Boat Source-Level Noise in Haro Strait: Relevance to Orca Whales

L. Galli, B Hurlbutt, W. Jewett, W. Morton, S. Schuster, Z. Van Hilsen

Abstract

Using a calibrated underwater hydrophone array located in the Haro Strait of Puget Sound, boat traffic noise was categorized and analyzed. Underwater sound spreading models were developed and applied to the collected boat noise observations. The boat noise levels at a reference distance of one meter from the sound source, shows that at higher engine RPM there is both a higher sound level and a the sound energy is distributed over a wider frequency. The source-level noise produced by the small boats observed, varies from 141dB to 161dB and the broadest frequency range observed was between 0.86 kHz to 8.0 kHz. Large commercial ships produced source levels of 180dB to 188dB with a frequency range from 0.1 kHz to 8 kHz. Both large ships and small boats produce frequencies that overlap with Orca vocalizations. Further study is needed to determine how the noise produced by boat traffic is affecting Orca behavior and communication.

I. Introduction

The Haro Strait, located between San Juan Island and Vancouver Island in the Puget Sound, hosts a variety of marine life as well as a large human population. The strait’s most spectacular marine species is the Orca whale. Three pods of resident Orcas, fish eating whales that generally do not travel far, live in Haro Strait feeding on its salmon. These whales communicate by producing a rapid series of clicks used for navigation and echolocation of salmon (Balcomb) along with underwater vocalizations that may, among other things, alert each other to the presence of prey. Research is being done in Haro Strait to better understand the whales and their complex communications, which may shed light on the recent decline in the Southern Resident Orca population.

The Haro Strait is Western Canada’s principle shipping channel as well as a major recreational destination. Now, nearly a million more people live in the area than did ten years ago, increasing the strait’s traffic to hundreds of boats each day. As recreational travelers, fisherman, logs, sand, oil, and goods from Asia are transported through the strait, marine life may be negatively affected.

Human activity in Haro Strait produces underwater noise within the hearing and vocalization range of Orcas. At high intensities, this may have a large impact on the whales, but no studies have been conducted on human-made underwater noise levels in this region. The goal of our study was to collect the underwater source noise levels of boats using controlled experiments for small commercial, recreational, and research boats and in situ observations of large commercial ships passing through the strait. We recorded sound samples under a variety of conditions. Sound frequencies and intensities were then analyzed to determine the received noise level as a function of distance between the sound source and the recording hydrophone. This information was used to create a model for the transmission of underwater sound that predicts Orca noise exposure (intensity and frequency spectra) as a function of vessel operating conditions and range.

II. Materials and Methods

Recordings were conducted in the Haro Strait of Puget Sound with an array of four hydrophones over a three-week period in March 2003. The hydrophone array diagramed in Figure 2 is ordinarily used to localize Orca calls. It consists of one ITC-6050C hydrophone and three ITC-4066 hydrophones, both made by International Transducer Corporation (www.itc-transducers.com). The 6050C hydrophone is equipped with an internal preamp while the 4066 hydrophones use a custom built preamp designed to eliminate noise. These custom preamps, built by Colorado College students, use an AD524 chip to produce a gain of 100. The hydrophones are sensitive to frequencies ranging from 100 Hz-10 kHz, covering the range of most boat noise and Orca vocalizations.

Computer programs developed by Dr. Val Veirs calculate the root mean square voltage and the frequency spectrum for signals from each hydrophone. The voltage displayed at the computer, V_c(rms), is a combination of the ambient ocean noise, V_b(rms), and the sound source we are measuring, V_s(rms). In order to accurately measure the sound level of a source, the background noise must be factored out. Statistical analysis of the root mean square yields[1],

V_s(rms) = (V²_c(rms) - V²_b(rms) )^(1/2).

A. Obtaining Sound Source Intensities from Measured Voltages

To convert an RMS voltage at the computer into a sound intensity at the source, we measure the voltage gain from the hydrophone to the computer, convert the rms voltage at the hydrophone to an in water sound intensity at the hydrophone, and calculate the transmission loss of the underwater sound wave as it propagates from the source to the hydrophone (Fig. 3).

Sound Þ Transmission Loss Þ Conversion Þ Amplifier Gain Þ Conversion

Wave from Spreading to Volts - Cable Loss to rms Volts

Fig.3 The path taken by a sound wave from a point source to a V_(rms) at the computer.

Underwater sound spreads out as it propagates from a source. Transmission loss from the sound source to the hydrophones can be calculated using both a spherical and cylindrical spreading model. Both of these models are based on the idea that the sound intensity of the total sound surface area is conserved as it spreads. To conserve energy,

I_sA_s = I_hA_h.

