OceTHE OFFICIALa MAGAZINEn ogOF THE OCEANOGRAPHYra SOCIETYphy

CITATION Dziak, R.P., J.H. Haxel, H. Matsumoto, T.-K. Lau, S. Heimlich, S. Nieukirk, D.K. Mellinger, J. Osse, C. Meinig, N. Delich, and S. Stalin. 2017. Ambient sound at Challenger Deep, Mariana Trench. Oceanography 30(2):186–197, https://doi.org/10.5670/ oceanog.2017.240.

DOI https://doi.org/10.5670/oceanog.2017.240

COPYRIGHT This article has been published in Oceanography, Volume 30, Number 2, a quarterly journal of The Oceanography Society. Copyright 2017 by The Oceanography Society. All rights reserved.

USAGE Permission is granted to copy this article for use in teaching and research. Republication, systematic reproduction, or collective redistribution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: [email protected] or The Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA.

DOWNLOADED FROM HTTP://TOS.ORG/OCEANOGRAPHY REGULAR ISSUE FEATURE

Ambient Sound at Challenger Deep, Mariana Trench

By Robert P. Dziak, Joseph H. Haxel, ABSTRACT. We present a record of ambient sound obtained using a unique deep-ocean Haruyoshi Matsumoto, instrument package and mooring that was successfully deployed in 2015 at Challenger Tai-Kwan Lau, Sara Heimlich, Deep in the Mariana Trench. The 45 m long mooring contained a hydrophone and Sharon Nieukirk, David K. Mellinger, an RBR™ pressure-temperature sensor. The hydrophone recorded continuously for 24 days at a 32 kHz sample rate. The pressure logger recorded a maximum pressure of James Osse, Christian Meinig, 11,161.4 decibars, corresponding to a depth of 10,829.7 m, where actual anchor depth was Nicholas Delich, and Scott Stalin 10,854.7 m. Observed sound sources included earthquake acoustic signals (T phases), baleen and odontocete cetacean vocalizations, ship propeller sounds, airguns, active sonar, and the passing of a Category 4 typhoon. Overall, Challenger Deep sound levels in the ship traffic band (20–100 Hz) can be as high as noise levels caused by moderate shipping, which is likely due to persistent commercial and military ship traffic in the region. Challenger Deep sound levels due to sea surface wind/waves (500 Hz to 1 kHz band) are as high as sea state 2, but can also be very low, equivalent to sea state 0. To our knowledge, this is the first long-term (multiday to week) broadband sound record, and only the fifth in situ measurement of depth, ever made at Challenger Deep. Our study indicates that Challenger Deep, the ultimate hadal (>6,000 m) environment, can be relatively quiet but is not as acoustically isolated as previously thought, and weather- related surface processes can influence the soundscape in the deepest parts of the ocean.

186 Oceanography | Vol.30, No.2 INTRODUCTION level measurements collected (Barclay these unique environments has turned Challenger Deep is an 11 km long, et al., 2009; Barclay and Buckingham, to free-falling landers (SOI, 2014) and 2.25 km wide, east-west trending basin 2010). Also, in 2010, researchers at the specialized deep-ocean moorings that located in the Mariana Trench, ~300 km Lamont-Doherty Earth Observatory remain stationary once they reach the southwest of the island of Guam within (LDEO) designed a deep-ocean seafloor seafloor (Dziak et al., 2015). Although the territorial waters of the Federated seismometer and pressure case capable of this approach is constrained to discrete States of Micronesia (Figure 1). Because surviving to extreme depths. The LDEO spatial sampling (point sampling) during Challenger Deep is the deepest point platform was deployed near Challenger each deployment, the ability to collect in the world ocean (10,984 m; Gardner Deep; however, seismo-acoustic sensors long-term information on time-varying et al., 2014), the extreme hydrostatic were not included in the test deployment physical and biological processes enables pressures there make obtaining even syn- (Maya Tolstoy, LDEO, pers. comm., 2013). baseline characterization of the environ- optic physical and biological measure- Thus, recordings of ambient sound levels mental conditions and marine ecosys- ments of the deep-sea environment dif- at Challenger Deep remained elusive. tems at hadal zones (Taira et al., 2004). ficult. Challenger Deep has, however, From July to November 2015, a team The ocean is nearly transparent to been the focus of renewed in situ explo- of National Ocean and Atmospheric low-frequency sound, and even weak ration as exemplified by recent dives by Administration (NOAA), University of sound signals can propagate over long both the Woods Hole Oceanographic Washington, and Oregon State University distances with little loss in signal strength Institution (WHOI) autonomous under- scientists and engineers designed, built, (Munk, 1994). The term “soundscape” water vehicle Nereus in 2009 (Lippsett deployed, and recovered a deep-ocean is used here to describe the acous- and Nevala, 2009) and James Cameron’s hydrophone and pressure sensor mooring tic sources and signals that are present solo dive in a manned submersible on at Challenger Deep. The goal of this proj- at a particular location and time; these March 26, 2012 (Than, 2012). Although ect was to record ambient sound levels in sounds are typically broken into three many important scientific observations this ultimate hadal zone environment in broad source categories: biological, geo- were made during these expeditions, order to establish baseline sound levels in physical, and anthropogenic (Pijanowski, rigorous ambient sound level measure- what was expected to be one of the most 2011). Moreover, man-made noise in ments at these extreme ocean depths acoustically isolated ocean ecosystems. the ocean is thought to have increased were not collected. Ambient sound mea- This paper presents the 24-day-long by a factor of three to four (or 9–12 dB) surements can provide insights into the hydrophone record of ambient sound in areas near major shipping lanes since health of a marine ecosystem by offer- successfully recorded at Challenger the 1960s, largely due to increases in ves- ing a picture of the diversity of vocaliz- Deep, discusses the temporal charac- sel traffic transiting the world ocean and ing marine animals and a sense of what ter of the various natural and man-made an increase in the gross tonnage of ships natural sources of sound contribute to sound sources, and compares the overall used in the modern shipping fleet (Frisk the ocean soundscape, and they can help sound levels at Challenger Deep to global et al., 2003; Hildebrand, 2009). Sounds us gauge the impact of man-made noise sound averages. We also present a direct generated by anthropogenic activities in the deep ocean. pressure-​based measurement of ocean such as shipping, oil and gas explora- There have been other recent efforts to depth made during the mooring deploy- tion, and military sonar can be harmful deploy deep-ocean moorings with sen- ment, one of only five direct depth mea- to marine mammals that rely on low-​​ sor packages to probe the world’s ocean surements ever made at Challenger Deep. frequency sound to send and receive trenches. In 2009, a team from Scripps acoustic-based information. For exam- Institution of Oceanography (SIO) EXPLORATION AND ple, studies show that increased sound deployed an untethered instrument plat- SOUNDSCAPE SIGNIFICANCE levels from anthropogenic activities can form equipped with a 30 kHz hydro- Hadal zones at deep-ocean trenches are hinder marine mammal communication phone (depth rated to 11 km), designed among the least explored environments (Hatch et al., 2012), alter communication to record during controlled descent on Earth. Research in these extreme envi- behavior (Parks et al., 2012), change loco- from the surface (Barclay et al., 2009). ronments is challenging, and scientific motive behavior (Goldbogen et al., 2013), A series of descents were made with discoveries are limited by access to the and induce stress (Rolland et al., 2012). To this instrument in the Mariana Trench few exploration vehicles that are capable what extent anthropogenic sounds might (maximum depth of 9,000 m) and in of working at depths >7,000 m. Given the impact marine animals at hadal zone depths the Philippine Sea (6,000 m), providing, recent loss of the deep submergence vehi- is largely unknown. at that time, the deepest ambient sound cle Nereus (WHOI, 2014), exploration of

