museum and Beam
Reach (TWMBR) in Haro Strait. The objectives of the SL measurements were to: 1) generate descriptive
statistics of SLs of multiple ship size and type classes, including bulk carriers, cargo and container ships,
fishing and passenger vessels, tankers, tugs, and vehicle carriers; 2) estimate SLs of container ships and
tugs transiting at different speeds; and 3) estimate SLs of berthing activities at Deltaport Terminal.
Results of this study provide Project area and regional area-specific measurements to fill data gaps and
increase confidence in the construction and operational acoustic modelling to be completed for the
Project. Estimates of Project acoustic effects rely on accurate estimates of SLs and the site-dependent
TL, which attenuates these sounds.
One-third octave band TL values were measured for input into the acoustic propagation model.
Geoacoustic inversions were performed to estimate the acoustic parameters of seafloor layers that
influence acoustic TL, as these were unknown at each site. The inversion results indicate that the seabed
sediments are acoustically absorptive at all sites, and that the compressional wave speeds are consistent
with soft, unconsolidated silts and clays. When possible, the range-dependent acoustic model (RAM) was
used because of its suitability to low frequency propagation and its ability to take into account complex
bathymetry and sub-bottom geoacoustic properties. At distances <100 m, a wavenumber integration
propagation model was used. Site-specific TL measurements allowed for more accurate estimates of SL,
and measures of TL and estimates of SL allowed for more accurate site-dependent sound
propagation models (JASCO 2014a, b).
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - ii - December 2014
SLs of three large container ships representative of those currently using Roberts Bank terminals (i.e.,
>320 m in length) were estimated for transiting speed (~20 knots) and lower speed (~10 knots). Overall,
SLs were higher for ships transiting at ~20 knots (i.e., 206.0, 203.9, and 200.5 dB re 1 µPa at 1 m) versus
~10 knots (i.e., 198.2, 191.9, 187.9 dB re 1 µPa at 1 m).
SLs were also estimated from 5,993 ship transits through Haro Strait recorded off of Lime Kiln Point State
Park, Washington. A multiple linear mixed effects model was used to explore the relationships between
ship SL and ship class, width, and speed over ground (SOG). There was a significant positive relationship
between SL and SOG with a slope of 0.47 (95% confidence intervals (CI): 0.37 to 0.57) indicating that on
average, SL increases by almost half a dB for every one knot increase in speed. Likewise, there was a
positive relationship between ship width and SL such that SL increased by 0.10 dB (95% CI: 0.04 to 0.17)
for every 1 m increase in width. Container ships >320 m and bulk carriers >250 m had the highest SL with
means of 190.3 and 190.7 dB re 1µPa at 1m, respectively. In comparison, transiting fishing vessels and
tugs had some of the lowest SLs with means of 177.6 and 179.3 dB re 1µPa at 1m, respectively.
SLs were estimated for a harbour tug, the Seaspan Resolution for three scenarios: 1) transiting at
different speeds (i.e., 4.0, 7.5, and 12.0 knots); 2) performing three simulations of berthing activities (i.e.,
half power, full power, and accelerating); and 3) berthing a container ship at Deltaport Terminal (included
a second tug, the Seaspan Raven). SLs for the harbour tug Seaspan Resolution increased with speed
(i.e., 3.4 dB per 1 knot increase) with broadband SLs of 162.1, 171.3, and 189.1 dB re 1 µPa at 1 m for
the tug transiting at 4.0, 7.5, and 12.0 knots, respectively. Broadband SLs for berthing simulation activities
were 180.2, 199.7, and 191.7 dB re 1 µPa at 1 m for the half power, full power, and acceleration
scenarios, respectively. Average broadband SL for the harbour tugs, Seaspan Raven and Seaspan
Resolution, berthing the Vienna Express were 186.2 dB re 1 µPa at 1 m over the entire operation and
191.2 dB re 1 µPa at 1 m over the loudest period.
The SLs of 5,993 transits of ships in Haro Strait are consistent with other studies (Bassett et al. 2012;
McKenna et al. 2012); while SLs of the three large container ships are the maximum SLs measured for
container ships >320 m and are between 10 and 16 dB higher than the mean SL for this ship class. This
may be due to the reduced ability of the TWMBR hydrophone at measuring ship noise below 50 Hz,
where ship noise typically peaks. The increase in SL with speed for most ships is also consistent with
previous studies (McKenna et al. 2013).
The results of this study have provided more accurate inputs into the cumulative underwater noise study
(JASCO 2014a), which uses known locations of ships, ship SL (adjusted by ship class and ship speed),
and acoustic propagation models adjusted by site-dependent TL to model noise levels in the study area.
Noise levels are then in turn used to estimate Project operational noise effects on southern resident killer
whales (SMRU 2014).
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - iii - December 2014
ACRONYMS
AICc Akaike Information Criterion with correction for finite sample size
AIS Automatic Identification System
AMAR Autonomous Multichannel Acoustic Recorder
CI confidence interval
CPA closest point of approach
PSD power spectral density
RAM range-dependent acoustic model
RL received level
rms root-mean-square
SL source level
SNR signal to noise ratio
SPL sound pressure level
SOG speed over ground
TL transmission loss
TWMBR the Whale Museum and Beam Reach
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - iv - December 2014
TABLE OF CONTENTS
EXECUTIVE SUMMARY ............................................................................................................................... I
ACRONYMS ........................................................................................................................................... III
Figure 14 Calculated 1/3 Octave Band SLs for Seaspan Resolution Performing Berthing
Simulations at Half Power (Blue), Full Power (Red), and Accelerating (Orange)............. 23
Figure 15 Calculated 1/3 Octave Band Composite SLs for Seaspan Resolution and Seaspan Raven
Berthing the Vienna Express. SLs Were Calculated from RLs Averaged Over The
Loudest Section (2 minutes, Red Line, 1596 m Slant Range) and Over the Section of the
Tugs Pushing Easy to Full (30 minutes, Blue Line, 1607 m Slant Range) ....................... 24
List of Appendices
Appendix A Warner, A., C. O’Neill, A. McCrodan, H. Frouin-Mouy, J. Izett, and A. MacGillivray. 2014.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia: Transmission
Loss, Vessel Source Levels, and Ambient Measurements. JASCO Document 00659,
Version 3.0. Technical report by JASCO Applied Sciences for Hemmera.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 1 - December 2014
1.0 INTRODUCTION
This section provides project background information, an overview of the study, and study components
and major objectives.
1.1 PROJECT BACKGROUND
The Roberts Bank Terminal 2 Project (RBT2 or Project) is a proposed new three-berth marine terminal at
Roberts Bank in Delta, B.C. that could provide 2.4 million TEUs (twenty-foot equivalent unit containers) of
additional container capacity annually. The Project is part of Port Metro Vancouver’s Container Capacity
Improvement Program, a long-term strategy to deliver projects to meet anticipated growth in demand for
container capacity to 2030.
Port Metro Vancouver has retained Hemmera to undertake environmental studies related to the Project.
JASCO Applied Sciences (Canada) Ltd and SMRU Canada Ltd have been subcontracted by Hemmera to
conduct the Ship Sound Signature Analysis Study, the results of which provide Project area and regional
area-specific measurements to fill data gaps and increase confidence in the construction and operational
acoustic modelling.
1.2 SHIP SOUND SIGNATURE ANALYSIS OVERVIEW
A review of existing information and state of knowledge was completed for underwater ship noise to
identify key data gaps and areas of uncertainty related to Project activities. Several limitations of available
data and data gaps exist.
Ships generate sound in the water from propulsion and power generation. While research has been
conducted to measure ship source levels (SLs) and to understand how operational parameters (i.e.,
operational speed, propeller design, etc.) change ship SLs (Urick 1983; Ross 1976), many of these
studies have been conducted on older ship designs not relevant to Project activities. Recent studies on
newer ships more relevant to the Project had small sample sizes, used simplistic TL models, or did not
include all ship classes or activities needed for accurate estimates of Project noise (Bassett et al. 2012;
McKenna et al. 2012, 2013); therefore, this study was conducted in Project-related areas to fill these data
gaps. Proxy measurements will still be needed for container ships and tugs that will be berthing at RBT2
because no SL measurements are available for the larger Mærsk E-class, and tug designs for RBT2 are
not available.
This chapter summarises study findings for key components identified from this gap analysis and that
will inform operational and regional noise modelling (JASCO 2014a) for the Project including:
Sound propagation properties at Roberts Bank from the RBT2 site; and
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 2 - December 2014
Sound levels from:
▫ current and future container ships traveling at different speeds;
▫ other ship classes using the study area;
▫ tugs travelling at different speeds; and
▫ ships during berthing.
An acoustic measurement program was undertaken during May and June 2013 to address data gaps
related to underwater noise within the general RBT2 area. These acoustic measurements in relation to
ship sound signatures had the following objectives:
Determine the most appropriate sound propagation models to use within the study area;
Measure SLs of container ships and tugs transiting at different speeds; and
Measure SLs of berthing container ships.
More details on the methods and results of the acoustic measurement program are provided in
Appendix A.
Study components, major objectives, and a brief overview are provided in Table 1.
Table 1 Ship Sound Signature Analysis Study Components and Major Objectives
Component Major Objective Brief Overview
1) Transmission loss (TL) measurements
To improve the accuracy and site-specificity of sound propagation models in the Project and larger regional area.
To allow for accurate estimates of ship SLs collected over multiple years by the Whale Museum and Beam Reach (TWMBR).
Wide-band (0.3 to 30 kHz) TL data were collected at Roberts Bank, Haro Strait, Lime Kiln, and the Victoria pilot site.
2) Source level (SL) estimates
To generate descriptive statistics of SLs of container ships and other ship classes
To measure SLs of container ships and tugs transiting at different speeds
To measure SLs of berthing container ship activities at Deltaport Terminal
Vessels transiting Haro Strait were recorded over a 26 month period and SLs were estimated
High-speed and low-speed SLs for three container ships were measured in Haro Strait and at the Victoria pilot site.
Multiple SLs for a harbour tug were measured at Roberts Bank.
Berthing SLs were measured at Deltaport Terminal.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 3 - December 2014
1.2.1 Transmission Loss (TL)
TL is a measure of how sound levels diminish between a source and receiver over distance. TL depends
on the frequency of the sound and the physical environment, including water sound speed profile,
bathymetry, and sub-bottom geoacoustic properties. TL is calculated from source and received levels
according to the equation:
RLSLTL
where SL is the source level (dB re 1 µPa at 1 m) and RL is the received sound pressure level or SPL (dB
re 1 µPa) , and TL is the transmission loss (dB re 1 m). Conducting TL measurements in an area provides
site-specific data on sound propagation and allows for more accurate SL measurements and noise
modelling.
1.2.2 Source Level (SL)
SL is a measure of the intensity of sound that a source emits at a standard reference distance of 1 m. For
point sources, such as a small transducer, SLs can be measured directly with a hydrophone at 1 m
distance. For larger sources, SLs must be determined indirectly by measuring RLs at larger distances and
back-propagating the levels to a reference distance of 1 m. For example, because ships radiate sound
from their hull and propeller, their SLs must be measured at a distance such that the TL from the different
points on the ship emitting sound is roughly the same. SLs are calculated by re-arranging the previous
equation to the following:
TLRLSL
1.3 STUDY AREA
Acoustic measurements for the Ship Sound Signature Analysis Study were collected in four locations
using Autonomous Multichannel Acoustic Recorders (AMARs): Roberts Bank near RBT2, in Haro Strait
(two recorders), and in Juan de Fuca Strait south of Victoria (Figure 1).
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 4 - December 2014
Figure 1 Study Area (AMAR Deployments). AMAR A = Roberts Bank, AMAR B = Lime Kiln, AMAR C and D = Haro Strait, AMAR E = Victoria, and AMAR F = Georgia Strait.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 5 - December 2014
2.0 TRANSMISSION LOSS: MEASUREMENTS AND MODEL
To validate sound propagation models, TL measurements were conducted at Roberts Bank, in Haro
Strait, and in Juan de Fuca Strait south of Victoria. TL was measured by playing underwater sounds from
a Lubell underwater speaker at several distances (200 to 5,000 m) from the AMARs and calculating the
difference between the RL on the AMAR and the SL measured by a hydrophone 1 m from the underwater
speaker. Detailed methodologies can be found in Appendix A.
2.1 TRANSMISSION LOSS MODELS
One-third octave band TL was required to determine SLs for the ships included in this study; however,
measured single-frequency TL at 1/3-octave band centre frequencies may not accurately represent TL for
an entire 1/3-octave band. TL at specific ranges, particularly at low frequencies, can be large due to
destructive interference between different propagation paths; therefore, single-frequency TL would not be
representative of 1/3-octave band TL if the measurement geometry aligned with these nulls. One-third
octave band TL values were calculated with an acoustic propagation model, more appropriate for lower
frequencies. The range-dependent acoustic model (RAM) was used because of its suitability to low
frequency propagation and its ability to take into account complex bathymetry and sub-bottom
geoacoustic properties, except at distances less than 100 m where a wavenumber integration model
(Jensen et al. 2000) was used.
Geoacoustic inversions were performed to estimate the acoustic parameters of seafloor layers that
influence acoustic TL, as these were unknown at each site. Table 2 lists the TL tracks used for each site,
and the model-measurement mismatch and geoacoustic properties that provided the best match to TL
measurements. The inversion results indicate that the seabed sediments are acoustically absorptive at all
sites, and that the compressional wave speeds are consistent with soft, unconsolidated silts and clays
(Table 2). Because the TL data were not very sensitive to sediment density and attenuation coefficient,
those parameters were less well-defined by the inversion; however, these uncertainties do not adversely
affect the model predictions, as water depth and the AMAR height above the seafloor were more
important parameters and better constrained by the inversion algorithm.