Energy is conserved if the intensity at the source, I_s, multiplied by the initial surface area of the wave, A_s, is equal to the intensity at the hydrophone, I_h, multiplied by the final surface area, A_h.

When the depth of the water is greater than the distance from the source to the hydrophone, sound propagates as a perfect sphere of radius R and area 4pR² (Fig 4).

The sound intensity at a reference distance, R_s, from the source is found in terms of the intensity at the hydrophone and the distance from the source to the hydrophone, R_h.

I_s = I_hR²_h/R²_s

In our model the reference distance for source levels is R_s = 1 meter. Converting to decibels, where dB = 10logI, and simplifying,

dB_s = 10logI_h + 20logR_h.

As outlined later, we know the pressure at the hydrophone, not the intensity. The intensity is proportional to the pressure squared. Converting to pressure, P, where 10logI = 20logP,

dB_s = 20logP + 20logR_h.

A cylindrical spreading model may more accurately describe sound propagation when the depth of the water is less than the distance from the sound source to the hydrophone (Fig. 5). The sound will spread spherically until it reflects off of the ocean floor and surface. The speed of sound in water differs greatly from the speed of sound in the air and the ocean floor. Therefore, we treat the sound energy as totally reflected when encountering both the surface and the bottom. Bound vertically, the sound intensity only spreads horizontally creating a cylinder of radius R, height H, and area 2pRH.

Fig. 5 As sound travels through a shallow channel, it reflects off the surface and floor. After a certain distance, the sound is dispersed evenly in a vertical column.

Once again using the conservation of energy at R_s = 1, and canceling the common term 2pH,

I_s = I_hR_h

dB_s = 10logI_h + 10logR_h

The sound wave spreads spherically for some distance until transforming into a cylindrically spreading wave (Urick). The term 10logL is added to this equation to account for the fact that cylindrical spreading does not begin at the source[2].

dB_s = 10logI_h + 10logR_h+10logL

In terms of pressure,

dB_s = 20logP + 10logR_h+10logL.

The sound wave compresses a crystal in the hydrophone and generates a voltage, V_h. This voltage is equal to a constant, the open circuit receiving response, multiplied by the pressure. The open circuit receiving response, C_r, is found on the hydrophone spec-sheet in logarithmic form. Solving for the pressure and taking C_r out of logarithmic form gives, P_h, the pressure at the hydrophone.

P_h = V_h/10^{^(Cr/20)}.

The voltage travels by wire from the hydrophone to an operational amplifier, which amplifies the signal before entering a National Instruments Board inside a Pentium PC. The board allows a computer to measure the voltages on the line. Voltage is attenuated on the long wires and amplified by the operational amplifier and the computer, resulting in a final voltage gain, g. V_h is equal to the final voltage at the computer, V_c, divided by the gain, so substituting this into the equation for pressure at the hydrophone,

P_h = V_c /(g*10^{^(Cr/20)})

Substituting this pressure into the spherical and cylindrical models, the intensity of the sound source is given in decibels.

dB_s = 20log (V_c(rms)/(g_*10^{^(response/20)})) + 20logR_h (Spherical)

dB_s = 20log (V_c(rms)/(g_*10^{^(response/20)})) + 10logR_h + 10logL (Cylindrical)

Water absorbs sound waves, but, consulting Urick, we concluded the waves are not significantly attenuated over the distance traveled in this experiment.

B. Calibration

The hydrophone array was calibrated by developing a sound intensity spreading model that fit data from an already calibrated speaker. To find the most accurate model, we used a Lubell LL916 underwater speaker (www.lubell.com) with a calibration curve from the U.S. Navy (Fig. 6). We broadcast a variety of tones and pulses that varied in amplitude, frequency, direction, depth, and distance from the hydrophones. We recorded all of the sounds simultaneously with all four hydrophones. After adjusting the signals received for both distance and gains and losses in the system using formulae stated above, we applied cylindrical spreading models for each hydrophone. We compared our results with the calibration curve of the speaker, and found that a cylindrical spreading model with a spherical to cylindrical spreading transition distance, L, equal to the depth at the hydrophone, most accurately fit. Though all four hydrophones had matched the Lubell transmit voltage response graph well, hydrophone 3, or H-3, fit the graph the best. (Fig. 7) We therefore used H-3 and the cylindrical spreading model, where L = 19m, to analyze our data. H-3 has a gain of g = 1500, and an open circuit receiving response of C_r = -193dB.

All of our testing occurred within 200 meters of the hydrophone array. Our spreading models therefore do not apply to sounds that originate from much greater distances. This means that ships traveling at approximately two kilometers over depths of up to 600 meters and variable ocean floor slopes require a different model. A spherical spreading model yielded results that most closely agreed with previously published results. Source levels from these ships are therefore only accurate as a relative comparison of one ship to another.