Oceanography | June 2017 187 HISTORICAL BACKGROUND relatively few direct measurements of was made that this was the deepest point In 1875, HMS Challenger became the depth at Challenger Deep using in situ at Challenger Deep. first ship to fully explore the Mariana pressure sensors. The first direct mea- What is clear from these previous Trench, sounding the abyss with explo- surement of depth occurred during the efforts is that it is very difficult to first sives that provided a depth estimate of historic dive of the manned bathyscaphe find the exact spatial coordinates of the 8,184 m (Gardner et al., 2014). In 1952, the Trieste in 1960. TheTrieste crew, the first deepest location and then to make an HMS Challenger II crew used explosives, humans to visit Challenger Deep, used accurate and precise measurement of the a hand-held stopwatch, and a wireline an onboard pressure sensor to measure a full ocean depth. Echosounder and multi- sounding machine to estimate the depth calibrated depth of 10,911 m (Piccard and beam sonar can provide only an averaged at 10,862 m (Carruthers and Lawford, Dietz, 1961). The next in situ measure- model of seafloor depth because the mea- 1952). Other noteworthy echosounder- ment did not occur until 2002, when the surements are (1) made from a rolling based measurements ensued over Japan Agency for Marine-Earth Science and pitching surface ship, and (2) reliant the decades, including estimates of and Technology (JAMSTEC) deployed on the accuracy of water-column sound (1) 11,034 m from the research ship the remotely operated vehicle (ROV) speed models. However, in situ pressure Vityaz (USSR) in 1957, (2) 10,933 m from Kaiko (Therberge, 2008). The ROV measurements collected by a free-falling R/V Thomas Washington (USA) in 1976, recorded a depth of 10,896 m, again using instrument package can also be driven (3) 10,920 ± 10 m by R/V Kairei (Japan) an onboard pressure sensor; however, no away from the correct depth location by in 1984, (4) and 10,903 m by R/V Kilo claim was made that this was the maxi- prevailing currents, and robotic and/or Moana (USA) in 2008 (Gardner et al., mum depth at Challenger Deep. The last manned submersibles can drift from the 2014). After comprehensive modeling, direct depth measurement, prior to this intended location despite efforts to main- the latest, most thorough effort using report, was performed by James Cameron tain position relative to the sea surface. echosounders occurred in 2011 from using his specially designed submers- USNS Sumner; it yielded an estimate of ible Deepsea Challenger. The National MOORING AND SENSOR the deepest point at 10,984 m ± 25 m Geographic Society (2012) reported that DESCRIPTION (Gardner et al., 2014). Cameron made a pressure-gauge-based The Challenger Deep mooring (see online In contrast, there are, understandably, estimate of 10,908 m. Again, no claim supplemental Figure S1) is 45 m in length and consists of a hydrophone instrument package, an RBR™ depth and temperature 142°8'E 142°10'E 142°12'E 142°14'E 142°16'E 9000 logger, and nine Vitrovex® glass spheres encased in plastic “hardhats.” The glass floats are specially designed for 12,000 m 10000 PMEL Full-Ocean Depth Mooring (2015) depth operations. The hydrophone and

11°22'N (Hydrophone and pressure sensor) pressure sensors are attached in a double-​ 10400 Depth = 10,854.7 m yoke frame to a 1 m long stainless steel 10600 rod shackled within the mooring for 10900 extra support. The top mooring assembly

10800 has an aluminum mast with a Xeos sat- 11°20'N ellite beacon (GPS/Iridium transmitter) 10600 and a flag for visibility in surface recov- Gardner Challenger Deep (2014) Depth = 10,984 m ery operations. Housed inside one of the two top floats is a redundant Xeos satellite

11°18'N beacon. This in-line system is attached to an anchor using a specially modified Km 100 Guam 10000 dual acoustic release package employed to increase the probability of recovery. 12°N The mooring is deployed from top to bot-

11°16'N 9600 UTM Zone 54 proj. map area tom using the anchor as weight to expe- 0 Km 2 142°E 144°E dite its free-fall descent to the seafloor. FIGURE 1. Bathymetric map of Challenger Deep, Mariana Trench (after Gardner et al., 2014). The For recovery, an acoustic release is trig- inset shows map location relative to Guam (~370 km from Challenger Deep). The yellow dot shows gered from the surface ship to detach the location of the deepest point estimated from bathymetry (Gardner et al., 2014). The red dot shows the surveyed location of the full-ocean depth hydrophone and pressure sensor mooring. the anchor and allow the platform to Map courtesy of S. Merle, NOAA/PMEL/EOI rise to the surface.

188 Oceanography | Vol.30, No.2 The NOAA/Pacific Marine Environ- this, the mooring shown in Figure 2 was the vertical velocity specification. All mental Laboratory (PMEL) designed a tested to ensure proper buoyancy ver- pressure-sensitive components, includ- hydrophone pressure housing capable of sus weight compensation to achieve the ing the hydrophone and the GPS bea- operational recording at full ocean depths optimum ascent/descent rates targeted con, were successfully tested at 11,000 m (Figure 2a,b). The pressure case is made of between 0.5–1.0 m s−1. The mooring hydrostatic pressure at Deep Sea Power 0.7 inch (1.8 cm) thick 6AL-4V titanium, system was tested in Puget Sound, and Light (San Diego, California) prior to allowing it to withstand the hydrostatic Washington, in 200 m of water to confirm shipment to Guam. pressure at 15,000 m depth, providing a factor of 1.5 safety margin. The low-​ a b power-consumption hydrophone elec- tronics are capable of continuous record- ing at a 32 kHz sample rate with 16-bit resolution, resulting in a three-week-long record of ambient sound. The data are stored on a 128 GB Compact Flash card, enabling rapid data transfer and backup upon recovery. The hydrophone used for this study is a low-power version of the HTI92WB/PC, with a built-in differential output manu- factured by High Tech Inc., Mississippi. c It is an oil-filled pressure-compensated hydrophone with a nominal element sen- sitivity of −182 dB re 1V/µPa, which can withstand extreme high pressure up to 20,000 psi. It has a relatively flat response below 1 kHz with two internal high-pass filters: one is mechanical and the other is electrical. The oil is exchanged between the inside and outside of the ceramic element through an orifice that serves as a first-​ order mechanical high-pass filter with a roll-off frequency of 7 Hz. The second internal high pass is a resistor-capacitor​ (RC) filter formed by the capacitance of the element and the input impedance of the first stage amplifier with a roll-off frequency at 50 Hz. It also has a built-in signal-​conditioning amplifier with a gain of 7 dB, which makes the hydrophone sensitivity −175 dB re 1V/µPa. The sen- d e f sitivity is omnidirectional at 13 kHz in an x-y plane. With the combined filters of the hydrophone and variable frequency gain of the pre-amp, the analog system’s noise floor stays below the Wenz sea state 0 noise curve up to 10 kHz. FIGURE 2. (a) Hydrophone and pressure mooring on deck of US Coast Guard Cutter Sequoia. Glass floats in orange plastic hardhats and the RF beacon mast are at the top of the picture, One caveat is that the hydrophone is dual releases are located at middle right, and the hydrophone pressure case and pressure sen- sensitive to rapid changes in pressure, sor, which are yoked to a 1 m long stainless steel rod shackled within the mooring, are at left. and its rate of descent/ascent should not (b) and (c) Hydrophone and mooring being deployed using the ship’s crane. The crane block is at −1 the top of the image with floats below it. (d) Top of mooring (flag, mast, and floats) floating at the exceed 5 m s or there is a risk of dam- sea surface after release from the seafloor. (e) Recovery of sensors and float onto ship using crane. aging the ceramic element. To account for (f) Hydrophone and pressure sensor on deck immediately after recovery.