TL measurements on AMAR D (Figure 1) were not used for the inversion because the source-receiver
range was not accurately known as that AMAR’s depth was affected by tidal currents. Figure 2 shows
examples of the TL measurements and predictions at four frequencies using the inversion results for the
Victoria TL transect. Figure 3 shows single-frequency TL measurements versus range for all transects at
frequencies above 10 kHz. Because spherical spreading (20 log R, where R is range in m) with
absorption matched TL measurements well at frequencies above 5 kHz, that method was used in the SL
back-propagation calculations to calculate 1/3-octave band TL for bands at 6.3 kHz and above.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 6 - December 2014
Table 2 Geoacoustic Properties at Each Site that Result in the Best Match of Measured Single-frequency TL to Model Predictions
Site Tracks Best rms
mismatch (dB) Compressional speed (m/s)
Density (g/cm3)
Attenuation (dB/λ)
Roberts Bank North of AMAR A,
West of AMAR A 5.8 1502 1.54 0.61
Haro Strait
North of AMAR C,
South of AMAR C,
West of AMAR B
6.1 1541 1.80 1.79
Victoria Southwest of AMAR E 4.8 1558 1.64 0.83
Figure 2 Measured (Blue Symbols) and Modelled (Black Lines) TL for Tones at 316, 501, 794, and 1995 Hz Along the Southwest Transect Near Victoria. Spherical Spreading Loss (20 Log R) is Shown as Red Lines
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 7 - December 2014
Figure 3 Single-frequency TL Measurements for Roberts Bank (AMAR A), Lime Kiln (AMAR B), Haro Strait (AMARs C and D), and Victoria Pilot Site (AMAR E) Tracks Along with Spherical Spreading (20 Log R) Plus Absorption (α) for Frequencies Above 10,000 Hz
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 8 - December 2014
2.2 TRANSMISSION LOSS MODEL IN HARO STRAIT
Due to the near–shore shallow placement of TWMBR Lime Kiln hydrophone and complex local
bathymetry, a TL study was conducted between the site and the shipping lane to allow for accurate SL
estimates in that location. One-third octave band TL was modelled from ships in the northbound and
southbound shipping lanes of Haro Strait to the location of the TWMBR hydrophone near Lime Kiln. TL to
Lime Kiln was measured by subtracting container ship RLs measured on AMAR B (at Lime Kiln) from SLs
of the same container ships measured on AMAR C and D (in the middle of Haro Strait). One-third octave
band RLs were averaged over a 30 second period when three container ships were at the closest point of
approach (CPA) along the southbound lane. Ship distances to AMAR B were 4.2, 4.4, and 4.2 km for the
Osaka Express, CMA CGM Attila, and Zim Los Angeles, respectively. Computed TL is compared with the
modelled TL from the southbound lane in Figure 4. TL measurements were computed from container
ship RLs on AMAR B. The TL modelled using RAM showed good agreement with the computed TL at
frequencies between 125 Hz and 5 kHz. At lower frequencies (≤ 100 Hz) the signal to noise ratio (SNR) of
the container ship data was poor due to the high levels of flow noise on AMAR B and the low-frequency
cut-off effect caused by the near-shore bathymetry drop-off at Lime Kiln. As a consequence, TL could not
be measured below 50 Hz for two of the container ships at Lime Kiln, and the remaining low-frequency TL
data exhibited a substantial amount of scatter.
Figure 4 Comparison of Modelled and Measured 1/3-Octave Band TL between the Southbound Haro Strait Shipping Lane and Lime Kiln Site*
* Frequency bands where the background noise on AMAR B obscured sound from the container ships were excluded from the analysis.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 9 - December 2014
Above 5 kHz, spherical spreading slightly overestimated the TL between the shipping lanes and Lime
Kiln. A least-squares fit of the TL measurements above 5 kHz determined that an 18 Log R curve, plus a
frequency-dependent absorption coefficient, gave the best fit to the data (Figure 4); therefore, TL from
the southbound and northbound lanes to Lime Kiln was calculated using RAM at frequencies ≤5 kHz and
using 18 Log R with absorption at frequencies ≥6.3 kHz. The resulting 1/3 octave band TL values from the
two shipping lanes to Lime Kiln are shown in Figure 5 and detailed in Appendix A. Due to the complex
pattern in the modelled 1/3 octave band TL values, use of these 1/3 octave band TL values to estimate
SL for ships recorded at Lime Kiln resulted in an erroneous and consistent peak is source spectrum levels
between ~400 to 800 Hz. To generate a set of working 1/3 octave band TL levels that would not generate
this erroneous peak, a polynomial equation was fit to the TL values (Figure 5), resulting in smoothing TL
values at lower frequencies and fitting more closely with the TL values at higher frequencies. These
polynomial smoothed TL values from 50 Hz to 20 kHz were then added to ship RL measured on TWMBR
hydrophone at Lime Kiln to estimate SLs for these ships. Only ships clearly in the northbound lane
(1.48 to 2.96 km) and southbound lane (4.07 to 3.70 km) were included in SL estimates.
Figure 5 One-third Octave Band TL from the Northbound and Southbound Haro Strait Shipping Lanes to the TWMBR Hydrophone near Lime Kiln. A Polynomial Equation was Fit to the 1/3 Octave Band TL to Smooth the Values at Lower Frequencies
55
60
65
70
75
80
85
90
10 100 1000 10000
Tran
smis
sio
n L
oss
(d
B)
Frequency (Hz)
Northbound Lane to TWMBRhydro
Southbound Lane to TWMBRhydro
Northbound Polynomial
Southbound Polynomial
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 10 - December 2014
3.0 SHIP SOURCE LEVELS
The continuous noise produced by the container ships or tugs passing the AMARs was quantified by
computing root-mean-square sound pressure levels (rms SPLs) over consecutive one second time
windows in 1/3 octave bands. The RLs were averaged over a 30 second time window centred around the
time of the CPA. Detailed methods are outlined in Appendix A.
Data collection on ship RLs on TWMBR hydrophone at Lime Kiln was achieved with a custom Python
program. An Automatic Identification System (AIS) receiver connected to a computer allowed the
recording system to trigger a 30 second acoustic recording whenever the AIS data indicated that a ship
was transiting in Haro Strait and was abeam of Lime Kiln (i.e., 240º from Lime Kiln). AIS ship data (i.e.,
ship name, Maritime Mobile Service Identity (MMSI; a nine digit ship or coast guard station unique
identifier), number, speed, location, etc.) were also automatically collected for each of these recordings.
The audio recordings were made with a Reson TC4032 hydrophone and a MOTU Traveller soundboard
that digitised the signal at 192 kHz sampling rate. Two second time windows were used to calculate rms
SPL, which were then averaged over the 30 seconds of CPA.
3.1 SHIPS TRANSITING (HARO STRAIT)
3.1.1 Container Ship Source Levels from AMARs
Three container ships from the largest container ship size class currently using the area (i.e., >320 m in
length) with high quality AMAR sound measurements were selected for analysis. The goal was to
estimate SL in the deep waters of Haro Strait at small CPA and regular operating speeds for comparison
with estimates from TWMBR hydrophone and with AMAR data from the Victoria Pilot site where these
ships would be traveling at lower speeds. SLs were estimated for the Osaka Express, CMA CGM Attila,
and Zim Los Angeles, as they travelled at transiting speeds (~20 knots) in Haro Strait and at lower
speeds (~10 knots) south of Victoria where ships must transfer pilots (Table 3). The estimated
1/3 octave band SLs of the three ships transiting in Haro Strait and past the Victoria Pilot Site are
depicted in Figure 6. Overall, these ships have higher SL while travelling at ~20 knots than when
travelling at ~10 knots. Peak energy in the SL of these ships is below 100 Hz, but the slope of the
spectrum tends to fall off more slowly with increasing frequency while the ships are near the Victoria Pilot
Site compared to these same ships in Haro Strait. This trend is especially evident in the 1/3 octave band
SL for the Osaka Express where levels are higher at the Victoria Pilot Site than Haro Strait for
frequencies >200 Hz, which is likely due to propeller cavitation as the ships accelerate away from the pilot
station. Cavitation is caused by rapid pressure changes in the water as the propeller spins, creating
bubbles in the water which implode resulting in noise.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 11 - December 2014
Table 3 Specifications for SLs of Selected Container Ships Measured While Transiting at Different Speeds
Ship Length
(m) Draft (m)
Operation Location Speed (knots)
SL (dB re 1µPa at 1m)
Date (2013)
Osaka Express 335 9.3 Transiting Haro Strait
Victoria
22.0
9.0
206.0
198.2 2 June
CMA CGM Attila 335 12.9 Transiting Haro Strait
Victoria
24.1
10.6
203.9
191.9 5 June
Zim Los Angeles 334 13.1 Transiting Haro Strait
Victoria
22.4
11.1
200.5
187.9 8 June
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 12 - December 2014
Figure 6 Calculated 1/3 Octave Band SLs for: a) the Osaka Express; b) the CMA CGM Attila; and c) the Zim Los Angeles, at Two Different Speeds. Haro Strait Values were Averaged between AMARs C and D.
b)
c)
a)
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 13 - December 2014
Figure 7 Ship Speed (knots) Versus Calculated Broadband SL for the Three Ships Measured in Haro Strait and Near the Victoria Pilot Site
3.1.2 Ship Source Levels from TWMBR Hydrophone
Between March 8, 2011 and October 8, 2013, 5,993 transits of ships past Lime Kiln were recorded during
which no other AIS transmitting ships were within six nautical miles (11.11 km) of Lime Kiln. Of these
transits, 2,187 were of unique ships (i.e., the rest were of ships already recorded in the dataset). On
average, container ships travelled at the highest speeds and tugs at the lowest, but there was typically
little variability in ship speed within each ship class (Table 4). For comparison to the AMAR ship speed
analysis and with other studies, a multiple linear mixed effects statistical model was implemented in R
(R Core Team 2012) to explore the relationships between ship SL and ship class, length, width,
deadweight, year built, and speed over ground (SOG). This type of statistical model is an extension of a
linear regression that allows for the input of more than one independent variable. A mixed effects model
was chosen because it was assumed that the variance in SL for each individual ship (i.e., repeated
measures of the same ship) would be smaller than the variance between ships. During model selection,
deadweight, length, and width were determined to be collinear (i.e., they are correlated enough that one
variable would predict another). As width had the highest explanatory value, it was retained in the
statistical model while the other two terms were dropped. The ‘dredge’ function in R was used to select
186.0
188.0
190.0
192.0
194.0
196.0
198.0
200.0
202.0
204.0
206.0
208.0
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Bro
ad
ba
nd
SL
(d
B r
e 1
µP
a)
Ship Speed (knots)
Atilla
Osaka Express
Zim LA
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 14 - December 2014
the model with the lowest Akaike Information Criterion with correction for finite sample size (AICc) values
and the statistical model that included ship class, SOG, and width was determined to be the most
appropriate model to use (i.e., it was ranked number 1; Table 5).
Table 4 Ship Speed (knots) and Sample Size (N) for Each Ship Class Included in Analyses
Ship Class Mean 95th %ile 75th %ile 50th %ile 25th %ile 5th %ile N
Bulk carrier <200 m
13.7 6.6 12.8 13.7 14.7 18.5 1,361
Bulk carrier 200-250 m
13.5 8.5 12.6 13.5 14.5 17.4 728
Bulk carrier >250 m
13.5 7.0 12.7 13.7 14.6 18.0 275
Cargo 14.5 8.3 13.4 14.5 15.6 19.0 333
Container ship <250 m
18.9 13.0 17.6 19.1 20.4 24.3 352
Container ship 250-320 m
20.1 12.5 18.9 20.2 21.5 25.4 833
Container ship >320 m
20.4 13.3 19.4 20.6 21.6 25.4 422
Fishing 9.3 1.5 8.0 8.9 10.5 22.1 104
Passenger 15.6 4.6 13.4 15.6 18.7 25.3 107
Tanker 14.1 8.4 13.1 14.1 15.0 20.4 349
Tug 8.7 3.0 7.0 8.6 10.1 16.8 650
Vehicle carrier 17.5 10.3 16.4 17.7 18.8 22.5 479
Table 5 Model Selection Table Output for the Four Models with the Lowest AIC Values (see below for explanations of model outputs)
Rank Intercept Ship Class
SOG Width Year Built AICc Delta
1 173.1 + 0.4682 0.1025 35209.2 0
2 176.1 + 0.476 35211.8 2.53
3 196 + 0.4741 0.1074 -0.01152 35217.2 7.99
4 180.3 + 0.4772 -0.0021 35220.3 11.10
To help the reader interpret the statistical model outputs in Table 5, the output parameters are explained.
Rank is the ranking of the statistical models based on their AICc (see explanation below). The Intercept
is the y intercept estimate for that statistical model, and is the estimate of the independent variable (in this
case SL) if all the dependent variables are held at their default value. Ship Class has a ‘+’ when this
categorical independent variable is included in the statistical model. If the SOG, Width, or Year Built has
an output in the table, it indicates that this continuous independent variable is included in the statistical
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 15 - December 2014
model. These three outputs are the slope estimates for those variables. For example, an SOG output of
0.4682 indicates that for every 1 knot increase in SOG, the SL increases by 0.4682 dB. The AICc output
is a measure of the relative quality of a statistical model using a particular dataset and is what is used by
the ‘dredge’ function in R to rank the statistical models. The lower the AICc, the better the statistical
model is for that dataset. Delta is the difference in AICc score from the highest ranked model to the
model in question.
The significant positive relationship between SL and SOG with a slope of 0.47 (95% CI: 0.37 to
0.57) indicates that, on average, SL increases by almost half a dB for every one knot increase in speed
(Figure 8). Likewise, there was a positive relationship between ship width and SL such that SL increased
by 0.10 dB (95% CI: 0.04-0.17) for every 1 m increase in width (Figure 9). The R-squared value for the
fixed effects (i.e., ship class and speed) was 0.14 and the R-squared value for both fixed and random
effects (i.e., multiple measures of the same ship) was 0.22. Container ships >320 m and bulk carriers
>250 m had the highest SLs with means of 190.3 and 190.7 dB re 1µPa @ 1m, respectively, while fishing
vessels and tugs had some of the lowest with means of 177.6 and 179.3 dB re 1µPa @ 1m, respectively
(Table 6). When viewed in 1/3 octave band SLs, the broadband trends in SL tend to hold (Figure 10).
The increase in 1/3 octave spectrum levels above 10 kHz has not been reported in other literature and
may be due to large estimated TL and absorption at those high frequencies. To investigate differences in
spectrum level slopes, the 1/3 octave band SL in the 100 Hz band was subtracted from the 10,000 Hz
band (Table 7). Tankers and bulk carriers had the highest dB drop, while tugs and fishing vessels had the
smallest decrease, indicating that tankers and bulk carriers generate relatively less high frequency sound
than low frequency sound when compared to tugs and fishing vessels. This result has relevance for
investigating frequency overlap between vessel noise and marine mammal sounds (i.e., killer whale calls
and clicks).
