The conversion from seismic to ocean-acoustic waves occurs in different places on the bottom of the ocean, often hundreds to thousands of kilometers away from the epicenter. Here, we investigate this conversion process by studying 15 large-magnitude earthquakes that occurred between 2014 and 2022 along the Kermadec Arc in the southwestern Pacific Ocean. To pinpoint the location where seismic-to-acoustic conversion takes places, we analyze hydroacoustic signals recorded by a hydrophone triplet station of the International Monitoring System in the Juan Fernández archipelago. Results from direction-of-arrival and travel-time calculations indicate that the location of the conversion zone largely matches segments of the Louisville Seamount Chain, its lateral extent ranging from approximately 300 to 1800 km, and its location depending on the geometry between earthquake epicenter and the seamounts.

The conversion from seismic to ocean-acoustic waves traveling in the oceanic water column, i.e., from primary (P) or secondary (S) to tertiary (T) waves, can occur in different places on the bottom of the ocean, often hundreds of kilometers away from the epicenter (e.g., de Groot-Hedlin and Orcutt, 2001; Graeber and Piserchia, 2004; Talandier and Okal, 1979). These conversion zones have been associated with large bathymetric gradients at seaward-dipping slopes along subduction zones and continental shelves, islands, and seamounts (Bottero , 2020; Freeman , 2013; Lin, 2001; Okal, 2008). The identification and location of areas and bathymetric features where seismic to ocean-acoustic conversion occurs presents a significant challenge due to the lack of high-quality and long-term seismic and underwater acoustic measurements. Oliveira (2022) analyzed T waves from two high-magnitude submarine earthquakes (Mw of 7.4 and 8.1) in 2021 in the Kermadec Arc (southwest Pacific Ocean) and suggested that conversion from seismic to ocean-acoustic waves could have occurred near the epicenter and hundreds of kilometers away along the Louisville Seamount Chain (LSC), an approximately 4300-km-long chain of submarine volcanoes (Vanderkluysen , 2014) (Fig. 1). The authors based their analysis on low-frequency underwater acoustic signals recorded at the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO) International Monitoring System (IMS) hydrophone stations in the Pacific Ocean and the South Atlantic Ocean.

FIG. 1.

(Color online) (A, B) Location maps of the earthquake epicenters (colored circles) and IMS hydrophone station HA03 in the Juan Fernández archipelago (yellow star). Bathymetric data are taken from the GEBCO Compilation Group (2023). (C) Positions of the two HA03 hydrophone triplets, moored approximately 15 km offshore to the north (H03N) and south (H03S) of Isla Robinson Crusoe. The average mooring depth of the H03S hydrophones is approximately 830 m below sea level (bsl) and the average seafloor depth is approximately 2070 m bsl. (D) Configuration of the H03S hydrophone triple.

FIG. 1.

(Color online) (A, B) Location maps of the earthquake epicenters (colored circles) and IMS hydrophone station HA03 in the Juan Fernández archipelago (yellow star). Bathymetric data are taken from the GEBCO Compilation Group (2023). (C) Positions of the two HA03 hydrophone triplets, moored approximately 15 km offshore to the north (H03N) and south (H03S) of Isla Robinson Crusoe. The average mooring depth of the H03S hydrophones is approximately 830 m below sea level (bsl) and the average seafloor depth is approximately 2070 m bsl. (D) Configuration of the H03S hydrophone triple.

Close modal

The IMS is a global network of seismic, hydroacoustic, infrasonic, and radionuclide sensors designed to continuously monitor the world for nuclear explosions underground, in the ocean, and in the atmosphere. The hydroacoustic component of the network records underwater signals propagating in the oceans using six stations of hydrophones suspended in the water column of the oceans (Fig. 2) and five coastal seismometer stations deployed at remote islands (Lawrence and Grenard, 1998). T phases (waterborne phases that originate from earthquakes or underground explosion events as opposed to in-water events) are a key feature among the signals recorded by the hydroacoustic component of the IMS network. The IMS hydrophone station HA03, located in the Juan Fernández archipelago (Chile, southeast Pacific Ocean) (Fig. 1), regularly records hydroacoustic signals related to seismic and volcanic activity in the “Ring of Fire,” especially from the Tonga-Kermadec trench (Matoza , 2022; Metz , 2018; Oliveira , 2022) and Peru-Chile trench (Matsumoto , 2016).

FIG. 2.