Oceanography | June 2017 189 Mooring Deployment from transducer to release approached beacons transmitted as planned, and and Recovery true water depth. The mooring stopped messages subsequently received pro- During July 20–23, 2015, the mooring was descending once a depth (slant range) of vided the precise location of the moor- deployed from US Coast Guard Cutter 10,752.61 m was reached; however, this ing at the surface. The mooring was spot- Sequoia, which is based in Apra Harbor, estimate was based on assuming a sound ted at a distance of 1.5 km, with seven Guam. The target deployment site, the speed of 1,500 m s−1 over the entire water floats bobbing horizontally at the sea sur- deepest point identified by Gardner et al. column, which is somewhat slower than face and the mast and flag extended ver- (2014), is located in a relatively small the expected sound speed at depth. The tically in the air (Figure 2d). The mooring basin within the main east-west-trending ship next circled the anchor drop location was hoisted on board using the crane, and trough of Challenger Deep (Figure 1). A while several slant ranges to the moor- both instruments were found to be intact first attempt to deploy the hydrophone ing were collected. These ranges and GPS and undamaged (Figure 2e,f). mooring there was made during January ship locations were then used to derive to March 2015. Despite difficulties caused mooring location and depth using a sim- SENSOR PERFORMANCE by the presence of Typhoon Mekkhala, ple nonlinear regression algorithm and Pressure and Temperature the mooring was successfully deployed an 11 km deep sound speed profile avail- Data Loggers and recovered. However, the hydro- able on the MB-System seafloor mapping In addition to the robust mooring and phone stopped recording once a depth software website (Caress et al., 2015). Our release performance, the RBR™ pres- of 1,700 m was reached, likely due to a calculations showed the mooring was sure and temperature sensors also per- faulty power unit (Dziak et al., 2015). The located at 11°20.127'N, 142°12.0233'E. formed as designed. The pressure data hydrophone problem was corrected, and This location is roughly 1 km north- log shows the sensor was subjected to a a second deployment from Sequoia fol- east of the target location of the deepest maximum pressure of 11,161.4 decibars lowed in summer 2015. depth identified by Gardner et al. (2014; (see Figure S2). The sensor depth was Once on site, ocean surface current Figure 1). The bathymetric depth here is then estimated from pressure using the speed and direction were estimated at also consistent with the initial mooring Thermodynamic Equation of Seawater-10 1 knot, moving from east to west. The survey depth of ~10,750 m. (TEOS-10) relationships that can be ship was then positioned 1 km to east of The mooring recovery cruise, once derived from software routines available the target deployment site for the anchor again onboard Sequoia, took place during online to perform the calculations (IOC, SCOR, IAPSO, 2010). We simply input the Challenger Deep measurements of pressure in decibars, the measured tem- …the success of this deep-ocean mooring perature (2.45°C), and the practical salin- system demonstrates the value of employing the ity (34.6 ppt) at Challenger Deep avail- able from the US Navy’s Generalized relatively low-cost ocean sampling methodology Digital Environmental Model (Allen, “ used here (as compared to robotic or manned 2012). The latitude and longitude loca- tion of the sensor are also required; we submersibles) to recover data from otherwise used the acoustically ranged location inaccessible deep-sea areas. . of the mooring while on the seafloor (11°20.127'N; 142°12.0233'E). However, the dynamic height anomaly and sea sur- face geopotential are typically ignored drop. The anchor and mooring were in November 2–4, 2015, which was delayed in the TEOS-10 model when converting the water at 10:00 Guam time (Figure 2c). from” the original planned date because between pressure and depth. These calcu- Using an Edgetech 8011M deck unit and of another typhoon, Champi, that passed lations resulted in a maximum depth esti- transducer, we were able to range to the near the deployment site. On-site acous- mate of 10,829.7 m. Because the sensor specially modified Benthos 865A dual tic communication between the moor- is 25 m above the anchor on the moor- acoustic releases on the mooring as it ing and the transducer was excellent, ing, the actual anchor depth of the moor- descended through the water column at even with the exceptional water depth. ing is 10,854.7 m. To our knowledge, this the designed rate of 0.6 m s−1. With a water The mooring released immediately and is only the fifth in situ measurement of depth in excess of 10 km, it took ~6 hours began to rise to the surface at roughly depth ever made at Challenger Deep. for the mooring to reach the seafloor. twice its descent rate (~1.2 m s−1), reach- There is, however, the question of what As it descended, the acoustic slant range ing the surface in ~2.5 hours. The satellite uncertainty should be assigned to this