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 16 - December 2014
Figure 8 Speed over Ground (knots) vs. SL (dB re 1µPa @ 1m). Red Line is the Linear Mixed Effects Model Estimate of this Relationship. Black Shadows around the Red Line are the 95% CI
Figure 9 Ship Width (m) vs. SL (dB re 1µPa @ 1m). Red Line is the Linear Mixed Effects Model Estimate of this Relationship. Black Shadows around the Red Line are the 95% CI
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 17 - December 2014
Table 6 SL (dB re 1µPa @ 1m) by Ship Class. Mean SLs were Calculated before Conversion to dB Scale
Ship Class Mean 95th %ile 75th %ile 50th %ile 25th %ile 5th %ile
Bulk carrier <200 m 188.3 167.4 174.8 182.0 188.1 194.2
Bulk carrier 200-250 m 187.1 166.4 173.9 181.1 187.4 193.3
Bulk carrier >250 m 190.7 172.6 180.1 185.5 190.1 196.3
Cargo 187.9 172.0 178.2 182.9 187.6 193.8
Container ship <250 m 187.0 170.9 177.2 182.4 187.3 192.6
Container ship 250-320 m 188.4 174.7 180.6 184.8 188.9 194.1
Container ship >320 m 190.3 176.5 181.8 185.6 189.5 196.4
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 18 - December 2014
Table 7 Mean Drop in the 1/3 Octave Band SL (dB) from the 100 to 10,000 Hz Band
Ship Class Mean
Bulk carrier <200 m -12.7
Bulk carrier 200-250 m -9.9
Bulk carrier >250 m -9.5
Cargo -9.8
Container ship <250 m -6.0
Container ship 250-320 m -5.1
Container ship >320 m -4.8
Fishing -3.9
Passenger -4.9
Tanker -11.9
Tug -4.2
Vehicle carrier -8.4
Sound speed profiles can change with seasons and therefore could cause changes in TL by season. To
investigate this possibility, the SLs of the five most commonly recorded container ships >320 m in
length were plotted for each ship. One-third octave levels were plotted with different colours by season.
Figure 11 is an example of these plots. While there is a large amount of variability in SL between transits,
a consistent seasonal change in the SL, either in general or at specific frequencies, is not apparent.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 19 - December 2014
Figure 11 One-third Octave Band SLs of Repeat Transits of the Vancouver Express, a 334 m Container Ship. The Colour Represents Seasons: Black is Winter, Magenta is Spring, Red is Summer, and Blue is Fall.
3.1.3 Discussion of Ship Source Levels and Comparison to Other Studies
The broadband SLs for all three container ships transiting at high speed in Haro Strait were similar (within
4 dB); however, levels above 1 kHz for the Osaka Express were approximately 20 dB lower than those
from the CMA CGM Attila and Zim Los Angeles. The high frequency SL of the Osaka Express was higher
when its speed was 9.0 knots at the Victoria pilot site than when it transited at 22.0 knots in Haro Strait.
This result, which is likely caused by increased cavitation from high load on the propellers as the Osaka
Express accelerated, was not observed with the other two container ships. These findings emphasise the
importance of cavitation in determining radiated noise levels at high frequencies that overlap with and
potentially mask killer whale call and clicks.
Broadband SLs from the three ships recorded on AMARs in Haro Strait are 15 to 20 dB higher than mean
SLs reported by McKenna et al. (2012) for container ships transiting at 20.0 knots and are above the 5th
percentile of container ships >320 m recorded with TWMBR hydrophone. The same transits of these
three ships were also recorded on TWMBR hydrophone. From that dataset, the estimated SLs of the
Osaka Express, CMA CGM Attila, and Zim Los Angeles were 187.0, 193.1 and 190.1 dB re 1 µPa at 1 m
respectively, which are 19, 10.8 and 10.4 dB lower than AMAR estimates in the middle of Haro Strait.
This may be due to the shallow depth of TWMBR hydrophone which would make it difficult to measure
frequencies below 50 Hz. Comparison of the 1/3 octave bands shows that the AMAR measurements of
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 20 - December 2014
these three ships are consistent with McKenna et al. (2012) above 200 Hz, but are higher from 20 to 160
Hz (Figure 12). Although the measurements from TWMBR hydrophone showed a great deal of variation,
the 50th percentile measurements are consistent with the different ship class measurements reported in
McKenna et al. (2012), suggesting that either the TWMBR hydrophone estimated SLs are underestimates
at low frequencies, or the AMAR estimates of SL are overestimates, or both. The relative ship-
hydrophone geometry could play a role in this anomaly (i.e., the AMAR’s much closer position was almost
directly underneath the ships, whereas the TWMBR hydrophone was to the side of them). Likewise, the
shallow water bathymetry around Lime Kiln could be filtering out lower frequencies that would cause an
underestimate of SL. The McKenna et al. (2012) recordings were in deep waters off California.
Findings of the multiple linear mixed effects model on TWMBR hydrophone data are generally in
agreement with McKenna et al. (2013), with increases in ship speed and size (either length or width, both
of which are correlated) leading to increased broadband SL. McKenna et al. (2013) found that for a 294 m
container ship, there is an increase of 1.1 dB for every 1 knot increase in ship speed. These values are
roughly double the slope predicted by the mulitple linear mixed effects model (0.47), which may be
because there was little variation in ship speed within ship class in Haro Strait (Table 4). The mixed
effects model is also lower than the power-law model developed by Ross (1976) to predict increased SL
with increased ship speed. Given the small variation in ship speed within each ship class in the
linear mixed effects model, it is probably best to use the Ross (1976) power-law model to adjust ship SL
for speed.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 21 - December 2014
Figure 12 One-Third Octave Band SLs for the Three Container Ships Measured in this Study and the Average Container Ship SLs Reported in McKenna et al. (2012)
3.2 APPROACHING PORT AND BERTHING (ROBERTS BANK)
SLs were measured for a harbour tug Seaspan Resolution, which is similar to those anticipated to be
used for berthing at RBT2, transiting at different speeds (4.0, 7.5, and 12.0 knots), performing simulations
of berthing activities at Roberts Bank, and berthing a container ship. To simulate berthing, the Seaspan
Resolution performed three activities in close proximity to the AMAR: 1) it oriented its twin propellers in
opposing directions (pushing against each other); 2) ran its engines at half and full power, generating
cavitation noise similar to that generated while manoeuvering a container ship into its berth; and 3)
accelerated from 7.1 to 11.9 knots. Engine rpm in the half power simulation, full power simulation, and
during acceleration were approximately 608 rpm, 795 rpm, and 891 rpm, respectively.
3.2.1 Tug Transiting
SLs for the Seaspan Resolution increased with its transit speed, with broadband SLs of 162.1, 171.3, and
189.1 dB re 1 µPa at 1 m at 4.0, 7.5, and 12.0 knots, respectively. These values result in an increase of
3.4 dB per knot increase on average. Cavitation noise at frequencies above 3 kHz varied substantially (by
as much as 45 dB) between the 4.0 and 12.0 knot scenarios (Figure 13).
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 22 - December 2014
Figure 13 Calculated 1/3 Octave Band SLs for Seaspan Resolution Transiting at 4.0, 7.5, and 12.0 knots
3.2.2 Tug Berthing Simulations
One-third octave band TL was calculated for the source-receiver geometry at CPA for the three berthing
simulation scenarios. SLs were calculated by taking the average of the RLs around the CPA and adding
them to TL (Figure 14). SLs for the three berthing scenarios of half power, full power, and acceleration
were 180.2, 199.7 and 191.7 dB re 1µPa at 1m, respectively.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 23 - December 2014
Figure 14 Calculated 1/3 Octave Band SLs for Seaspan Resolution Performing Berthing Simulations at Half Power (Blue), Full Power (Red), and Accelerating (Orange)
3.2.3 Tug Berthing
SLs were estimated for berthing of the Vienna Express (335 m) at Deltaport Terminal with the assistance
of the tugs Seaspan Raven and Seaspan Resolution. Average broadband SL for this berthing was 186.2
dB re 1 µPa at 1 m over the entire operation and 191.2 dB re 1 µPa at 1 m over the loudest period.
Bathymetry data and source-receiver geometry were insufficiently detailed at Deltaport Terminal to use
the acoustic model RAM to accurately predict TL during the berthing operation; therefore, TL in
1/3 octave bands was calculated instead assuming simple spherical spreading loss (20 Log R). The
1/3 octave band SLs are depicted in Figure 15.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 24 - December 2014
Figure 15 Calculated 1/3 Octave Band Composite SLs for Seaspan Resolution and Seaspan Raven Berthing the Vienna Express. SLs Were Calculated from RLs Averaged Over The Loudest Section (2 minutes, Red Line, 1596 m Slant Range) and Over the Section of the Tugs Pushing Easy to Full (30 minutes, Blue Line, 1607 m Slant Range)
Cavitation noise levels (> 1 kHz) measured during the berthing simulation, where cavitation was induced
by opposing the twin propellers on the tugs, were higher than those measured during this actual berthing
operation, particularly above 10 kHz (Figure 14 and Figure 15). SLs for the half power berthing
simulation were lower than the SLs from the berthing operation except at frequencies above 10 kHz.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 25 - December 2014
4.0 DISCUSSION OF KEY FINDINGS
TL measurements in five locations in the Salish Sea combined with acoustic inversion analysis allowed
development of a site-specific acoustic transmission model and determination of geoacoustic properties
in particular areas. Acoustic transmission models require this information to determine how the ocean
bottom reflects and absorbs sound as it travels away from its source. While the acoustic inversion
determined that all sites contained acoustically absorptive silts and clays, variation between sites will help
better parameterise noise models (see JASCO 2014a, b).
The attenuation model that best fit the TL data depended on frequency. Wave-equation based models
(parabolic equation or wavenumber integration) worked best for frequencies below 5 kHz. Above 5 kHz,
simple spherical spreading with absorption was the best match. The difficulties with fine-tuning the TL
model at Lime Kiln are indicative of the challenge of modelling acoustic transmission in complex
bathymetry.
SL estimates of vessels in this study were consistent with other recent studies but provided larger sample
sizes (and therefore more robust estimates), extended SL measurements to other ship types and
activities including tugs and berthing, and characterised the relationship between ship speed and size,
which will again help to better parameterise noise models. During transit, container ships have some of
the highest SLs while tugs have some of the lowest; however, when tugs are accelerating or berthing a
ship, cavitation of their propellers generates higher SL. The increase in tug SL with speed is about three
times higher than for container ships, which is likely due to the very different vessel designs.
5.0 DATA GAPS AND LIMITATIONS
Measuring TL from playback studies and the RL of vessels in the inland waters of the Salish Sea
presents a number of challenges because of the complex and sometimes shallow bathymetry.
Acoustic inversion analysis resulted in geoacoustic estimates in only three locations. To model noise in
the entire study area, these geoacoustic properties will need to be applied over much larger areas with
variable geoacoustic properties, which may lead to errors in noise model estimates in some areas.
There were discrepancies in the SL estimates of the three large container ships in Haro Strait between
the AMAR and TWNBR datasets and a great deal of variability within TWMBR dataset. An estimate of SL
for these large ships that combines the strengths of these different datasets will be the most appropriate
approach to noise modelling.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 26 - December 2014
6.0 CLOSURE
Major authors and reviewers of this technical data report are listed below, along with their signatures.
Report prepared by: SMRU Canada Ltd.
Jason Wood, PhD Senior Research Scientist Hemmera Envirochem Inc.
Elly Chmelnitsky, M.Sc. Marine Biologist The following persons contributed to this TDR and are authors of the report, Underwater Acoustic Measurements in Haro Strait and Strait of Georgia, included in Appendix A of this document: JASCO Graham Warner Caitlin O’Neill Andrew McCrodan Heloise Frouin-Mouy Jonathan Izett Alexander MacGillivray Report peer reviewed by: Hemmera Envirochem Inc.
Sonya Meier, M.Sc., R.P.Bio., Senior Biologist Acknowledgements: Data collection at TWMBR hydrophone made possible by: Val Veirs, PhD Professor Emeritus, Colorado College Scott Veirs, PhD President, Beam Reach
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 27 - December 2014
7.0 REFERENCES
Bassett, C., B. Polagye, M. Holt, and J. Thomson. 2012. A vessel noise budget for Admiralty Inlet, Puget
Sound, Washington (USA). Journal of the Acoustical Society of America 132: 3706-3719.
JASCO. 2014a. Roberts Bank Terminal 2 technical report: Regional commercial vessel traffic underwater
noise exposure study. Prepared for Port Metro Vancouver, Vancouver, B.C. in Port Metro
Vancouver (PMV). 2015. Roberts Bank Terminal 2 Environmental impact statement: Volume 2.
Environmental Assessment by Review Panel. Submitted to Canadian Environmental Assessment
Agency.
JASCO. 2014b. Roberts Bank Terminal 2 technical report: Construction activities and terminal vessel
operations noise modelling study. Prepared for Port Metro Vancouver, Vancouver, B.C. in Port
Metro Vancouver (PMV). 2015. Roberts Bank Terminal 2 Environmental impact statement:
Volume 2. Environmental Assessment by Review Panel. Submitted to Canadian Environmental
Assessment Agency.
Jensen, F. B., W. A. Kuperman, M. B. Porter, and H. Schmidt. 2000. Computational Ocean Acoustic. AIP
Series in Modern Acoustics and Signal Processing. AIP Press - Springer.
McKenna, M.F., D. Ross, S.M. Wiggins, and J.A. Hildebrand. 2012. Underwater radiated noise from
modern commercial ships. Journal of the Acoustical Society of America 131: 92-103.
McKenna, M. F., S. M. Wiggins, and J. A. Hildebrand. 2013. Relationship between container ship
underwater noise levels and ship design, operational and oceanographic conditions. Scientific
Reports 3: 1760.
R Core Team. 2012. R: A language and environment for statistical computing. Vienna, Austria.
http://www.r-project.org.
Ross, D. 1976. Mechanics of Underwater Noise. Pergamon, New York.
SMRU. 2014. Roberts Bank Terminal 2 technical report: Southern resident killer whale underwater noise
exposure and acoustic masking study. Prepared for Port Metro Vancouver, Vancouver, B.C. in
Port Metro Vancouver (PMV). 2015. Roberts Bank Terminal 2 Environmental impact statement:
Volume 3. Environmental Assessment by Review Panel. Submitted to Canadian Environmental
Assessment Agency.
Urick, R. J. 1983. Principles of Underwater Sound. 3rd edition. McGraw-Hill, New York, NY.
Port Metro Vancouver Hemmera, SMRU Canada, and JASCO RBT2 – Ship Sound Signatures - 28 - December 2014
8.0 STATEMENT OF LIMITATIONS
This report was prepared by Hemmera Envirochem Inc. (“Hemmera”) and SMRU Canada Ltd. (“SMRU”),
based on fieldwork conducted by JASCO, for the sole benefit and exclusive use of Port Metro Vancouver.