Fig. 6 The spec sheet for the Lubell LL916 speaker graphs transmit voltage response versus frequency on a logarithmic scale.

Fig. 7 The transmitted voltage response for H-3 calculated with the cylindrical spreading model in red with standard deviation error bars, closely fits the speaker’s actual transmitted voltage response shown in blue.

C. Data Collection

We made four passes over H-3 with six boats selected to represent a variety of boats that are often operated close to Orca whales. Passes were made back and forth from north to south, or parallel to the shore, and east to west, or away from and towards the shore. At least three recordings were made on each pass. A constant RPM was maintained during each pass, but RPM was changed from high (cruising) to low (moving slowly) from pass to pass. We also recorded each boat while idling. We chose to record high RPM and idling to get the full range underwater noise for each boat. We also recorded a lower RPM to have noise data for typical whale watching operating conditions. GPS positions were marked for each sound sample recorded at the computer. Ambient noise level was recorded for each situation. Boat type, engine type, and engine horsepower were noted for each boat.

We also made recorded the sounds from five large commercial vessels transiting Haro Strait. When recording passing ships we estimated the range with the reticule in a pair of Leitz binoculars. Operating conditions and engine types are unknown, but we photographed the tankers and noted their boat type, size, and direction of travel.

III. Results

The experimental results are summarized in figure 8. Data was taken for the controlled boats on April 10, 2003, under near calm, quiet ocean conditions. Boat type and operating conditions are noted. Data was taken for the tankers under variable conditions on April 4 and April 7. Boat type is noted. Nothing is known about the operating conditions of the tankers. For all boats, the sound intensity at one meter from the source is given in decibels along with a range of frequencies that significantly contributed to the sound intensity.

A sound frequency spectrum displays the intensity of sound at various frequencies versus time. The spectrum indicates at what frequencies the sound is loudest. From the spectrum in Figure 9 we can see that the Annie Mae produces sound at a wide range of frequencies, .8-8 kHz, while the ship produced constant low frequency, 0-1 kHz, sound with a spike throughout higher frequencies at repeating intervals.

It is difficult to understand what a sound file will sound like just from looking at the amplitude (voltage) vs. time.[3]
Controlled Boats:

Boat Name	Boat Type	Engine Type, Horsepower	Movement relative to H-3	RPM	dB/ umPa/V @1m	Frequency Range (kHz)
Cat’s	Catamaran	Westerbeck	none (idle)	950	145	0.2 – 2
Cradle		Diesel, 27 HP	parallel	2000	147 + 2	0.2 – 3.1
			parallel	3000	154 + 3	0.9 – 4.4
			away	3100	151 + 1	0.5 –3.6
			towards	3100	154 + 3	0.1 – 3.5
Stellar Sea	Whale	Twin 3208	none (idle)	700	148 + 1	0.2 – 1.8
	Watch	Caterpillar	parallel	1900	158 + 1	0.2 – 1.6
		inboard, 210	parallel	2400	162 + 3	0.2 – 4.6
		HP	away	1900	156 + 1	0.2 – 2.5
			towards	1900	157 + 1	0.2 – 3.0
Auklet	Flat	Mercury 90	none	Idle	141 + 2	0.5 – 1.7
	Bottom	outboard, 90 HP	parallel	Medium speed	146 + 2	0.1 – 1.8
			parallel	High speed	163 + 2	0.1 - 3
			away	Medium speed	144 + 3	0.1 – 1.1
			towards	Medium speed	143 + 1	0.1 – 1.3
Orca[4]	Flat	4-stroke	none (idle)	600	159	0.1 - 2
	Bottom	outboard,	parallel	3000	159 + 5	0.1 – 3.64
	Boston	115 HP	parallel	4500	156 + 2	0.2 – 4.4
	Whaler		away	3000	158	0.1 – 4.2
			towards	3000	158 + 1	0.1 – 5.1
Putt-Putt	Dinghy	Mercury, 4	none	Idle	140	0.4 – 1.7
		HP	parallel	Full throttle	144 + 5	0.2 – 1.2
			away	Full throttle	143 + 0	0.1 – 1
			towards	Full throttle	141 + 1	0.1 – 1.2
Annie Mae	Hard Top	Yamaha outboard,	none (idle)	900	135 + 2	0.1 – 2
		200 HP	parallel	2900	154 + 1	0.26 – 5.5
			parallel	4400	161 + 3	0.86 – 8.0
			away	3100	155 + 2	0.1 – 4.1
			towards	3100	153 + 1	0.1 – 4.8

Commercial Shipping Boats:

Name	Operation	Range (m)	Frequency range (kHz)	dB/ umPa/V @1m	dB/u/mPa/V@100m
China Shipping Line	Full container, southbound	3000	.1 - 5.4