190 Oceanography | Vol.30, No.2 maximum depth estimate. McDougall temperature of 28.4°C was recorded in attenuation rates of these high-frequency et al. (2003) evaluated the TEOS-10 the surface waters prior to deploying the signals, for example, roughly 0.5 dB km−1 relationship’s fit to the equation of state anchor over the side of the ship, and a at 10 kHz (Ainslie and McColm, 1998); (e.g., Feistel and Hagen, 1995) over a minimum 1.5°C was recorded for a short and the fact that click sounds are unlikely range of pressures, temperatures, and time as the mooring reached the seafloor. to be directed at a sensor on the sea- salinity in the ocean water column. This However, the seafloor temperatures then floor, as these clicks are more commonly region forms a polygonal shape of sea­ stabilized at 2.45°C for the remainder emitted at small angles to the horizontal. water referred to as an “oceanographic of the deployment. Lastly, there is a clear increase in broad- funnel,” which represents a section of band sound levels beginning on August 1, ocean with realistic parameters of envi- Full-Ocean Depth Hydrophone 2015, that is sea surface wind and wave ronmental relevance from the sea surface During the four-month deployment, the noise caused by the Category 4 Typhoon down to the depth equivalent to a pressure hydrophone operated successfully at full Soudelor as it moved from east to west of 8,000 decibars. The TEOS-10 funnel-​ bandwidth (32 kHz) for 24 days (July 13 through the region north of Saipan. This test algorithm for the Challenger Deep to August 6, as planned), including part of the record shows that weather- depth measurement of 11,161.4 decibars recording the full descent from the sur- related surface processes can affect ambi- well exceeds the maximum decibar fun- face to 10,852 m while passing through a ent sound levels and clearly influence nel limit, which casts doubt on the abso- 25.9°C temperature range. Analysis of the the soundscape in the deepest parts lute accuracy of our depth estimate as hydrophone acoustic recordings reveals a of the ocean. presented here. However, the TEOS-10 number of interesting sound sources that model does state that the uncertainty in contribute to ambient levels at Challenger MARIANA TRENCH SOUND depth estimates is up to 4 m at 5,000 deci- Deep. A spectrogram of the entire 24-day PROPAGATION, SOURCES, bars. This implies that the uncertainty recording period (Figure 3) is dominated AND LEVELS in our depth estimate should be roughly by broadband seismo-acoustic energy To aid in our analysis of the individ- twice this value, or ±8.9 m. The only other from earthquake activity. The earth- ual sound sources identified in the method to infer uncertainty is to use the quake acoustic energy is focused under Challenger Deep record, we constructed manufacturer’s estimate of sensor accu- 4 kHz, but registers as high as 10–12 kHz. acoustic ray trace propagation mod- racy: ±0.05% for the full-scale depth at Additionally, anthropogenic sounds from els of signal loss using the software 11 km (https://rbr-global.com/products/ surface vessels, baleen whale vocaliza- Bellhop (Porter et al., 2011). Figure 4 sensors). This suggests an uncertainty tions, sperm whale clicks, and several shows three of these simulations for sea of ±5.4 m at the sensor depth, which odontocete whistles have been identi- surface sound sources, directly over- is less than the uncertainty from the fied within the record. Detection of the head and to the north and south of the model calculation. whistles and clicks is somewhat surpris- Challenger Deep location in the Mariana Further analysis of the pressure record ing given the generally low source level Trench. The propagation models indi- shows that there is a two- to three-week of odontocete whistles (e.g., Au, 1993; cate that in the deepest sections of the period during which the RBR™ gauge Soldevilla et al., 2008); the expected high Mariana Trench there is likely a 10–15 dB acclimates to the ambient pressure con- ditions. This acclimation period is seen 2 in Figure S2 as the gradual, asymptotic dB re 1 uPa /Hz 60 70 80 90 100 110 120 decline in recorded pressure during the first 40 days of the total deployment, 12,000 which, with the mean removed, levels 10,000 z) Sonar off at a Δ depth of −0.5 m. Additionally, Typhoon (H 8,000 the spring-neap and semidiurnal tidal Soudelor 6,000 cycles at Challenger Deep are well Earthquakes Baleen equen cy 4,000 Whale resolved in power spectral density esti- Fr 2,000 mates of the pressure time series (see Figure S2, bottom), indicating that 15-Jul-2015 20-Jul-2015 25-Jul-2015 30-Jul-2015 04-Aug-2015 even at these extreme ocean depths, FIGURE 3. Frequency spectra of the entire 24-day recording period (1 Hz to 12.5 kHz bandwidth). tidal processes dominate fluctuations in Short-duration (several minute) bursts of broadband energy are from ocean crust earthquakes ambient pressure. along the Mariana arc and trench. Relatively high-frequency energy on July 25 is from active sonar, which is short duration and in the 1 Hz to 10 kHz band. Broadband energy from August 2 to August 6 The temperature gauge on the is wind and surface wave noise from Typhoon Soudelor (Category 4) near Saipan. Continuous, nar- RBR™ also showed that a maximum row-band frequency signals are ongoing vocalizations of baleen whales.

Oceanography | June 2017 191 reduction in received level signal strength occurring deep within the oceanic crust moderate-sized earthquakes (mb ≥4) that from sound sources generated north and of the Mariana region will introduce are detected by global seismic networks south of the trench axis, where acoustic acoustic energy into the basin via prop- every few days. Thus, we expected the rays outside of the trench attenuate rap- agation paths through the trench floor or hydrophone would detect a significant idly due to multiple seafloor-sea sur- walls. The following sections describe in number of earthquakes, especially given face reflections. This observation is sup- more detail the variety of acoustic sources that the hydrophone was deployed in situ ported by previous acoustic studies at the recorded at Challenger Deep. and therefore would detect much smaller Tonga Trench that showed the bottom magnitude events. of the trench is shadowed from surface Geophysical Sources Figure 5a shows an example of the noise by the trench walls (Barlcay and The Mariana Trench is an active accre- acoustic arrival (T-phase) packet from

Buckingham, 2014). This implies that the tionary boundary where the Pacific a body wave magnitude (mb) 5.0 earth- sound sources comprising the baseline Plate is being subducted beneath quake occurring 50 km northeast of sound levels at Challenger Deep mostly the Philippine Plate at the rate of Challenger Deep on July 16, 2015, derive from sources at the surface and/ 1.0–4.6 cm yr−1 (Smoczyk et al., 2013). 23:28 GMT. The main shock is followed or in the water column directly above the Consequently, the Mariana Trench by dozens of aftershocks that are readily basin. However, it also seems likely that and the Challenger Deep are very seis- observed within the record. This record seismo-acoustic energy from earthquakes mically active regions and produce is typical of T-phase signals from earth- quakes recorded in the deep ocean, char- acterized by broadband energy (mainly 0 focused <100 Hz), with signal dura- 2,000 tions of 60–120 s. The onset of a T-phase 4,000 packet from a large earthquake will also usually exhibit a fast-propagating seismic 6,000 phase (P wave) that has traveled entirely Depth (m) 8,000 through oceanic crust. The P-wave sig- 10,000 nal is the low-frequency, impulsive sig- (a) North Source nal that arrives ~10 s before the T-phase –50 0 50 100 150 0 packet (Figure 5a). As noted earlier, the earthquake T-phase signal packets on the 2,000 long-term spectrogram are very high fre- 4,000 quency (≥10–12 kHz). Previous obser- 6,000 vations of seismo-acoustic energy are

Depth (m) not nearly as broadband, being typi- 8,000 cally under 100 Hz (e.g., Wilcock et al., 10,000 2011). The high frequency of the seismo- (b) Central Source acoustic energy observed here may be –150 –100 –50 0 50 0 caused by the very close proximity of the hydrophone to these large earthquake 2,000 sources and/or the high sample rate of 4,000 the full ocean-depth hydrophone, which 6,000 employs a sample rate not typically used

Depth (m) for seismic studies. 8,000 Figure 5b shows a histogram count 10,000 of earthquakes per day recorded on the (c) South Source Challenger Deep hydrophone. A total 250 200 150 100 50 0 of 3,814 T-phase signals were detected Range (km) during the 24-day recording period, with FIGURE 4. Ray tracing model simulations using the software Bellhop (Porter, 2011) with a 100 Hz source at 30 m depth (present near surface, water column source) from (a) north, (b) central, and an average of 159 T-phase events detected (c) south of the Challenger Deep basin. Red is high energy, blue is lowest energy, representing a per day. A total of 996 T-phase events had 60 dB relative transmission loss. Most of the energy from sound sources at the sea surface north P-wave seismic arrivals associated with and south of Challenger Deep is scattered prior to reaching the bottom of the trench, while acous- tic energy from a source directly above the trench axis propagates to full basin depth. Image cour- them, at an average of 41.5 P-wave detec- tesy of J. Caplan-Auerbach, Western Washington University tions per day. Typically, both P waves