The material in it reflects Hemmera, SMRU and JASCO’s best judgment in light of the information
available at the time of preparing this Report. Any use that a third party makes of this Report, or any
reliance on or decision made based on it, is the responsibility of such third parties. Hemmera, SMRU and
JASCO accept no responsibility for damages, if any, suffered by any third party as a result of decisions
made or actions taken based on this Report.
Hemmera, SMRU and JASCO have performed the work as described above and made the findings and
conclusions set out in this Report in a manner consistent with the level of care and skill normally
exercised by members of the environmental science profession practicing under similar conditions at the
time the work was performed.
This Report represents a reasonable review of the information available to Hemmera, SMRU and JASCO
within the established Scope, work schedule and budgetary constraints. The conclusions and
recommendations contained in this Report are based upon applicable legislation existing at the time the
Report was drafted. Any changes in the legislation may alter the conclusions and/or recommendations
contained in the Report. Regulatory implications discussed in this Report were based on the applicable
legislation existing at the time this Report was written.
In preparing this Report, Hemmera, SMRU and JACO have relied in good faith on information provided by
others as noted in this Report, and have assumed that the information provided by those individuals is
both factual and accurate. Hemmera, SMRU and JASCO accept no responsibility for any deficiency,
misstatement or inaccuracy in this Report resulting from the information provided by those individuals.
APPENDIX A
Underwater Acoustic Measurements
in Haro Strait and Strait of Georgia
Transmission Loss, Vessel Source Levels, and Ambient Measurements
Submitted to: Hemmera
Authors: Graham Warner Caitlin O’Neill Andrew McCrodan Heloise Frouin-Mouy Jonathan Izett Alexander MacGillivray
26 September 2014
P001140-004 Document 00659 Version 4.0
JASCO Applied Sciences (Canada) Ltd. 2305–4464 Markham Street
Victoria, BC V8Z 7X8 Canada Tel: +1-250-483-3300
Fax: +1-250-483-3301 www.jasco.com
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
i
Technical Report/Technical Data Report Disclaimer
The Canadian Environmental Assessment Agency determined the scope of the proposed Roberts
Bank Terminal 2 Project (RBT2 or the Project) and the scope of the assessment in the Final
Environmental Impact Statement Guidelines (EISG) issued January 7, 2014. The scope of the
Project includes the project components and physical activities to be considered in the
environmental assessment. The scope of the assessment includes the factors to be considered
and the scope of those factors. The Environmental Impact Statement (EIS) has been prepared in
accordance with the scope of the Project and the scope of the assessment specified in the EISG.
For each component of the natural or human environment considered in the EIS, the geographic
scope of the assessment depends on the extent of potential effects.
At the time supporting technical studies were initiated in 2011, with the objective of ensuring
adequate information would be available to inform the environmental assessment of the Project,
neither the scope of the Project nor the scope of the assessment had been determined.
Therefore, the scope of supporting studies may include physical activities that are not included in
the scope of the Project as determined by the Agency. Similarly, the scope of supporting studies
may also include spatial areas that are not expected to be affected by the Project.
This out-of-scope information is included in the Technical Report (TR)/Technical Data Report
(TDR) for each study, but may not be considered in the assessment of potential effects of the
Project unless relevant for understanding the context of those effects or to assessing potential
cumulative effects.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
ii Version 4.0
Contents
GLOSSARY .................................................................................................................................. VIII
LIST OF ACRONYMS ....................................................................................................................... XI
Figure 15. Measured single-frequency TL along the Haro Strait west track to AMAR B at Lime Kiln. ... 21
Figure 16. Measured single-frequency TL along the southwest track near Victoria to AMAR E. ............. 22
Figure 17. Measured single-frequency TL along the north track near Deltaport Terminal to
AMAR A. .............................................................................................................................................. 23
Figure 18. Measured single-frequency TL along the west track near Deltaport Terminal to AMAR A. ... 24
Figure 19. Measured single-frequency TL for the station 2,500 m NW of AMAR A. ............................... 24
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
iv Version 4.0
Figure 20. Measured (blue symbols) and modelled (black lines) TL for tones at 316, 501, 794, and
1995 Hz along the southwest transect near Victoria. Spherical spreading loss (20 Log R) is shown
as red lines. ............................................................................................................................................ 26
(AMAR B), Haro Strait (AMARs C and D), and Victoria pilot site (AMAR E) tracks along with
spherical spreading (20 Log R) plus absorption (α) for frequencies above 10,000 Hz. ........................ 27
Figure 22. Comparison of modeled and measured 1/3-octave-band TL between the southbound Haro
Strait shipping lane and Lime Kiln. ....................................................................................................... 29
Figure 23. One-third-octave-band TL from the northbound and southbound Haro Strait shipping
lanes to the Whale Museum and Beam Reach hydrophone off Lime Kiln. ........................................... 30
Figure 24. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Osaka Express while the
vessel transited south through Haro Strait at 22.0 kts measured by (left) AMAR C and (right)
AMAR D. .............................................................................................................................................. 31
Figure 25. Spectrogram of Osaka Express (870 m CPA) measured by AMAR C at 219 m depth.
Times are in UTC. ................................................................................................................................. 32
Figure 26. Spectrogram of Osaka Express (850 m CPA) measured by AMAR D at 34 m depth.
Times are in UTC. ................................................................................................................................. 33
Figure 27. PSD of the Osaka Express at CPA measured at 846 m horizontal range by (blue)
AMAR C at 219 m depth and (red) AMAR D at 34 m depth. Levels were averaged over 10 one-
second windows, with 50% overlap using a Hanning window. ............................................................ 33
Figure 28. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Osaka Express while the
vessel transited south past the Victoria pilot site at a CPA (200 m) at speed of ~9.0 kts, measured
on AMAR E. .......................................................................................................................................... 34
Figure 29. Spectrogram of Osaka Express (200 m CPA) measured by AMAR E at 76 m depth. Times
are in UTC. ............................................................................................................................................ 35
Figure 30. PSD of the Osaka Express at CPA (200 m), measured by AMAR E. Levels were averaged
over 10 one-second windows, with 50% overlap using a Hanning window. ........................................ 35
Figure 31. Calculated 1/3-octave-band source levels for the Osaka Express transiting at two different
Figure 32. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the CMA CGM Attila while
the vessel transited south through Haro Strait at 24.1 kts measured by (left) AMAR C and (right)
AMAR D. .............................................................................................................................................. 37
Figure 33. Spectrogram of CMA CGM Attila including 1120 m CPA measured by AMAR C at
219 m depth. Times are in UTC. ........................................................................................................... 38
Figure 34. Spectrogram of CMA CGM Attila including 1100 m CPA measured by AMAR D at 72 m
depth. Times are in UTC. ...................................................................................................................... 39
Figure 35. PSD of the CMA CGM Attila at CPA measured at 1098 m horizontal range by (blue)
AMAR C at 219 m depth and (red) AMAR D at 72 m depth. Levels were averaged over 10 one-
second windows, with 50% overlap using a Hanning window. ............................................................ 39
Figure 36. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the CMA CGM Attila while
the vessel transited south past the Victoria pilot site at CPA (360 m) at speed of 10.6 kts,
measured on AMAR E. .......................................................................................................................... 40
Figure 37. Spectrogram of CMA CGM Attila including 360 m CPA measured by AMAR E at 76 m
depth. Times are in UTC. ...................................................................................................................... 41
Figure 38. PSD of the CMA CGM Attila at CPA (360 m), measured by AMAR E. Levels were
averaged over 10 one-second windows, with 50% overlap using a Hanning window. ......................... 41
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 v
Figure 39. Calculated 1/3-octave-band source levels for the CMA CGM Attila transiting at two
different speeds. ..................................................................................................................................... 42
Figure 40. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Zim Los Angeles while the
vessel transited south through Haro Strait at 22.4 kts measured by (left) AMAR C and (right)
AMAR D. .............................................................................................................................................. 43
Figure 41. Spectrogram of Zim Los Angeles including 830 m CPA measured by AMAR C at 219 m
depth. Times are in UTC. ...................................................................................................................... 44
Figure 42. Spectrogram of Zim Los Angeles including 810 m CPA measured by AMAR D at 43.8 m
depth. Times are in UTC. ...................................................................................................................... 45
Figure 43. PSD of the Zim Los Angeles at CPA measured at 794 m horizontal range by (blue)
AMAR C at 219 m depth and (red) AMAR D at 43.8 m depth. Levels were averaged over 10 one-
second windows, with 50% overlap using a Hanning window. ............................................................ 46
Figure 44. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Zim Los Angeles while the
vessel transited south past the Victoria pilot site at CPA (67 m) at speed of 11.1 kts, measured on
AMAR E. The higher levels at 600 m range are due to additional noise from a pilot boat. .................. 46
Figure 45. Spectrogram of Zim Los Angeles including 67 m CPA measured by AMAR E at 76 m
depth. Times are in UTC. ...................................................................................................................... 47
Figure 46. PSD of the Zim Los Angeles at CPA (67 m), measured by AMAR E. Levels were
averaged over 10 one-second windows, with 50% overlap using a Hanning window. ......................... 47
Figure 47. Calculated 1/3-octave-band source levels for the Zim Los Angeles transiting at two
different speeds. ..................................................................................................................................... 48
Figure 48. Broadband (50 Hz to 64 kHz) received levels as a function of time of the Seaspan
Resolution and Seaspan Raven berthing the Vienna Express. Range is between the Seaspan
Resolution and the AMAR. Data marked in red indicate the 30 minute time window of data used
for computing average source levels. Times are in UTC. ...................................................................... 49
Figure 49. PSD of the Seaspan Resolution and Seaspan Raven berthing the Vienna Express. The
spectral measurements were computed from averaging received levels over the loudest section (2
minutes, red line, 1,596 m slant range) and over the section of the tugs pushing easy to full (30
minutes, blue line, 1,607 m slant range). ............................................................................................... 50
Figure 50. Calculated 1/3-octave-band composite source levels for Seaspan Resolution and Seaspan
Raven berthing the Vienna Express. Source levels were calculated from received levels averaged
over the loudest section (2 minutes, red line, 1,596 m slant range) and over the section of the tugs
pushing easy to full (30 minutes, blue line, 1,607 m slant range). ........................................................ 50
Figure 51. Spectrogram of container ship loading measurements. ............................................................. 52
Figure 52. Broadband (up to 18 kHz) received SPL as a function of perpendicular distance to the
CMA CGM La Scala. ............................................................................................................................. 53
Figure 53. 1/3-octave band source levels for loading of the container ship CMA CGM La Scala.
Source levels above 16 kHz (in red) were calculated from linear extrapolation of received band
levels between 5-16 kHz ........................................................................................................................ 53
Figure 54. Received levels as a function of slant range to the Seaspan Resolution transiting at 4 kts,
7.5 kts, and 12 kts. Data at distances less than 168 m for the 12 kt transit are not shown as they
exceeded the amplitude limit of the recorder. ........................................................................................ 54
Figure 55. PSD from 5-second averages near CPA, of the Seaspan Resolution transiting at 4 kts at
48.9 m slant range, at 7.5 kts at 50.1 m slant range, and 12 kts at 168 m slant range. .......................... 55
Figure 56. Calculated 1/3-octave-band source levels for Seaspan Resolution transiting at 4, 7.5, and
simulations at half power (blue), full power (red), and accelerating (orange). ...................................... 57
Figure 60. Deltaport Terminal (AMAR A): (top) Broadband and decade band SPL (30 minute
average) from 30 May to 19 Jun; (bottom) Ambient noise spectrogram (30 minute average) over
the same period. ..................................................................................................................................... 59
Figure 61. Lime Kiln (AMAR B): (top) Broadband and decade band SPL (30 minute average) from
29 May to 11 Jun; (bottom) Ambient noise spectrogram (30 minute average) over the same
The AMARs were calibrated with a Pistonphone Type 42AA precision sound source (G.R.A.S.
Sound & Vibration A/S) before they were deployed and again when they were retrieved. The
pistonphone calibrator produces a constant tone at 250 Hz at the hydrophone sensor. The
recorded level of the reference tone on the AMAR yields the total pressure sensitivity for the
instrument, i.e., the conversion factor between digital units and pressure. To determine absolute
SPLs, this gain was applied during data analysis. The calibration variance using this method is
typically less than 0.5 dB absolute pressure.
Five AMARs were used for this study, one of which was recovered part way through the field
measurement and redeployed in a different location (AMAR B was redeployed as AMAR F).
The deployment locations are listed in Table 2 and shown in Figure 2. All AMARs were
deployed from either the R/V Coastal Geosciences or R/V Buzzard. Figure 3 shows AMAR B
being deployed. AMARs A, B, C, E, and F were mounted 3 m above the seafloor; AMAR D was
anchored to the seafloor on a 200 m long low-drag mooring. During slack tide, the AMAR was
at 20 m depth; during strong tidal flows, the drag on the AMAR’s mooring caused it to go as
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
6 Version 4.0
deep as 120 m. A Star-Oddi depth logger recorded the AMAR’s depth as a function of time
(Figure 4). AMARs A, C, D, and F were retrieved by activating an acoustic release, allowing the
mooring to float to the surface. AMARs B and E were retrieved by grappling. Sounds from the
acoustic release transducer were not played in U.S. waters.
Table 2. AMAR deployment and retrieval details.
AMAR Station Latitude Longitude Water Depth (m)
AMAR Depth (m)
Deployment Date/Time (UTC)
Retrieval Date/Time (UTC)
A Deltaport
Terminal 49°0.255′ N 123°9.295′ W 55 52 29 May 17:03 18 June 19:05
B Lime Kiln 48°30.942′ N 123°9.196′ W 16 13 28 May 18:59 11 June 23:15
C Haro Strait
48°30.440′ N 123°11.856′ W 219 216 30 May 17:55 21 June 16:56
D Haro Strait
48°30.448′ N 123°11.838′ W 221 20-120* 30 May 18:42 21 June 18:07
E Victoria 48°22.959′ N 123°23.994′ W 76 73 28 May 16:15 17 June 19:25
F Georgia Strait
48°57.147′ N 123°18.860′ W 184 181 21 June 22:33 2 July 19:19
*Due to strong tidal currents in Haro Strait, the depth of this AMAR changed between slack tide and running tide.
Figure 2. AMAR deployment locations.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 7
Figure 3. Deployment of AMAR B off Lime Kiln from the R/V Buzzard. The white satellite beacon, AMAR with orange cylindrical float collars, and yellow tandem releases (left to right) can be seen prior to the main anchor deployment. Photo credit: Jason Wood.