192 Oceanography | Vol.30, No.2 dB re μPa2/Hz and T phases are recorded from local a 10 15 40 65 90 and regional earthquakes. Large (mb >7) 200 global earthquakes will produce detect-

z) 150 P-wave able P waves only (Dziak et al., 2004). At (H Challenger Deep, the majority of earth- 100 T-phase

quakes detected during the recording equen cy 50 period were local and/or regional events, Fr 0 while during the same time only a total of 1.5 six large mb 6.0–7.0 earthquakes occurred 1.0 worldwide. Therefore, almost all earth- 0.5 quakes detected generated P waves and lt s 0.0 Vo T phases, and the P-wave and T-phase –0.5 counts are well correlated. Moreover, a –1.0 –1.5 search of the US national earthquake 0 60 120 180 240 300 360 420 480 online database (https://earthquake. Seconds b usgs.gov) shows that a total of 84 earth- 30 quakes of 3.9 ≤ mb ≤ 5.3 occurred within T-phase Counts 1,000 km of Challenger Deep along 25 P-wave Counts the Mariana, Ryuku, and Philippine Trenches during the hydrophone deploy- 20 ment period. This implies the full-ocean depth hydrophone likely detected several 15 hundred more earthquakes below the Event Counts 10 mb 3.9 detection threshold of land-based seismic networks in the region. Figure 5c shows received lev- 5 els of T-phase signal packets. Because 0 there are several earthquake-generated 07/16 07/21 07/26 07/31 08/05 Date T-phase signals each day on the hydro- c phone record, we use the daily median 160 of the received T-phase energy lev- els (peak to peak) to represent the over- 140 all earthquake-produced sound levels in the 1–100 Hz frequency band. The daily z 120 median levels range from 79.9 dB to /H 2 102.9 dB re 1 μPa2/Hz during the 24-day a recording period, with an overall median 100 of 83.9 dB re 1 μPa2/Hz. Also, there are dB re μP several times during the 24 days when there are clear increases in the num- 80 ber of earthquakes and received energy levels: July 15, July 21, July 24, and 60 August 1. All of these increased acous- 07/16 07/21 07/26 07/31 08/05 Date tic energy levels correspond to clusters of FIGURE 5. (a) Time-series and spectrogram of the July 15, 2015, 5.0 m earthquake located three to five m 4–5 earthquakes near the b b ~50 km northeast of Challenger Deep at 10 km depth. The impulsive, broadband first arrival islands of Guam, Agrihan, and Farallon is the direct seismic P wave; a long-duration emergent T phase can be seen arriving one de Pajaros. We estimate the total acous- to two minutes later. Four aftershocks associated with this mainshock were also recorded. (b) Histogram of T-phase signal packets per hour recorded from local and regional earth- tic received level of all T-phase signals quakes (<500 km distance). Also shown are counts of P-wave earthquake arrivals. Because recorded over the 24-day period to be the majority of earthquakes detected were local to regional events and generated detect- 167.12 dB re 1 μPa2/Hz. Thus, natural seis- able P waves and T phases, the P-wave and T-phase counts are well correlated. All P- and T-phases signals were manually identified by an analyst. (c) Received levels of T-phase sig- mic energy is a significant contributor to nal packets from a 10-second window of maximum peak-to-peak amplitude over the 1 Hz to ambient sound levels at Challenger Deep. 100 Hz band. Highest levels correspond to large, local events.

Oceanography | June 2017 193 a 10 Biological Sources z) 8 The Mariana Island region is host to a wide vari- y (kH 6 ety of mysticetes (seven species) and odontocetes 4

equenc (22 species; Fulling et al., 2011). Figure S3a,b shows Fr 2 example spectrograms of odontocete and mys- 0 1.5 ticete vocalizations recorded on the Challenger 1.0 Deep hydrophone. The most common type of 0.5 baleen whale call observed resembles a call recently lts 0.0 Vo –0.5 recorded and characterized during acoustic glider –1.0 surveys over the Mariana Trench (southeast of –1.5 0 20 40 60 80 100 120140 160180 Guam and east of the Challenger Deep) in late 2014 Time (min) and early 2015 (Hill et al., 2016; Nieukirk et al., b 2016). These calls are characterized by a ~38 Hz har- 10 z) monic tone followed by a broad-​frequency metal- 8

y (kH lic sweep up to 7.5 kHz, and are considered most 6 likely to be from a minke whale (Nieukirk et al., 4 equenc 2016). Figure S3c shows a histogram of the percent Fr 2 0 of the day these vocalizations were observed on the 1.5 Challenger Deep hydrophone. Generally, the call 1.0 was seen on the hydrophone >5 times per day, the 0.5

lts 0.0 exception being July 24–25 and August 1–2 when Vo –0.5 calls decreased to <5 per day. –1.0 While many odontocete whistles are identified –1.5 010203040506070 within the record, we were not able to use the clicks Time (sec) to identify the variety of possible source species. c 10 However, given that both mysticete and odontocete z) 8 whales are not known to dive deeper than 3,000 m y (kH 6 (Baird et al., 2006), obtaining a record of their calls 4 on the seafloor at Challenger Deep demonstrates equenc

Fr 2 the favorable propagation conditions for acoustic 0 rays traveling vertically from the sea surface. 1.5 1.0 0.5 Anthropogenic Sources

lts 0.0 Over one-third of global shipping activity is made Vo –0.5 up of large (>10,000 gross tons) commercial ships –1.0 (container ships and product tankers) that produce –1.5 061207240 360 480 00 20840 tonal noise at ~40 Hz and bulk carriers that produce Time (sec) noise at ~100 Hz (McKenna et al., 2012). The com- 2 dB re μPa /Hz mercial port of Guam is a major shipping hub for the 5 10 20 40 50 northwestern Pacific and Micronesia. Guam is also FIGURE 6. Examples of time series and spectrogram of sound sources recorded at the location of US Naval Base Guam, a key military Challenger Deep. (a) An example of ship propeller sound (recorded on July 29, 2015) hub that is home to dozens of US Pacific Fleet units. seen in the spectra as a series of relatively narrow bands of energy that change fre- quency due to changing interference patterns as a ship passes over the hydrophone. Typically, ship propeller sounds are seen as continu- The time series shows ship sounds loud enough to clip the hydrophone record ous, discrete-frequency bands of energy in the spec- (that is, the amplitudes of the signals are truncated, or clipped, meaning the signals trogram at 2, 4, 6, and 10 kHz (and can be much exceeded the dynamic range of the recording system). (b) Four airgun bursts, each exhibiting short-duration (~1 s) energy over the 1 Hz to 8 kHz band. Airgun records are less than 40 Hz), although these bands can exhibit clipped as well. (c) Example of intense, active sonar pings recorded on July 25, 2015. constructive and destructive interference patterns Pings are each ~80 s in duration, occur at ~1.5 minute spacing, and are characterized of several kilohertz over relatively short (2–3 min- by a broadband, short-duration ping followed by three narrow-band pulses centered at ~2.85, 3.5, and 4.25 kHz. ute) periods (Figure 6, top). Figure 7a shows the histogram of the number of hours per day that ship propeller sounds were observed on the Challenger Deep hydrophone. With the exception of the time

194 Oceanography | Vol.30, No.2 when Typhoon Soudelor was in the region Overall Sound Levels world. This is consistent with the per- (August 3–5), ship propeller sounds were To place the sound levels observed at sistent military and commercial ship traf- observed 10–24 hours per day. Thus, Challenger Deep in context with global fic in and out of the Port of Guam. In the sounds of ship traffic are a signifi- averages, Figure 8 compares the ambient the 500 Hz–1 kHz wind-driven sea state cant contribution to the ocean sound- sound levels observed at Challenger Deep band, Challenger Deep ambient sound scape around Guam and at the southern with the global sound levels estimated in levels are ~40 dB to 60 dB re 1μPa2/Hz. Mariana Trench. Unfortunately, many of the 1960s. These curves (Wenz, 1962) Thus, noise levels at Challenger Deep due the records of the ship propeller sounds show the average sound spectral energy to wind/waves are relatively low at sea are clipped, which can affect the spectral levels for different measures of ship traf- states of 0 to 2, and therefore it seems that character of the received sound levels. fic (heavy to moderate) and sea state due Typhoon Soudelor did not have a sub- Seismic airguns, commonly used for to variation in wind speeds (Figure 8a). stantial impact on overall noise levels oil and natural gas exploration beneath Anthropogenically induced noise from during the recording period. Indeed, as the seafloor, are one of the main sources heavy shipping creates the highest sound can be seen in Figure 8b, ambient sound of anthropogenic sound below 100 Hz level values of ~90 dB re 1μ Pa2/Hz in the levels are so low that the hydrophone is (Tolstoy et al., 2004). Organized in 20–100 Hz band, whereas wind has high- recording mainly hydrophone system multi-unit arrays, each airgun gener- est noise levels of 70–75 dB re 1μ Pa2/Hz at noise starting at 5 kHz on the 10th per- ates a bubble that expands and then con- sea states 3 to 6 in the 500 Hz–1 kHz band. centile curve, and at 10 kHz on the 50th tracts, releasing pressurized air under- In comparison, Figure 8b shows the and 90th percentile curves. These very water and creating a loud transient signal Challenger Deep sound level distribu- low sound levels at Challenger Deep are (<0.1 s, 235–240 dB re 1 Pa at 1 m in the tions at the 10th, 50th, and 90th percentiles. not too surprising given the high atten- 2–188 Hz frequency band) that can pen- Challenger Deep exhibits sound levels of uation of acoustic energy from regional etrate the seafloor (Hatch and Wright, 70–80 dB re 1μ Pa2/Hz in the 20–100 Hz sound sources that are not directly above 2007). Figure 6b shows the hydrophone ship traffic band. Thus, noise levels from the Mariana Trench, as well as the low records of airgun bursts recorded at ship traffic at Challenger Deep appear sea states that can persist in the region Challenger Deep, and Figure 7b shows the equivalent to noise levels caused by between passing high-wind events. hours per day when airgun pulses were moderate shipping in other parts of the Lastly, in Figure 8b, an interesting observed. Airgun activity was relatively phn intermittent during this 24-day recording e period, occurring on seven different days 0 a between 1 and 10 hours on each of these 20 days. We estimated the median received 10 level of the airgun pulses (1 Hz–14 kHz frequency range) to be 99.4 dB re 1Pa. 0

In addition to ship propeller and air- 0 b gun noise, the Challenger Deep hydro- e Da 20 phone also recorded persistent signals

e 10 from active sonar, which we assumed to be military in origin (Southall et al., 0 0 2012). Each of these sonar signal pack- c ets are ~80 s in duration (and occur at 20

~1.5 minute spacing), and are charac- 10 terized by an initial broadband, short- me e 0 duration (~5–10 sec) burst followed by 0 three narrow band pulses centered at d ~2.85, 3.5, and 4.25 kHz. These following 20 pulses may also possibly be reflections/ 10 reverberations. The pings were observed 0 4–12 hours per day for four days from 12 1 26g02g0 July 22–25 (Figure 7c). We estimate the FIGURE 7. (a) Histogram showing the number of hours per day that ship propeller noise received levels of the broadband ping was observed on hydrophone record. (b) Number of hours per day airgun sounds were observed. (c) Number of hours active sonar was observed. (d) Number of hours and narrow-band pulses range from per day baleen whale calls were observed. This histogram is the same as Figure S3c, 58.1­–68.0 dB re 1Pa. and is shown here to highlight how it compares with anthropogenic source counts.

Oceanography | June 2017 195 (b) Challener ee (a) en () uluut 0 0 hp 80 80 e n Depenent )

tm 0 0 e 2 a 60 60 0 e pe 50 6 50

e 50 ( e 1 40 40 e e 0 1 0 10 ea tat 20 0 20 10 100 1,000 10,000 10 1001,000 10,000 een () een () FIGURE 8. (a) Diagram showing global estimates of sound levels due to ship traffic noise and wind-driven sea state (Wenz, 1962; adapted from NRC, 2003). Shaded areas highlight the 20–100 Hz “ship traffic” noise band (red lines), and the maximum sound levels due to surface winds and various sea states (blue lines; 500 Hz–1 kHz band). The green line shows the ambient sound level due to microseism background energy. (b) Challenger Deep sound level distributions at the 10th, 50th, and 90th percentiles. Shaded areas highlight the same frequency bands as Wenz curves to compare sound levels. Challenger Deep ship noise (90th percentile) can be as high as moderate ship-traffic noise in Wenz curves 2 (~70–80 dB re 1μPa /Hz). Maximum noise levels due to wind/waves in Challenger Deep are equivalent to sea states 2 th 0 to 2 (~60 dB re 1μPa /Hz at the 90 percentile). Spectral spikes at ~8 kHz and 12 kHz are due to instrument noise. Hydrophone system noise starts at 5 kHz in 10th percentile, and becomes 50th to 90th percentile above 10 kHz. Spectral peak at 40 Hz in 10th percentile band are part of a complex baleen whale call recently identified in the Mariana Trench by Nieukirk et al. (2016; Figure S3). The peak at 900 Hz is also likely a baleen whale call, but its origin is currently not known.