Figure 4. Depth of AMAR D versus time (UTC) as measured on the Star-Oddi DST sensor.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
8 Version 4.0
2.3. Transmission Loss Measurements
TL was measured by playing underwater sounds at several distances from the AMARs and
calculating the difference between the received level on the AMAR and the source level
measured by a hydrophone at the transducer (Equation 3). The transducer was a Lubell Labs
9162 sound source driven using waveforms played from a Sound Devices 722 digital audio
recorder, amplified with a 300 W JL Audio Model 300/2 amplifier. The system was powered
with a 12 V battery and the rms voltage driving the transducer was adjusted to limit the
maximum source level to approximately 180 dB re 1 µPa at 1 m. The source waveform played at
each station was a sequence of 20 five-second long tones at 1/3-octave-band centre frequencies
between 300 Hz and 30 kHz.
The source levels were recorded using a second 722 recorder with a RESON TC4043
hydrophone (nominal sensitivity −201 dB re 1 V/µPa) mounted 1 m from the transducer
(Figure 5). While the source was broadcasting, a continuous recording was made at 96,000
samples per second with 24 bit samples. The TC4043 hydrophone and 722 recorder were
calibrated as a system with the 42AA Pistonphone calibrator.
The source waveform was played at several stations along tracks near Deltaport Terminal, in
Haro Strait, and at the Victoria pilot site. At each source station, the transducer was aimed at the
receiving AMAR, with the RESON hydrophone positioned in line with the transducer and the
AMAR. Tables 3, 4, and 5 list the coordinates and Figures 6, 7, and 8 show the tracks of the
playback locations near Deltaport Terminal, in Haro Strait, and at the Victoria pilot site,
respectively.
Salinity, temperature, and depth profiles were taken with a Minos-X CTD (conductivity,
temperature, and depth) at each of these three TL sites prior to the playbacks. Water sound speed
profiles were calculated from the CTD measurements according to the equation of Coppens
(1981). Appendix C shows temperature, salinity, and water sound speed profiles measured at
each TL site.
During acoustic playbacks a marine mammal observer was present to record marine mammal
observations and implement mitigation measures recommended by Fisheries and Oceans Canada
(Appendix D) as appropriate.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 9
Figure 5. Lubell Labs 9162 transducer (blue) with RESON TC4043 hydrophone mounted 1 m from the transducer on a PVC pipe. The transducer was mounted on a 6 m galvanised steel pole.
Figure 6. Deployment location for AMAR A near Deltaport Terminal and the approximate TL tracks (black).
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
10 Version 4.0
Table 3. Coordinates of the playback stations along the north and west transects, and the northwest playback station at the Deltaport Terminal site for AMAR A.
Direction from AMAR
Distance from AMAR (m)
Latitude Longitude
North 200 49°00.366′ N 123°09.313′ W
North 400 49°00.474′ N 123°09.314′ W
North 600 49°00.582′ N 123°09.314′ W
North 800 49°00.689′ N 123°09.314′ W
North 1000 49°00.797′ N 123°09.315′ W
West 500 49°00.257′ N 123°09.723′ W
West 1000 49°00.257′ N 123°10.133′ W
West 1500 49°00.256′ N 123°10.543′ W
West 2000 49°00.255′ N 123°10.954′ W
West 2500 49°00.255′ N 123°11.364′ W
West 3000 49°00.254′ N 123°11.774′ W
West 3500 49°00.253′ N 123°12.184′ W
West 4000 49°00.252′ N 123°12.594′ W
West 4500 49°00.252′ N 123°13.005′ W
West 5000 49°00.251′ N 123°13.415′ W
Northwest 2500 49°00.791′ N 123°11.196′ W
Figure 7. Deployment locations for AMAR B off Lime Kiln and AMARs C and D in Haro Strait. The approximate TL tracks are shown in black.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 11
Table 4. Coordinates of the playback stations along the north, south, and west transects in Haro Strait.
Direction from AMAR Distance from AMAR (m) Receiver AMAR Latitude Longitude
North 500 C/D 48°30.698′ N 123°11.989′ W
North 1000 C/D 48°30.950′ N 123°12.133′ W
North 1500 C/D 48°31.202′ N 123°12.277′ W
North 2000 C/D 48°31.454′ N 123°12.421′ W
South 500 C/D 48°30.194′ N 123°11.699′ W
South 1000 C/D 48°29.942′ N 123°11.552′ W
South 1500 C/D 48°29.690′ N 123°11.406′ W
South 2000 C/D 48°29.439′ N 123°11.260′ W
South 2500 C/D 48°29.187′ N 123°11.113′ W
South 3000 C/D 48°28.935′ N 123°10.967′ W
South 3500 C/D 48°28.684′ N 123°10.821′ W
South 4000 C/D 48°28.432′ N 123°10.674′ W
South 4500 C/D 48°28.181′ N 123°10.528′ W
South 5000 C/D 48°27.929′ N 123°10.382′ W
West 4000 B 48°30.401′ N 123°12.245′ W
West 3500 B 48°30.461′ N 123°11.906′ W
Figure 8. Deployment location for AMAR E at the Victoria pilot site and the approximate TL track (black).
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
12 Version 4.0
Table 5. Coordinates of the playback stations for AMAR E along the southwest transect off Victoria.
Distance from AMAR (m)
Latitude Longitude
500 48°22.750′ N 123°24.062′ W
1000 48°22.517′ N 123°24.266′ W
1500 48°22.284′ N 123°24.470′ W
2000 48°22.051′ N 123°24.673′ W
2500 48°21.818′ N 123°24.877′ W
3000 48°21.585′ N 123°25.081′ W
3500 48°21.352′ N 123°25.285′ W
4000 48°21.119′ N 123°25.489′ W
4500 48°20.886′ N 123°25.693′ W
5000 48°20.652′ N 123°25.897′ W
Source sound tone levels for each station and frequency were computed from recordings on the
TC4043 source hydrophone. The raw pressure waveform data were scaled according to the mean
calibrated sensitivity of the 722 digital audio recorder with the TC4043 hydrophone. Power
spectral density (PSD) was computed for each tone (5 s of data) according to Welch’s method
(Oppenheim and Schafer 1999), using a normalised Hanning window with 50% overlap. The fast
Fourier transform (FFT) length for the PSD calculation was 96,000 points and the frequency bin
width was 1 Hz. Source SEL was computed for each tone by integrating the PSD over its
frequency bandwidth and time duration. The bandwidth over which source SEL was calculated
was based on the bandwidth of the received tone. The bandwidth for tones below 6 kHz was
7 Hz centred around the tone frequency; the bandwidth was increased to 11 Hz for tones above
6 kHz. The reason for this increase is discussed below.
Received tone levels for each station and frequency were computed from recordings on the
AMARs. Received SELs were calculated from the sum of the direct-path tone and reverberation
inside 10 second windows. The raw pressure waveform data were scaled according to the mean
calibrated sensitivity of the AMAR with the M8E hydrophone. PSD and SEL were computed for
each tone as was done for the source level calculation, however, the measured frequency of the
tones were occasionally shifted from the playback frequency because the R/V Coastal
Geosciences was drifting during the playbacks. The movement caused a shift and spread in
frequency (Doppler effect) most prominent at higher frequencies. The increased frequency
spread required larger bandwidths to calculate SELs for tones above 6 kHz. The received levels
were calculated by centring the 7 Hz or 11 Hz wide windows over the peak PSD frequency.
Background noise levels were calculated by averaging the PSD over a larger bandwidth centred
around the PSD peak but excluding the frequencies used to calculate the tone levels. For
frequencies below 6 kHz the background bandwidth was 42 Hz and for frequencies above 6 kHz
the background bandwidth was 70 Hz. Received tones with a signal-to-noise ratio (SNR) less
than 6 dB were excluded from the analysis.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 13
Figure 9 shows examples of the process for calculating received levels at 794 and 30,000 Hz for
tones played 2,500 m from AMAR E at the Victoria pilot site. The reverberation at 794 Hz was
more than 6 dB above the background noise so the received level was the sum of the tone and
reverberation levels. The reverberation at 30,000 Hz was less than 6 dB above the background
noise so the reverberation was not included in the received level for that tone. The plot for the
30,000 Hz tone also shows the shift and spread in frequency due to the Doppler effect. Source
levels were computed using the same method, though reverberation and frequency shift effects
were not present on the source hydrophone.
Figure 9. Received PSD levels for the 794 Hz (left) and 30,000 Hz (right) tones played 2,500 m from AMAR E (Victoria pilot site). Black lines show the tone levels from the first half of the 10 second window; the blue lines show the reverberation levels from the second half of the 10 second window. Tone and reverberation levels (solid horizontal lines) were calculated from PSD levels at frequencies marked with asterisk symbols (*). Noise levels were computed from PSD levels over frequencies spanned by the dashed horizontal lines. Note that tone, reverberation, and noise levels (horizontal lines) are in SEL units (dB re 1 µPa2s).
2.4. Transmission Loss Model
One-third-octave band TL was required to determine source levels for several vessel operations
in this study; however, measured single-frequency TL at 1/3-octave-band centre frequencies may
not accurately represent TL for an entire 1/3-octave-band. Particularly at low frequencies, TL at
specific ranges can be large due to destructive interference between different propagation paths.
Single-frequency TL would not be representative of 1/3-octave-band TL if the measurement
geometry aligned with these nulls.
Previous vessel source level studies (e.g., McKenna et al. 2012 and Bassett et al. 2012) have
approximated TL using frequency independent empirical spreading loss equations of the form
N Log R, where N is a spreading coefficient (equal to 20 for spherical spreading) and R is the
source-receiver range. This approximation is adequate for high frequencies but is not suitable for
low frequencies, particularly in environments with complex bathymetry.
One-third-octave-band TL values were therefore calculated with an acoustic propagation model.
The range-dependent acoustic model RAM was used because of its suitability to low frequency
propagation and its ability to take into account complex bathymetry and subbottom geoacoustic
properties. The geoacoustic properties of the subbottom at each site were unknown so RAM was
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
14 Version 4.0
used to invert for these properties by modelling single-frequency TL corresponding to each TL
measurement. The geoacoustic properties were adjusted to minimise the difference between the
single-frequency TL measurements and model predictions in the least squared sense. This was
done using a simulated annealing algorithm that inverted for the compressional wave speed,
density, and compressional wave attenuation of the seabed. The algorithm also inverted for the
AMAR height above the seafloor (which was known within a few metres from the mooring
design, but not accurately enough for RAM) and an offset in water depth to account for the R/V
Coastal Geosciences sounder’s depth. Sound speed profiles were derived from CTD casts taken
during the TL measurements. Water depths were obtained from sounder measurements and from
digital bathymetry charts (National Geophysical Data Center 2013).
TL measurements at 316, 501, 794, 1,259, 1,585, and 1,995 Hz were used for the geoacoustic
inversion; TL measurements at 398, 631, and 1,000 Hz were found to destabilise the inversion so
they were excluded from the analysis. TL measurements above 2 kHz were also excluded to
decrease the computational time required for the inversion. The TL data were not sensitive to
seabed layering, i.e., the TL measurements could not be better matched by allowing a layered
substrate with depth-dependent geoacoustic properties. The substrate was therefore
parameterised as a fluid half-space.
Figure 10 shows an example of the simulated annealing run for the Victoria TL transect. The rms
mismatch is shown as a function of the inversion parameters. The geoacoustic properties of each
site were determined from the model that had the lowest mismatch.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 15
Figure 10. Model-measurement mismatch versus inversion parameters for the Victoria TL transect simulated annealing run. Compressional wave speed, delta water depth, and receiver height off the bottom are well determined but density and attenuation are not.
Once the geoacoustic properties at each site were determined, either RAM or a wavenumber
integration model (Jensen et al. 2000) was used to calculate 1/3-octave-band TL by averaging
single-frequency TL at 11 frequencies spaced evenly within each 1/3-octave-band. The
wavenumber integration model was used for measurements at less than 100 m range where the
wide-angle parabolic-equation (PE) approximation used in RAM was not sufficient to accurately
predict TL. The acoustic propagation model was used up to 5 kHz. Above 5 kHz spherical
spreading (20 Log R) with absorption (François and Garrison 1982a, 1982b) was used to back-
propagate source level measurements because it was found that it agreed well with single-
frequency TL measurements at short-range.
2.5. Source Level Measurements
2.5.1. Vessels Transiting and Berthing
The continuous noise produced by the container ships or tugs passing the AMARs was quantified
by computing rms SPLs over consecutive 1 s time windows in 1/3-octave bands. The received
levels were averaged over a 30 second time window centred around the time of the closest point
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
16 Version 4.0
of approach (CPA). Shorter averaging times were used if the ships passed very close to the
AMARs and the TL, estimated by spherical spreading (20 Log R), varied by more than 3 dB
over the minimum and maximum ranges in a 30 second window. This corresponds to a
maximum/minimum distance ratio of greater than 141%. For these cases, the averaging time
window was the longest window for which TL, estimated by spherical spreading 20 Log R,
varied by less than 3 dB over the minimum and maximum ranges. Time-stamped locations and
drafts for each vessel were obtained from Automatic Identification System (AIS) records
provided by MarineTraffic.com.
One-third-octave-band TL was calculated as described in Section 2.4 with the acoustic source
depth estimated according to the procedure listed in Wright and Cybulski (1983). The source of
radiated noise was assumed to be at a point midway between the shaft and the top of the
propeller disk, and the depth of the bottom of the propeller disk was assumed to be equal to the
ship’s draft. The propeller diameters of the container ships measured in this study were unknown
so they were estimated to be 8.7 m, based on propeller diameters from ships of similar size
(Wright and Cybulski 1983). The propeller diameter of the tug Seaspan Resolution was taken to
be 2.7 m, based on information provided on the vessel data sheet (Seaspan Marine 2013). The
source-receiver range was usually the CPA; however, for some measurements, farther ranges had
to be used because the recorder was saturated or the highest sound levels were measured when
the vessels were more distant from the AMAR.
2.5.2. Container Ship Loading
Underwater sound was measured during the loading of containers on the M/V CMA CGM La
Scala at Deltaport Terminal on 18 June 2013 from approximately 13:15 to 14:00 PDT.
Recordings were made with a Sound Devices 722 recorder and RESON TC4032 hydrophone
(nominal sensitivity −170 dB re 1 V/µPa). The recorder was calibrated before the recordings
with the same G.R.A.S. pistonphone calibrator as was used for the AMARs. Continuous
recordings were made at 96,000 samples per second at 24-bit resolution. The hydrophone was
deployed over the side of the R/V Coastal Geosciences at 10 m depth. R/V Coastal Geosciences’
engines were turned off and the vessel drifted for all measurements. A GPS was used to track the
R/V Coastal Geosciences’ position and a laser rangefinder was used to determine perpendicular
distance to the CMA CGM La Scala. Measurements began at the stern of the CMA CGM La
Scala and were collected as the R/V Coastal Geosciences drifted toward the bow. The R/V
Coastal Geosciences was occasionally repositioned to record sounds along the entire length of
the CMA CGM La Scala.
2.6. Ambient Noise Measurements
Ambient noise measurements were obtained at five different locations inside the study area, over
a one-month period (Table 2 and Figure 2). The objective of the ambient measurements was to
provide a statistical description of baseline ambient noise conditions throughout the study area.
The raw acoustic data were processed using JASCO’s ambient noise analysis software to
compute SPL and PSD over the deployment period. The raw pressure waveform data were
scaled according to the mean calibrated pressure sensitivity of the AMAR and adjusted for the
frequency response of the hydrophone sensor. Time domain pressure waveforms were analysed
to find rms SPL for each minute of data. Mean SPL for each minute of data was computed in
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 17
four decade bands (10 to100 Hz, 100 to 1,000 Hz, 1,000 to 10,000 Hz, and > 10,000 Hz)
spanning the sampling bandwidth of the acoustic recorders.
PSD was computed for each minute of data according to Welch’s method (Oppenheim and
Shafer 1999), using a normalised Hamming window with 50% overlap. The FFT length for the
PSD calculation was 128,000 samples and the frequency bin width was 1 Hz. Cumulative
probabilities were computed from the 1 minute PSDs over the entire recording period to
determine the statistical distribution of the recorded ambient noise levels. The cumulative
probabilities were plotted as percentiles (5%, 25%, 50%, 75%, and 95%); these depicted the
proportion of time when PSDs at each frequency exceeded a particular sound level.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
18 Version 4.0
3. Results
The following section presents the results of the Underwater Acoustic Measurements in Haro
Strait and Strait of Georgia Study.
3.1. Transmission Loss Measurements
Sections 3.1.1 to 3.1.4 show single-frequency TL for the AMARs in Haro Strait, at Lime Kiln,
the Victoria pilot site, and near Deltaport Terminal. Not all frequencies are included in each plot
because for some tones, the received levels were not sufficiently above background noise levels
to calculate TL (Section 2.3).
3.1.1. Haro Strait
Figures 11 and 12 show single-frequency TL measured on AMAR D for the north and south
tracks, respectively; Figures 13 and 14 show measured TL from AMAR C for the north and
south tracks, respectively.
Figure 11. Measured single-frequency TL along the Haro Strait north track to AMAR D (shallow).
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 19
Figure 12. Measured single-frequency TL along the Haro Strait south track to AMAR D (shallow).
Figure 13. Measured single-frequency TL along the Haro Strait north track to AMAR C (seabed-mounted).
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
20 Version 4.0
Figure 14. Measured single-frequency TL along the Haro Strait south track to AMAR C (seabed-mounted).
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 21
3.1.2. Lime Kiln
Figure 15 shows single-frequency TL along the west track in Haro Strait for AMAR B.
Figure 15. Measured single-frequency TL along the Haro Strait west track to AMAR B at Lime Kiln.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
22 Version 4.0
3.1.3. Victoria Pilot Site
Figure 16 shows single-frequency TL along the southwest track near Victoria for AMAR E.
Figure 16. Measured single-frequency TL along the southwest track near Victoria to AMAR E.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 23
3.1.4. Deltaport Terminal
Figures 17 and 18 show single-frequency TL near Deltaport Terminal along the north and west
tracks, respectively. The TL for the 2,500 m station northwest of AMAR A (i.e., the station that
did not align with either the north or west tracks) is shown in Figure 19.
Figure 17. Measured single-frequency TL along the north track near Deltaport Terminal to AMAR A.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
24 Version 4.0
Figure 18. Measured single-frequency TL along the west track near Deltaport Terminal to AMAR A.
Figure 19. Measured single-frequency TL for the station 2,500 m NW of AMAR A.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 25
3.2. Transmission Loss Model
3.2.1. Geoacoustic Inversion
TL measurements were inverted for the compressional speed, density, and attenuation of the
seabed near Deltaport Terminal, in Haro Strait, and near Victoria (Section 2.4). Table 6 lists the
TL tracks used for each site, and the model-measurement mismatch and geoacoustic properties
that provided the best match to TL measurements. The accuracy of the geoacoustic properties are
discussed in Section 4.1. TL measurements on AMAR D were not used for the inversion because
the source-receiver range was not accurately known as the AMAR depth was controlled by tidal
currents. Figure 20 shows examples of the TL measurements and predictions at four frequencies
using the inversion results for the Victoria TL transect.
At higher frequencies, i.e., above 10 kHz, sound does not penetrate to deeper subbottom layers
and its reflectivity is governed mainly by the properties of the top few metres of seabed
sediments. Figure 21 shows single-frequency TL measurements versus range for all transects at
frequencies above 10 kHz. Because spherical spreading with absorption matched TL
measurements well at frequencies above 5 kHz, that method was used in the source level back-
propagation calculations (Sections 3.3 to 3.5) to calculate 1/3-octave-band TL for bands at 6.3
kHz and above.
Table 6. Geoacoustic properties at each site that result in the best match of measured single-frequency TL to model predictions.
Site Tracks Best rms Mismatch (dB)
Compressional Speed (m/s)
Density (g/cm3)
Attenuation (dB/λ)
Deltaport Terminal
North of AMAR A,
West of AMAR A 5.8 1,502 1.54 0.61
Haro Strait
North of AMAR C,
South of AMAR C,
West of AMAR B
6.1 1,541 1.80 1.79
Victoria Southwest of AMAR E
4.8 1,558 1.64 0.83
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
26 Version 4.0
Figure 20. Measured (blue symbols) and modelled (black lines) TL for tones at 316, 501, 794, and 1995 Hz along the southwest transect near Victoria. Spherical spreading loss (20 Log R) is shown as red lines.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 27
Figure 21. Single-frequency TL measurements for Deltaport Terminal (AMAR A), Lime Kiln (AMAR B), Haro Strait (AMARs C and D), and Victoria pilot site (AMAR E) tracks along with spherical spreading (20 Log R) plus absorption (α) for frequencies above 10,000 Hz.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
28 Version 4.0
3.2.2. One-third-Octave-Band TL—Haro Strait Shipping Lanes to Lime Kiln
One-third-octave-band TL was modelled from vessels in the northbound and southbound
shipping lanes of Haro Strait to the location of the Whale Museum and Beam Reach hydrophone
off Lime Kiln. Table 7 lists the source and receiver coordinates. The source was modelled at
5.5 m depth and the receiver was at 9 m depth in 10 m of water. The water depth along the track
was obtained from digital bathymetry charts and the range-dependent acoustic propagation
model RAM was used to compute TL as described in Section 2.4.
In order to provide an independent verification of the model predictions, TL to Lime Kiln was
measured by subtracting container ship received levels measured on AMAR B from source
levels of the same container ships measured on AMAR C and D (see Section 3.3). One-third-
octave band received levels were averaged over a 30 second period when three container ships
were at CPA along the southbound lane. The source-receiver ranges to AMAR B were 4.2, 4.4,
and 4.2 km for the Osaka Express, CMA CGM Attila, and Zim Los Angeles, respectively. The
locations of the container ships at CPA were slightly different than source point used in the TL
model. The difference in propagation path was expected to introduce only a small error (< 2.6 dB
rms) into the comparison because the bathymetry between the CPAs and Lime Kiln was very
close to that of the modelled transect from the southbound lane.
Table 7. Source and receiver coordinates for modelling TL from the northbound and southbound Haro Strait shipping lanes to Lime Kiln.
Location Latitude Longitude Distance to Lime Kiln Hydrophone (km)
Source Depth (m)
Receiver Depth (m)
Northbound lane 48°30.254’ N 123°10.932’ W 2.500 5.5 N/A
Southbound lane 48°29.601’ N 123°12.623’ W 4.908 5.5 N/A
Lime Kiln Hydrophone
48°30.931’ N 123°9.175’ W N/A N/A 9
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 29
Figure 22. Comparison of modeled and measured 1/3-octave-band TL between the southbound Haro Strait shipping lane and Lime Kiln. Modeled TL was computed using RAM at frequencies ≤5 kHz and using 18 log R + absorption at frequencies ≥6.3 kHz. TL measurements were computed from container ship received levels on AMAR B. Frequency bands where the background noise on AMAR B obscured sound from the container ships were excluded from the analysis.
The computed TL is compared with the modelled TL from the southbound lane in Figure 22. The
TL modelled using RAM showed good agreement with the computed TL at frequencies between
125 Hz and 5 kHz. At lower frequencies (≤ 100 Hz) the SNR of the container ship data was poor
due to the high levels of flow noise on AMAR B (see Section 3.6) and due to the low-frequency
cut-off effect caused by the near-shore bathymetry drop-off at Lime Kiln. As a consequence, TL
could not be measured below 50 Hz for two of the container ships, and the remaining low-
frequency TL data exhibited a substantial amount of scatter. Despite this fact, the comparison in
Figure 22 suggests that RAM overestimated the TL below 100 Hz between the southbound lane
and Lime Kiln. This is likely because the geoacoustic model in RAM was based on an inversion
of TL measurements above 300 Hz, and could not resolve deeper subbottom layering that would
typically influence low frequency TL in this environment.
Above 5 kHz, spherical spreading was found to slightly overestimate the TL between the
shipping lanes and Lime Kiln. A least-squares fit of an N log R geometrical spreading loss law to
the TL measurements above 5 kHz determined that an 18 Log R curve, plus a frequency-
dependent absorption coefficient, gave the best fit to the data in Figure 22. Therefore TL from
the southbound and northbound lanes to Lime Kiln was calculated using RAM at frequencies
≤5 kHz and using 18 Log R with absorption at frequencies ≥6.3 kHz. The 1/3-octave band TL
values from the two shipping lanes to Lime Kiln are shown in Figure 23 and listed in Appendix
A.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
30 Version 4.0
Figure 23. One-third-octave-band TL from the northbound and southbound Haro Strait shipping lanes to the Whale Museum and Beam Reach hydrophone off Lime Kiln. See Table 7 for source and receiver coordinates.
3.3. Container Ship Source Levels
Container ships’ transit speeds in Haro Strait are high (~20 knots) relative to their speeds near
the Victoria pilot site, where the vessels slow down to rendezvous with much smaller boats for
pilot transfer. Sound levels from three container ships transiting southbound were measured at
both locations to determine source levels for vessels transiting at different speeds. Additional
container ship sound levels were recorded at Haro Strait and Victoria, however sound level
recordings of the Osaka Express, CMA CGM Attila and Zim Los Angeles were the highest quality
and therefore selected for detailed analysis. Table 8 lists container ship specifications; the vessel
drafts are the last drafts received on AIS before each operation. Tables of 1/3-octave-band source
levels are listed in Appendix B.1.
Table 8. Container ship specifications for source levels of transiting vessels.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 31
3.3.1. Osaka Express
Figure 24 shows sound levels versus range for the Osaka Express transiting at 22 kts past
AMARs C and D in Haro Strait. Figures 25 and 26 show decade-band SPLs versus time and
spectrograms from recordings on these respective recorders. The spectrograms clearly show that
received levels 5 minutes prior to and 5 minutes after the time of CPA (approximately 3 km) are
higher on AMAR D than on AMAR C due to low-frequency flow noise on the shallow recorder.
Figure 27 shows the PSD levels measured at the CPA.
Figure 28 shows sound levels versus range for the Osaka Express transiting at 9 kts past
AMAR E off Victoria. Figure 29 shows decade-band SPLs versus time and a spectrogram from
recordings on AMAR E. Figure 30 shows the PSD levels at the CPA. One-third octave-band TL
was calculated for the source-receiver geometry at CPA for each AMAR (Section 2.4). Source
levels were calculated by taking the average received levels around the CPA (Section 2.5.1) and
adding them to TL (Figure 31).
Figure 24. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Osaka Express while the vessel transited south through Haro Strait at 22.0 kts measured by (left) AMAR C and (right) AMAR D.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
32 Version 4.0
Figure 25. Spectrogram of Osaka Express (870 m CPA) measured by AMAR C at 219 m depth. Times are in UTC.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 33
Figure 26. Spectrogram of Osaka Express (850 m CPA) measured by AMAR D at 34 m depth. Times are in UTC.
Figure 27. PSD of the Osaka Express at CPA measured at 846 m horizontal range by (blue) AMAR C at 219 m depth and (red) AMAR D at 34 m depth. Levels were averaged over 10 one-second windows, with 50% overlap using a Hanning window.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
34 Version 4.0
Figure 28. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Osaka Express while the vessel transited south past the Victoria pilot site at a CPA (200 m) at speed of ~9.0 kts, measured on AMAR E.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 35
Figure 29. Spectrogram of Osaka Express (200 m CPA) measured by AMAR E at 76 m depth. Times are in UTC.
Figure 30. PSD of the Osaka Express at CPA (200 m), measured by AMAR E. Levels were averaged over 10 one-second windows, with 50% overlap using a Hanning window.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
36 Version 4.0
Figure 31. Calculated 1/3-octave-band source levels for the Osaka Express transiting at two different speeds. Haro Strait values were averaged between AMARs C and D.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 37
3.3.2. CMA CGM Attila
Figure 32 shows broadband sound levels versus range for the CMA CGM Attila transiting at
24.1 kts past AMARs C and D in Haro Strait. Figures 33 and 34 are octave-band levels versus
time and spectrograms from recordings on these AMARs. Figure 35 shows the PSD levels at the
CPA.
Figure 36 shows sound levels versus range for the CMA CGM Attila transiting at 10.6 kts past
AMAR E off Victoria. Figure 37 shows decade-band SPLs versus time and a spectrogram from
recordings on AMAR E. Figure 38 shows the PSD levels at the CPA. One-third-octave-band TL
was calculated for the source-receiver geometry at CPA for each AMAR (Section 2.4). Source
levels were calculated by taking the average of the received levels around the CPA (Section
2.5.1) and adding them to TL (Figure 39).
Figure 32. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the CMA CGM Attila while the vessel transited south through Haro Strait at 24.1 kts measured by (left) AMAR C and (right) AMAR D.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
38 Version 4.0
Figure 33. Spectrogram of CMA CGM Attila including 1120 m CPA measured by AMAR C at 219 m depth. Times are in UTC.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 39
Figure 34. Spectrogram of CMA CGM Attila including 1100 m CPA measured by AMAR D at 72 m depth. Times are in UTC.