spectral peak can be seen at ~900 Hz in ocean sampling methodology used here Allen, R.L. Jr., and US Navy; Naval Oceanographic th Office. 2012. Global gridded physical profile the 10 percentile curve. We speculate (as compared to robotic or manned sub- data from the U.S. Navy’s Generalized Digital that this sustained signal is biological in mersibles) to recover data from otherwise Environmental Model (GDEM) product database (NODC Accession 9600094). Version 1.1. National origin, as it is unknown what else would inaccessible deep-sea areas. Oceanographic Data Center, NOAA. Dataset, produce a somewhat narrow-band peak Although based on a relatively short https://data.nodc.noaa.gov/cgi-bin/iso?id=gov.noaa. nodc:9600094. over almost the entire duration of the 24-day sample size, the overall hourly Au, W.W.L. 1993. The Sonar of Dolphins. Challenger Deep recording. In detailed counts of anthropogenic and natural Springer, New York, https://doi.org/10.1007/​ 978-1-4612-4356-4. review of the spectral data, no single call sound sources indicate earthquakes, ship Barclay, D.R., and M.J. Buckingham. 2010. can be clearly identified, the frequency noise, and baleen whale calls are a com- Ambient noise in the Mariana Trench. Journal of the Acoustical Society of America 128:2300, does not match that of any known whale mon component of the Challenger Deep https://doi.org/10.1121/1.3508084. call in the area, and the spectral peak soundscape. Additionally, it is clear that Barclay, D.R., and M.J. Buckingham. 2014. On the spatial properties of ambient noise in the Tonga appears merely as a rise in energy around sea surface wind and wave noise caused Trench, including effects of bathymetric shad- 900 Hz. The origin of this signal thus by large storms can penetrate to the owing. Journal of the Acoustical Society of America 136:2497, https://doi.org/10.1121/1.4896742. remains a subject of conjecture. However, deepest parts of the ocean and domi- Barclay, D.R., F. Simonet, and M.J. Buckingham. 2009. another spectral peak that can be seen nate the ambient sound field. Given Deep sound: A free-falling sensor platform for th depth-profiling ambient noise in the deep ocean. at 40 Hz in the 10 percentile band is the relatively common occurrence of Marine Technology Society Journal 43(5):144–150, part of the complex baleen whale call typhoons in the western Pacific Ocean https://doi.org/10.4031/MTSJ.43.5.19. Baird, R.W., D.L. Webster, D.J. McSweeney, A.D. Ligon, recently identified in the Mariana Trench (http://www.ssd.noaa.gov/PS/TROP/ G.S. Schorr, and J. Barlow. 2006. Diving behav- (see Figure S3). Basin_WestPac.html), storm noise is ior of Cuvier’s (Ziphus cavirostris) and Blainville’s (Mesoplodon densirostris) beaked whales in likely a significant part of the long-term Hawai’i. Journal of Zoology 84:1,120–1,128, SUMMARY soundscape in the region. https://doi.org/10.1139/z06-095. The successful deployment and recov- Caress, D.W., D.N. Chayes, and C. dos Santos Ferreira. 2015. MB-System seafloor mapping soft- ery of a hydrophone and pressure sensor ONLINE SUPPLEMENTAL MATERIAL ware: Processing and display of swath sonar data, Supplemental Figures S1–S3 are available online at mooring provided, to our knowledge, the http://www.mbari.org/products/research-software/ https://doi.org/10.5670/oceanog.2017.240. mb-system. first multiday, broadband record of ambi- Carruthers, J.N., and A.L. Lawford. 1952. The deep- ent sound made at Challenger Deep, as est oceanic sounding. Nature 169:601–603, REFERENCES https://doi.org/10.1038/169601a0. well as only the fifth direct depth measure- Ainslie, M.A., and J.G. McColm. 1998. A simpli- Dziak, R.P., D.R. Bohnensteihl, H. Matsumoto, fied formula for viscous and chemical absorp- ment. Moreover, the success of this deep- C.G. Fox, D.K. Smith, M. Tolstoy, T.-K. Lau, tion in sea water. Journal of the Acoustical Society J.H. Haxel, and M.J. Fowler. 2004. P- and T-wave ocean mooring system demonstrates the of America 103(3):1,671–1,672, https://doi.org/​ detection thresholds, Pn velocity estimate, and 10.1121/1.421258. value of employing the relatively low-cost detection of lower mantle and core P-waves on