Figure 35. PSD of the CMA CGM Attila at CPA measured at 1098 m horizontal range by (blue) AMAR C at 219 m depth and (red) AMAR D at 72 m depth. Levels were averaged over 10 one-second windows, with 50% overlap using a Hanning window.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
40 Version 4.0
Figure 36. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the CMA CGM Attila while the vessel transited south past the Victoria pilot site at CPA (360 m) at speed of 10.6 kts, measured on AMAR E.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 41
Figure 37. Spectrogram of CMA CGM Attila including 360 m CPA measured by AMAR E at 76 m depth. Times are in UTC.
Figure 38. PSD of the CMA CGM Attila at CPA (360 m), measured by AMAR E. Levels were averaged over 10 one-second windows, with 50% overlap using a Hanning window.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
42 Version 4.0
Figure 39. Calculated 1/3-octave-band source levels for the CMA CGM Attila transiting at two different speeds. Haro Strait source level measurements were averaged between AMARs C and D.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 43
3.3.3. Zim Los Angeles
Figure 40 shows sound levels versus range for the Zim Los Angeles transiting at 22.4 kts past
AMARs C and D in Haro Strait. Figures 41 and 42 are octave-band levels versus time and
spectrograms from recordings on AMARs C and D, respectively. Figure 43 shows the PSD
levels at the CPA.
Figure 44 shows sound levels versus range for the Zim Los Angeles transiting at 11.1 kts past
AMAR E off Victoria. Figure 45 shows decade-band SPLs versus time and a spectrogram from
recordings on AMAR E. Figure 46 shows the PSD levels at the CPA. One-third-octave-band TL
was calculated for the source-receiver geometry at CPA for each AMAR (Section 2.4). Source
levels were calculated by taking the average of the received levels around the CPA
(Section 2.5.1) and adding them to TL (Figure 47).
Figure 40. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Zim Los Angeles while the vessel transited south through Haro Strait at 22.4 kts measured by (left) AMAR C and (right) AMAR D.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
44 Version 4.0
Figure 41. Spectrogram of Zim Los Angeles including 830 m CPA measured by AMAR C at 219 m depth. Times are in UTC.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 45
Figure 42. Spectrogram of Zim Los Angeles including 810 m CPA measured by AMAR D at 43.8 m depth. Times are in UTC.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
46 Version 4.0
Figure 43. PSD of the Zim Los Angeles at CPA measured at 794 m horizontal range by (blue) AMAR C at 219 m depth and (red) AMAR D at 43.8 m depth. Levels were averaged over 10 one-second windows, with 50% overlap using a Hanning window.
Figure 44. Broadband (10 to 64,000 Hz) SPL (rms) versus range from the Zim Los Angeles while the vessel transited south past the Victoria pilot site at CPA (67 m) at speed of 11.1 kts, measured on AMAR E. The higher levels at 600 m range are due to additional noise from a pilot boat.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 47
Figure 45. Spectrogram of Zim Los Angeles including 67 m CPA measured by AMAR E at 76 m depth. Times are in UTC.
Figure 46. PSD of the Zim Los Angeles at CPA (67 m), measured by AMAR E. Levels were averaged over 10 one-second windows, with 50% overlap using a Hanning window.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
48 Version 4.0
Figure 47. Calculated 1/3-octave-band source levels for the Zim Los Angeles transiting at two different speeds. Haro Strait values were averaged between AMARs C and D.
3.4. Berthing and Loading Source Levels
Noise from two berthing-related operations (berthing and loading) was analysed to determine
their representative source levels. Table 9 lists the container ship specifications for each
operation. The vessel drafts are the last drafts received on AIS before each operation. Source
levels are listed in Appendix B.2.
Table 9. Container ship specifications for berthing and loading operations.
Ship Length (m)
Container Ship Draft (m)
Operation Location Date (2013)
Vienna Express with tugs Seaspan Raven and Seaspan Resolution
335 12.7 Berthing Deltaport Terminal
16 June
CMA CGM La Scala 334 11.5 Loading Deltaport Terminal
18 June
3.4.1. Container Ship Berthing
Measurements of the tugs Seaspan Resolution and Seaspan Raven berthing the container ship
Vienna Express were made at Deltaport Terminal on AMAR A. Received levels were high-pass
filtered at 50 Hz to remove background noise from tidal currents. Received levels as a function
of time are shown in Figure 48 and PSD levels at the loudest period of the berthing operation are
shown in Figure 49.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 49
Bathymetry data and source-receiver geometry were insufficiently detailed at Deltaport Terminal
to use the acoustic model RAM to accurately predict TL during the berthing operation, so TL in
1/3-octave-bands was instead calculated using the assumption of simple spherical spreading loss
(20 Log R). During the berthing operation, the position of the Seaspan Resolution was logged by
an on-board observer using a hand-held GPS. Positions for the Seaspan Raven and the Vienna
Express were unavailable, so source to receiver range for these calculations were made using the
positions of the Seaspan Resolution, assuming all three vessels were close together. Received
broadband levels for the entire duration of the berthing operation (30 minutes) were averaged, as
well as the period of highest received levels during the berthing operations, which lasted 2
minutes (see Figure 48 and Figure 49). These two sets of averaged received levels were added to
TL to get 1/3-octave-band source levels (Figure 50). Source levels are not provided for
frequencies below 50 Hz, due to the high-pass filtering.
Figure 48. Broadband (50 Hz to 64 kHz) received levels as a function of time of the Seaspan Resolution and Seaspan Raven berthing the Vienna Express. Range is between the Seaspan Resolution and the AMAR. Data marked in red indicate the 30 minute time window of data used for computing average source levels. Times are in UTC.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
50 Version 4.0
Figure 49. PSD of the Seaspan Resolution and Seaspan Raven berthing the Vienna Express. The spectral measurements were computed from averaging received levels over the loudest section (2 minutes, red line, 1,596 m slant range) and over the section of the tugs pushing easy to full (30 minutes, blue line, 1,607 m slant range).
Figure 50. Calculated 1/3-octave-band composite source levels for Seaspan Resolution and Seaspan Raven berthing the Vienna Express. Source levels were calculated from received levels averaged over the loudest section (2 minutes, red line, 1,596 m slant range) and over the section of the tugs pushing easy to full (30 minutes, blue line, 1,607 m slant range).
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 51
3.4.2. Container Loading
Figure 51 shows the spectrogram of noise from container loading, measured as the acoustics
work boat R/V Coastal Geosciences drifted alongside the container ship CMA CGM La Scala
from the stern to the bow. Because the hydrophone was suspended directly below the R/V
Coastal Geosciences, noise from the boat’s depth sounder contaminated the acoustic data above
18 kHz. An 18 kHz low pass filter was therefore applied to the data to remove the depth sounder
noise above 18 kHz before computing broadband levels from the container loading.
Broadband levels are shown as a function of perpendicular distance to the container ship
(Figure 52). The R/V Coastal Geosciences was repositioned twice to obtain measurements along
the entire length of the container ship. Sound levels varied with time, perpendicular distance
from the ship, and distance along the ship.
Several tones associated with machinery and/or engines on the CMA CGM La Scala dominated
the broadband received levels. Sound levels were highest at ~20:29 UTC (corresponding to
“Time 1” annotation in Figure 51) where the perpendicular range to the container ship was
minimum and the R/V Coastal Geosciences was adjacent to the container ship’s stacks. At
“Time 3”, an unknown engine or piece of machinery turned on, increasing levels at frequencies
above 1 kHz. Occasional impulses from containers landing on the deck of the container ship
were louder than container ship engine and machinery noise (e.g., Times 4 and 5 when the R/V
Coastal Geosciences was near the bow of the container ship).
For determining source levels of container loading, received levels were averaged over a 30
second period centred at Time 2; this time window resulted in the highest continuous source
levels (source levels could occasionally be higher for one or two seconds when containers touch
down onto the ship). Received levels at frequencies above 18 kHz were extrapolated from levels
in the six 1/3-octave bands centred between 5 and 15.8 kHz.
The averaged received levels were back-propagated using the wavenumber integration model
with the source depth of 5.75 m (half the container ship’s draft), water depth of 20 m, receiver
depth of 10 m, geoacoustic parameters from the TL measurements, and averaged water column
properties from CTD measurements. The 1/3-octave-band source levels for loading the CMA
CGM La Scala are shown in Figure 53 and listed in Appendix B.2.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
52 Version 4.0
Figure 51. Spectrogram of container ship loading measurements. Gaps in the spectrogram correspond to times when the acoustics work boat R/V Coastal Geosciences was repositioned. Time is in UTC.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 53
Figure 52. Broadband (up to 18 kHz) received SPL as a function of perpendicular distance to the CMA CGM La Scala.
Figure 53. 1/3-octave band source levels for loading of the container ship CMA CGM La Scala. Source levels above 16 kHz (in red) were calculated from linear extrapolation of received band levels between 5-16 kHz
3.5. Harbour Tug Source Levels
Noise from the harbour tug Seaspan Resolution was measured while it performed the following
activities near Deltaport Terminal:
• Transiting at 4.0 kts;
• Transiting at 7.5 kts;
• Transiting at 12 kts;
• Berthing simulation: half power;
• Berthing simulation: full power; and
• Berthing simulation: acceleration.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
54 Version 4.0
3.5.1. Transiting
For the 4 and 7.5 kt transits, received levels were high-pass filtered at 20 Hz to remove flow
noise on the hydrophone and background noise originating from the nearby Roberts Bank
Terminals. Acoustic data for the 12 kt transit exceeded the amplitude limit of the recorder at
ranges less than 168 m, and therefore were excluded from the source level analysis. Received
levels as a function of slant range for the transiting measurements are shown in Figure 54. PSD
levels at the CPA are shown in Figure 55.
One-third octave-band TL was calculated using the source-receiver geometry at CPA (Section
2.4) for the 4 and 7.5 kt transits. Source levels were calculated by taking the average of the
received band levels near CPA (Section 2.5.1) and adding them to corresponding band TLs
(Figure 56). Source levels are not provided for frequencies below 20 Hz for these two transit
scenarios due to the high-pass filtering. Source levels for the 12 kt transit were computed from
measurements before and after CPA (Section 2.4), while received levels were within the
recorder’s valid operating amplitude range. These two measurements were averaged for each
1/3-octave-band (Section 2.5.1) and used to compute the source levels.
Figure 54. Received levels as a function of slant range to the Seaspan Resolution transiting at 4 kts, 7.5 kts, and 12 kts. Data at distances less than 168 m for the 12 kt transit are not shown as they exceeded the amplitude limit of the recorder.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 55
Figure 55. PSD from 5-second averages near CPA, of the Seaspan Resolution transiting at 4 kts at 48.9 m slant range, at 7.5 kts at 50.1 m slant range, and 12 kts at 168 m slant range.
Figure 56. Calculated 1/3-octave-band source levels for Seaspan Resolution transiting at 4, 7.5, and 12 kts.
3.5.2. Berthing Simulation
The Seaspan Resolution performed half power and full power berthing simulations by
positioning the two propellers toward the sides of the tug (pushing against each other). The
engines in the half power simulation were at approximately 608 rpm, and those in the full power
simulation were at approximately 795 rpm. The Seaspan Resolution also accelerated from 7.1 to
11.9 kts with engines at approximately 891 rpm. Received levels were high-pass filtered at 20 Hz
to remove background noise (flow noise on the hydrophone and background noise originating
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
56 Version 4.0
from the nearby Roberts Bank Terminals) for the half power and acceleration scenarios. Time
plots of received sound levels measured during the berthing simulations are shown in Figure 57.
PSD levels, averaged over 5 seconds centred around the CPA, for the Seaspan Resolution
performing berthing simulations are shown in Figure 58.
One-third octave-band TL was calculated for the source-receiver geometry at CPA (Section 2.4)
for the three berthing simulation scenarios. Source levels were calculated by taking the average
of the received levels around the CPA (Section 2.5.1) and adding them to TL (Figure 59). Source
levels are not provided for frequencies below 20 Hz for the half power and acceleration scenarios
due to the high-pass filtering.
Figure 57. Broadband (20 Hz to 64 kHz) received SPLs as a function of time as the Seaspan Resolution performed berthing simulation at half power (top left), full power (top right), and acceleration (bottom left).
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 57
Figure 58. PSD from 5 seconds of averaged data centred at CPA, of the Seaspan Resolution performing berthing simulations at half power at 214 m slant range (blue), at full power at 219 m slant range (red), and accelerating at 294 m slant range (orange).
Figure 59. Calculated 1/3-octave-band source levels for Seaspan Resolution performing berthing simulations at half power (blue), full power (red), and accelerating (orange).
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
58 Version 4.0
3.6. Ambient Noise Levels
Acoustic recordings from AMARs A to F were analysed to determine PSD levels and decade-
band SPLs over the entire deployment duration for each of the five measurement sites. Figures
60 to 65 show time-dependent spectrograms (lower panels) and decade-band SPLs (upper
panels) of the acoustic data collected at each AMAR. Figure 66 shows the cumulative statistical
distribution of 1 minute PSD levels at each AMAR. The presence of strong, low-frequency tones
throughout the data indicate that shipping and other anthropogenic sources were the dominant
source of ambient noise at all measurement stations. The diel periodicity of noise levels at many
of the stations is associated with the day-night cycle of human activity. In general, ambient noise
levels were strongly related to the amount of shipping present and the proximity of shipping to
the recorder.