196 Oceanography | Vol.30, No.2 ocean sound-channel hydrophones at the Mid- Lippsett, L., and A.E. Nevala. 2009. Nereus dives to Wenz, G.M. 1962. Acoustic ambient noise in the Atlantic Ridge. Bulletin of the Seismological the ocean’s deepest trench. Oceanus 47(3):30. ocean: Spectra and sources. Journal of the Society of America 94(2):665–677, https://doi.org/​ Nieukirk, S.L., S. Fregosi, D.K. Mellinger, and Acoustical Society of America 34:1,936–1,956, 10.1785/0120030156. H. Klinck. 2016. A complex baleen whale https://doi.org/10.1121/1.1909155. Dziak, R.P., J.H. Haxel, H. Matsumoto, C. Meinig, call recorded in the Mariana Trench Marine Wilcock, W.S.D., K.M. Stafford, R.K. Andrew, and N. Delich, J. Osse, and M. Wetzler. 2015. National Monument. Journal of the Acoustical R.I Odom 2014. Sounds in the ocean at 1–100 Hz. Deployment and recovery of a full-ocean depth Society of America 140:EL274, https://doi.org/​ Annual Review of Marine Sciences 6:117–140, mooring at Challenger Deep, Mariana Trench. 10.1121/1.4962377. https://doi.org/10.1146/annurev-marine-121211- In Oceans 2015 MTS/IEEE, Marine Technology NRC (National Research Council). 2003. Ocean Noise 172423. Society and Institute of Electrical and Electronics and Marine Mammals. The National Academies Engineers, Washington, DC, October 19–22, 2015. Press, 204 pp. Feistel, R., and E. Hagen. 1995. On the GIBBS ther- Parks, S.E., M.P. Johnson, D.P. Nowacek, and ACKNOWLEDGMENTS modynamic potential of seawater. Progress in P.L. Tyack. 2012. Changes in vocal behav- This project was supported by the NOAA Ocean Oceanography 36:249–327, https://doi.org/​ ior of North Atlantic right whales in increased Exploration and Research Program. We also wish 10.1016/0079-6611(96)00001-8. noise. Advances in Experimental Medicine to thank LCDR Worst, LT Adams, LT West, and Frisk, G., D. Bradley, J. Caldwell, G. D’Spain, and Biology 730:317–320, https://doi.org/​ LTjg Strassfield, the crew of the USCG Sequoia, J. Gordon, M. Hastings, and D. Ketten. 2003. 10.1007/978-1-4419-7311-5_70. NOAA Corps Officer LCD T. Pelzer, M. Fowler, Ocean Noise and Marine Mammals, 3rd ed. Piccard, J., and R.S. Dietz 1961. Seven Miles A. Turpin, and J. Colmanero, whose hard work and National Academies Press, Washington, DC, Down: The Story of the Bathyscaphe Trieste. dedication made this project possible. We also thank 221 pp. G.P. Putnam’s Sons, New York. B. Spychalski with HTI for assistance with descrip- Fulling, G.L., P.H. Thorson, and J. Rivers. 2011. Pijanowski, B.C., L.J. Villanueva-Rivera, S.L. Dumyahn, tion of the hydrophone sensitivity, and the Federated Distribution and abundance estimates for A. Farina, B.L. Krause, B.M. Napoletano, States of Micronesia for allowing us access to cetaceans in the waters off Guam and the S.H. Gage, and N. Pieretti. 2011. Soundscape ecol- Challenger Deep. The propagation model in Figure 4 Commonwealth of the Northern Mariana Islands. ogy: The science of sound in the landscape. is courtesy of J. Caplan-Auerbach. The authors wish Pacific Science 65(3):321–343, https://doi.org/​ BioScience 61:203–216, https://doi.org/​10.1525/ to thank W. Wilcock and three anonymous review- 10.2984/65.3.321. bio.2011.61.3.6. ers for their very helpful edits, which greatly improved Gardner, J.V., A.A. Armstrong, B.R. Calder, and Porter, M.B. 2011. The BELLHOP Manual and User’s the manuscript. This paper is PMEL contribution J. Beaudoin. 2014. So, how deep is the Mariana Guide: Preliminary Draft. Heat, Light, and Sound number 4605. Trench? Marine Geodesy 37:1–13, https://doi.org/​ Research Inc., La Jolla, California, 57 pp. 10.1080/01490419.2013.837849. Rolland, R.M., S.E. Parks, K.E. Hunt, M. Castellote, AUTHORS Goldbogen, J.A., B.L. Southall, S.L. DeRuiter, P.J. Corkeron, D.P. Nowacek, and S.K. Wasser. Robert P. Dziak ([email protected]) is J. Calambokidis, A.S. Friedlander, E.L. Hazen, 2012. Evidence that ship noise increases stress Acoustics Program Director, National Oceanic E.A. Falcone, G.S. Schorr, A. Douglas, D.J. Moretti, in right whales. Proceedings of the Royal and Atmospheric Administration/Pacific Marine and others. 2013. Blue whales respond to simu- Society B 279:2,363–2,368, https://doi.org/10.1098/ Environmental Laboratory (NOAA/PMEL), lated mid-frequency military sonar. Proceedings of rspb.2011.2429. Newport, OR, USA. Joseph H. Haxel is Assistant the Royal Society B 280, https://doi.org/10.1098/ Smoczyk, G.M., G.P. Hayes, M.W. Hamburger, Professor, Haruyoshi Matsumoto is Associate rspb.2013.0657. H.M. Benz, A. Villaseñor, and K.P. Furlong. 2013. Professor, Tai-Kwan Lau is Applied Mathematician, Hatch, L.T., C.W. Clark, S.M. van Parijs, A.S. Frankel, Seismicity of the Earth 1900–2012 Philippine Sea Sara Heimlich is Faculty Research Assistant, and D.W. Ponirakis. 2012. Quantifying loss Plate and vicinity. US Geological Survey Open- Sharon Nieukirk is Senior Faculty Research Assistant, of acoustic communication space for right File Report 2010–1083-M, scale 1:10,000,000, and David K. Mellinger is Professor, all at the whales in and around a US National Marine https://doi.org/10.3133/ofr20101083m. Cooperative Institute for Marine Resources Studies, Sanctuary. Conservation Biology 26:983–994, Soldevilla, M.S., E.E. Henderson, G.S. Campbell, Oregon State University, and NOAA/PMEL, Newport, https://doi.org/10.1111/j.1523-1739.2012.01908.x. S.M. Wiggins, J.A. Hildebrand, and M.A. Roch. OR, USA. James Osse is Research Engineer, Hatch, L.T., and A.J. Wright. 2007. A brief review of 2008. Classification of Risso’s and Pacific white- Joint Institute for Studies of the Atmosphere and anthropogenic sound in the oceans. International sided dolphins using spectral properties of Ocean, University of Washington, Seattle, WA, Journal of Comparative Psychology 20:121–133. echolocation clicks. Journal of the Acoustical USA. Christian Meinig is Director, Engineering Hildebrand, J.A. 2009. Anthropogenic and nat- Society of America 124:609–624, https://doi.org/​ Development Division, Nicholas Delich is Physical ural sources of ambient noise in the ocean. 10.1121/1.2932059. Science Technician, and Scott Stalin is Deputy Marine Ecology Progress Series 395:5–20, Southhall, B.L., D. Moretti, A. Bruce, J. Calambokidis, Division Leader, Engineering Development Division, https://doi.org/10.3354/meps08353. S.L. DeRuiter, and P.L. Tyack. 2012. Marine mammal all at NOAA/PMEL, Seattle, WA, USA. Hill, M.C., E.M. Oleson, S. Baumann-Pickering, behavioral response studies in southern California: A.M. VanCise, A.D. Ligon, A.R.Bendlin, A.C. Ü, Advances in technology and experimental meth- J.S. Trickey, and A.L. Bradford. 2016. Cetacean ods. Marine Technology Society Journal 46:48–59, ARTICLE CITATION monitoring in the Mariana Islands Range Complex, https://doi.org/10.4031/mtsj.46.4.1. Dziak, R.P., J.H. Haxel, H. Matsumoto, T.-K. Lau, 2015. Data Report DR-16-01, NOAA Pacific Islands Taira, K., S. Kitagawa, T. Yamashiro, and S. Heimlich, S. Nieukirk, D.K. Mellinger, J. Osse, Fisheries Science Center, Honolulu, HI, 55 pp. D. Yanagimoto. 2004. Deep and bottom currents C. Meinig, N. Delich, and S. Stalin. 2017. Ambient IOC, SCOR, and IAPSO. 2010. The international in the Challenger Deep, Mariana Trench, mea- sound at Challenger Deep, Mariana Trench. thermodynamic equation of seawater–2010: sured with super-deep current meters. Journal of Oceanography 30(2):186–197, https://doi.org/10.5670/ Calculation and use of thermodynamic properties. Oceanography 60:919–926, https://doi.org/10.1007/ oceanog.2017.240. Intergovernmental Oceanographic Commission, s10872-005-0001-y. Manuals and Guides No. 56, UNESCO (English), Than, K. 2012. James Cameron completes 196 pp. record-breaking Mariana Trench dive. National McDougall, T.J., D.R. Jackett, D. G. Wright, and Geographic News, March 25, 2012, http://news. R. Feistel. 2003. Accurate and computation- nationalgeographic.com/news/2012/03/120325- ally efficient algorithms for potential temperature james-cameron-mariana-trench-challenger- and density of seawater. Journal of Atmospheric deepest-returns-science-sub. and Oceanic Technology 20(5):730–741, Therberge, A. 2008. Thirty years of discovering the https://doi.org/10.1175/1520-0426(2003)20​ Mariana Trench. Hydro International 12:38–39. <730:AACEAF>2.0.CO;2. Tolstoy, M., J.B. Diebold, S.C. Webb, McKenna, M.F., D. Ross, S.M. Wiggins, and D.R. Bohnenstiehl, E. Chapp, R.C. Holmes, J.A. Hildebrand. 2012. Underwater radiated and M. Rawson. 2004. Broadband calibra- noise from modern commercial ships. Journal of tion of R/V Ewing seismic sources. Geophysical the Acoustical Society of America 131:92–103, Research Letters 31:1–4, https://doi.org/10.1029/​ https://doi.org/10.1121/1.3664100. 2004GL020234. National Geographic Society. 2012. James WHOI. 2014. Robotic deep-sea vehicle lost on dive to Cameron’s Deepsea Challenge, http://www. 6-mile depth. News release, http://www.whoi.edu/ deepseachallenge.com. news-release/Nereus-Lost.

Oceanography | June 2017 197