Tidal currents were responsible for the noise peaks at frequencies below 200 Hz in the ambient
data. This is primarily pseudo-noise caused by turbulent flow around the hydrophones during
periods of high current velocity. Periods of high flow noise appear as ~24 hour, low-frequency
fluctuations in the ambient noise levels, which are particularly prominent on AMARs B, D, and
F. Tidal currents also induced strong, low-frequency tonal vibrations on some of the moorings
(e.g., at 140 Hz on AMAR B and 120 Hz on AMAR D), which were picked up by the
hydrophones. During slack-tide periods, the ambient measurements below 200 Hz were free of
pseudo-noise interference and these time periods are representative of the true ambient noise
fields.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 59
Figure 60. Deltaport Terminal (AMAR A): (top) Broadband and decade band SPL (30 minute average) from 30 May to 19 Jun; (bottom) Ambient noise spectrogram (30 minute average) over the same period. Frequency scale is logarithmic.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
60 Version 4.0
Figure 61. Lime Kiln (AMAR B): (top) Broadband and decade band SPL (30 minute average) from 29 May to 11 Jun; (bottom) Ambient noise spectrogram (30 minute average) over the same period. Frequency scale is logarithmic.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 61
Figure 62. Haro Strait (AMAR C): (top) Broadband and decade band SPL (30 minute average) from 31 May to 21 Jul; (bottom) Ambient noise spectrogram (30 minute average) over the same period. Frequency scale is logarithmic.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
62 Version 4.0
Figure 63. Haro Strait (AMAR D): (top) Broadband and decade band SPL (30 minute average) from 31 May to 21 Jul; (bottom) Ambient noise spectrogram (30 minute average) over the same period. Frequency scale is logarithmic.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 63
Figure 64. Georgia Strait (AMAR F): (top) Broadband and decade band SPL (30 minute average) from 22 Jun to 2 Jul; (bottom) Ambient noise spectrogram (30 minute average) over the same period. Frequency scale is logarithmic.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
64 Version 4.0
Figure 65. Victoria pilot site (AMAR E): (top) Broadband and decade band SPL (30 minute average) from 29 May to 17 Jun; (bottom) Ambient noise spectrogram (30 minute average) over the same period. Frequency scale is logarithmic.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 65
Figure 66. Exceedance percentiles of ambient noise PSD (1-minute average) at the five monitoring stations. The Nth percentile corresponds to the sound level that was exceeded by N% of the data. Note that narrow-band peaks in the spectra above 25 kHz were due to low-level electronic noise on the AMARs and are not real features of the ambient noise field.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
66 Version 4.0
4. Discussion and Conclusions
4.1. Transmission Loss Measurements
TL measurements at the centre frequencies of 1/3-octave bands between 200 Hz and 20 kHz
from the southbound shipping lane in Haro Strait to Lime Kiln, indicated relatively higher losses
at low frequencies than other sites. This is an expected result since the Lime Kiln AMAR B
recorder was in shallow water just a few tens of metres from shore. Low frequency energy is
absorbed through interactions with the seafloor as it enters shallow water near the shore. TL was
generally lowest at the Victoria pilot site for frequencies less than a few kHz. The geoacoustic
inversion showed this site had the highest compressional wave speed in the seafloor, which
would reflect more sound energy back into the water.
Geoacoustic inversions were performed to estimate the acoustic parameters of seafloor layers
that influence acoustic TL. The inversion results indicate that the seabed sediments are
acoustically absorptive at all sites. The compressional wave speeds are consistent with soft,
unconsolidated silts and clays (Table 6). Because the TL data were not very sensitive to sediment
density and attenuation coefficient, those parameters were less well-defined by the inversion.
Higher uncertainties in those parameters do not adversely affect the model predictions. Water
depth and the AMAR height above the seafloor were more important parameters and they were
better-constrained by the inversion algorithm. The model-measurement misfit was lowest when
AMARs A and B were within 0.1 m of the seafloor, which is less than the actual ~3 m height of
the moorings. A very soft and non-reflective top-seafloor layer could account for this
discrepancy. Water currents may also have pushed the moorings lower.
Jones and Wolfson (2006) modelled ship noise in Haro Strait at 3.6 kHz and investigated the
effects of bottom and surface roughness and geoacoustic properties on the expected received
levels. They found that a soft subbottom of sand/mud (with sound speed of 1,550 m/s) matched
measurements of ship noise better than harder subbottoms of rock and sand. They also found that
sea surface roughness and bottom roughness did not substantially change results and that
bathymetry effects dominated. These findings are consistent with this study’s geoacoustic
inversion results and observations of bathymetry effects on container ship noise measurements
(discussed in the next section).
4.2. Container Ship Measurements
4.2.1. Received Sound Levels
Received levels from container ships were strongly affected by the bathymetry in Haro Strait.
Broadband received levels decreased by as much as 15 dB when upslope bathymetry (e.g., a
seamount) was between the container ship and the AMAR. Sound energy is absorbed into the
sub-bottom when it encounters steep upslope bathymetry. For example, sound levels from the
Osaka Express decreased by about 15 dB as the container ship passed a seamount at 2 km range
that obstructed the source-receiver path (see Figure 24).
Figures 67 to 69 show received levels for the Osaka Express, CMA CGM Attila, and the Zim Los
Angeles passing through Haro Strait. The plots also indicate times corresponding to 0, 45, 60,
and 75 degrees off broadside in the forward and aft directions as well as the bathymetry along
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 67
each vessel’s path. Received levels were similar in the forward and aft directions although levels
from the forward direction were sometimes higher (up to 10 dB at the same range), which is
most likely due to bathymetry effects; the seamount north of AMARs C and D was farther than
the seamount south of the AMARs, causing received levels in the forward direction to be higher
at a farther range than levels in the aft direction (all three vessels were travelling southbound).
Any potential source directionality effects will likely have much lower magnitude compared to
these bathymetry effects.
Figure 67. Plot of rms SPL and water depth versus time along the vessel track to examine directionality of the Osaka Express, measured in Haro Strait by AMAR C. Times and ranges when the angle from the ship’s broadside to AMAR C is 0°, 45°, 60°, and 75° are annotated on the top plot. Times before and after 0° correspond to the forward and aft directions, respectively.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
68 Version 4.0
Figure 68. Plot of rms SPL and water depth versus time along the vessel track to examine directionality of the CMA CGM Attila, measured in Haro Strait by AMAR C. Times and ranges when the angle from the ship’s broadside to AMAR C is 0°, 45°, 60°, and 75° are annotated on the top plot. Times before and after 0° correspond to the forward and aft directions, respectively.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 69
Figure 69. Plot of rms SPL and water depth versus time along the vessel track to examine directionality of the Zim Los Angeles, measured in Haro Strait by AMAR C. Times and ranges when the angle from the ship’s broadside to AMAR C is 0°, 45°, 60°, and 75° are annotated on the top plot. Times before and after 0° correspond to the forward and aft directions, respectively.
In Haro Strait, noise levels measured at the seabed differed from those measured near the sea
surface for frequencies below 30 Hz because the fast tidal currents in Haro Strait caused low
frequency flow noise on the near-surface hydrophone (e.g., Figure 27). There were no substantial
differences in received levels above 30 Hz.
Jones and Wolfson (2006) measured sounds from ships in Haro Strait. Their closest
measurement of an unidentified cargo ship was at 1 km range; the received levels were
approximately 90 and 80 dB re 1 µPa2/Hz at 3.6 and 10.4 kHz respectively. These levels agree
with the PSD levels for the CMA CGM Attila and Zim Los Angeles at approximately 1 km range
in Haro Strait. Because the Osaka Express produced less cavitation noise, its levels were lower
by approximately 20 dB.
The opportunistic nature of this study’s measurements limited knowledge of certain parameters
such as engine speeds and propulsion system parameters. AIS data were used to determine vessel
positions relative to the recorders, but readings were often minutes apart, leading to some
positional uncertainty. Frequency dependent directivity was observed for the Osaka Express and
Zim Los Angeles at the Victoria pilot site (Figures 29 and 45, respectively). The band level
versus time plots show maxima of high frequency received sound levels occurred before the
maxima of lower frequencies. This suggests that high frequency sounds were emitted from a
source located forward of the machinery that emitted the majority of lower-frequency sounds
(presumably the propellers). The spectrograms for all three container ships at the Victoria pilot
site (Figures 29, 37, and 45) show times of abrupt changes in levels at all frequencies that are not
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
70 Version 4.0
associated with changes in tone frequencies. The increases in levels are likely due to increased
loading on the propellers as the ships accelerated. Received levels in the 1 to 10 kHz and 10 to
64 kHz bands for the Osaka Express transiting in Haro Strait (Figures 25 and 26) varied by up to
17 dB in a periodic pattern, oscillating approximately 2.25 times every minute. The cause of this
oscillation is unknown but it is an interesting phenomena.
4.2.2. Source Level Measurements
Source levels for the container ships transiting at constant speed (22 to 24 knots) in Haro Strait
were higher than those for the same ships accelerating (9 to 11 knots) at the Victoria pilot site.
Because of increased loading on the propellers during acceleration, the source levels at slower
speeds are likely higher than they would have been had the vessels been travelling at constant
speed.
The broadband source levels for all three container ships transiting at high speed in Haro Strait
were similar (within 4 dB); however, the levels above 1 kHz for the Osaka Express were
approximately 20 dB lower than those from the CMA CGM Attila and Zim Los Angeles. The high
frequency source levels for the Osaka Express were higher when it accelerated at 9 knots at the
Victoria pilot site than when it transited at a constant speed of 22 knots in Haro Strait. This was
not observed with the other two container ships and it is likely caused by increased cavitation
from high load on the propellers as the vessel accelerated. These results emphasise the
importance of cavitation in determining radiated noise levels at high frequencies.
Broadband source levels from this study are 15 to 20 dB higher than mean source levels reported
by McKenna et al. (2012) for container ships transiting at 20 knots. Comparison of the 1/3-
octave bands shows that these measurements are consistent with McKenna et al. above 200 Hz,
but are higher from 20 to 160 Hz (Figure 70). McKenna et al. used a simpler TL model (20 Log
R) in their back-propagation calculation, but this is not the primary reason for the source level
difference, since this study’s results would only have been 3 to 5 dB lower using their approach.
Therefore, this study’s results suggest that the three container ships measured generate more low-
frequency noise than the container ships measured by McKenna et al. (2012).
Figure 70. 1/3-octave band source levels for the three container ships and the average container ship source levels reported in McKenna et al. (2012).
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 71
Spectral received levels for a cargo ship at 1.5 km range, reported by Bassett et al. (2012), were
similar to measurements recorded of the CMA CGM Attila and Zim Los Angeles at approximately
1 km range, although this study’s measurements showed higher levels at frequencies below
50 Hz. The broadband source levels for container ships in Bassett et al. (2012) were between 181
and 186 dB re 1 µPa, whereas the broadband levels for the three container ships measured in this
study were between 200 and 206 dB re 1 µPa. Bassett et al. (2012) calculated source levels by
back-propagating received levels using a 15 Log R TL model. They justified their choice based
on parabolic-equation model predictions at 50, 100, and 250 Hz. This study’s measurements and
modelling support 20 Log R as being a more suitable TL model. Using 20 Log R instead of
15 Log R for Basset et al.’s closest measurement (1 km range) would add 15 dB to their source
levels, making their results consistent with this study’s source level measurements.
4.3. Tugs, Berthing, and Loading Measurements
Source levels for the harbour tug Seaspan Resolution increased with its transit speed. Cavitation
noise at frequencies above 3 kHz varied substantially (by as much as 45 dB) between the 4 and
12 knot scenarios (Figure 56). Source levels for the simulated berthing scenarios were
approximately 18 dB lower when at half power than at full power (Figure 59). Source levels for
acceleration were between the half and full power levels, but tended toward the full power levels.
Cavitation noise levels (> 1 kHz) measured during the berthing simulation, where cavitation was
induced by opposing the twin propellers on the tugs, were higher than those measured during the
actual berthing operation (tugs Seaspan Resolution and Seaspan Raven berthing the container
ship Vienna Express), particularly above 10 kHz (Figure 50). Source levels for the half power
berthing simulation were lower than the source levels from the berthing operation except at
frequencies above 10 kHz.
The measurements of loading the container ship CMA CGM La Scala at Deltaport Terminals
showed that the highest continuous noise levels originated from the aft of the vessel, near the
engine stacks, with the maximum occurring in the 50 Hz 1/3-octave band. Received levels were
briefly higher when containers were lowered to and contacted the ship’s deck, particularly in
measurements near the bow of the ship.
4.4. Ambient Noise
Ambient acoustic measurements showed that human-generated sounds, primarily from ship
traffic, dominated ambient noise at all locations measured. The overall spectral trend of median
ambient PSD levels (Figure 71) was consistent at all five measurement stations. Differences
below 200 Hz were primarily due to varying levels of tidally-induced pseudo-noise (water flow
noise on hydrophones) between sites. Ambient levels were highest near Deltaport Terminal
(AMAR A), which was situated directly adjacent to three busy marine terminals (Deltaport
Terminal, Westshore Terminals, and the BC Ferries Terminal at Tsawwassen). Similarly,
ambient levels were high at the Victoria pilot site (AMAR E), which was situated just outside the
Victoria and Esquimalt harbours. Ambient noise levels were ~5 to 10 dB lower on the remaining
four AMARs, which were situated near the inbound and outbound shipping lanes in Georgia
Strait and Haro Strait. The shapes of the ambient spectra were consistent between all sites, with a
mean slope of −7.1±0.7 dB per octave.
Underwater Acoustic Measurements in Haro Strait and Strait of Georgia JASCO APPLIED SCIENCES
72 Version 4.0
A comparison of the data from AMARs C and D in Haro Strait showed that ambient levels
measured at the seabed and near the surface were almost identical above 200 Hz. Therefore
ambient noise measurements obtained using seabed-mounted hydrophones in this environment
are expected to be relevant for assessing noise effects for marine mammals present near the sea
surface. Seafloor-mounted hydrophones do not require vertical mooring cables that often vibrate
and introduce contaminating noise.
Figure 71. Comparison plot of median ambient noise levels measured at the five monitoring stations.
JASCO APPLIED SCIENCES Underwater Acoustic Measurements in Haro Strait and Strait of Georgia
Version 4.0 73
Literature Cited
ANSI S1.1-1994. R2004. American National Standard Acoustical Terminology. American National Standards
Institute, New York.
Bassett, C., B. Polagye, M. Holt, and J. Thomson. 2012. A vessel noise budget for Admiralty Inlet, Puget Sound,
Washington (USA). Journal of the Acoustical Society of America 132(6):3706-3719.
Coppens, A. B. 1981. Simple equations for the speed of sound in Neptunian waters. Journal of the Acoustical
Society of America 69(3):862-863.
François, R. E., and G. R. Garrison. 1982a. Sound absorption based on ocean measurements: Part I: Pure water and
magnesium sulfate contributions. The Journal of the Acoustical Society of America 72(3):896-907.
François, R. E., and G. R. Garrison. 1982b. Sound absorption based on ocean measurements: Part II: Boric acid
contribution and equation for total absorption. Journal of the Acoustical Society of America 72(6):1879-
1890.
Jensen, F. B., W. A. Kuperman, M. B. Porter, and H. Schmidt. 2000. Computational ocean acoustics. AIP Series in
Modern Acoustics and Signal Processing. AIP Press - Springer.
Jones, C. D., and M.A. Wolfson. 2006. Acoustic environment of Haro Strait: Preliminary propagation modeling and
data analysis. APL-UW TM 3-06. Washington University Seattle Applied Physics Lab, Washington. 53 p.