(Color online) Generic deep-water sound speed profile and schematic outline of an IMS-type hydrophone triplet, side and top-down view (not to scale). Note that the mooring depth of the hydrophone sensors, riser, and trunk cable length depend on the depth of the local sound speed minimum (“SOFAR channel axis”), seafloor topography, and distance from the shore facility; these parameters vary from triplet to triplet.

FIG. 2.

(Color online) Generic deep-water sound speed profile and schematic outline of an IMS-type hydrophone triplet, side and top-down view (not to scale). Note that the mooring depth of the hydrophone sensors, riser, and trunk cable length depend on the depth of the local sound speed minimum (“SOFAR channel axis”), seafloor topography, and distance from the shore facility; these parameters vary from triplet to triplet.

Close modal

Several large-magnitude earthquakes (Mw > 8) have occurred in the Kermadec Arc region since 1900 (Romano , 2021; Todd and Lay, 2013). Here, we focus on studying the conversion from seismic to underwater sound waves along the LSC for 15 large earthquakes that occurred in the Kermadec Arc between 2014 and 2022 (Fig. 1). Signals recorded at IMS station HA03 are used to locate the conversion zone.

According to the U.S. Geological Survey (2023) earthquake database, between April 1, 2014 and January 1, 2023, 16 earthquakes with Mw greater than or equal to 6.5 occurred in the Kermadec Arc region; i.e., between 27.35 °S and 38.33 °S and 175 °W and 182 °W (Table I; Fig. 1). In this study, we limit our analysis to 15 of these events. Hydroacoustic arrivals at HA03 from the 2014-06-23 19:21:45 earthquake were partially masked by those of the preceding 2014-06-23 19:19:15 event. Eleven of the earthquakes had Mw between 6.5 and 6.9, four between 7.0 and 7.4, and the strongest earthquake had Mw of 8.1. One of the events classifies as an intermediate-depth earthquake (70–300 km deep) and the others as shallow earthquakes (0–70 Km deep), ten earthquakes were shallower than 26 km, and five were between 26 and 50 km deep. For the purpose of this study, the earthquakes are divided into four groups according to the location of their epicenters, most of which are scattered along the seaward-dipping slope facing the Kermadec Trench: Group A consists of the northernmost earthquake (#5) only. Group B consists of eight earthquakes (#2, 3, 4, 6, 7, 10, 14, and 16), located between 29.56 °S and 30.65 °S. The three earthquakes of Group C (#1, 9, and 11) have epicenters between 31.75 °S and 33.29 °S. Earthquakes in Group D (#12, 8, and 13) are located in the southern part of the Kermadec Arc adjacent to New Zealand.

TABLE I.

Main characteristics of the earthquakes considered in this study. For each earthquake, the arrival time at H03S (south component) and the duration of the signal related to seismic to underwater sound conversion along the LSC are presented. The last column gives the length of the conversion zone along the LSC.

Kermadec Arc earthquakes Hydroacoustic arrivals at H03S LSC
Group No. Date and time (UTC) Latitude (°) Longitude (°) Depth (km) Mw Arrival time (UTC) Duration (s) Conversion zone length (km)
2021-03-04 23:12:59  −28.52  −176.68  24.0  6.5  00:49:00  128  343 
2022-03-02 12:52:07  −30.08  −177.73  24.0  6.6  14:22:09  429  1147 
2022-01-29 02:46:39  −29.56  −176.72  8.0  6.5  04:19:27  282  819 
2021-06-20 17:05:50  −30.22  −177.84  25.0  6.5  18:36:26  404  1120 
2021-03-04 19:28:33  −29.72  −177.28  28.9  8.1  20:59:35  532  1369 
2021-03-04 17:41:23  −29.68  −177.84  43.0  7.4  19:12:07  510  1327 
10  2019-06-15 22:55:04  −30.64  −178.10  46.0  7.3  00:25:25  422  1107 
14  2014-06-23 20:06:20  −29.94  −177.61  26.6  6.7  21:39:16  275  800 
16  2014-06-23 19:19:15  −29.98  −177.72  20.0  6.9  20:50:14  373  949 
2022-08-14 13:44:19  −32.74  −179.01  30.0  6.6  15:14:29  433  1070 
2020-06-18 12:49:53  −33.29  −177.86  10.0  7.4  14:13:39  786  1880 
11  2018-09-10 04:19:02  −31.74  −179.37  115.0  6.9  05:45:54  637  1615 
2021-03-04 13:27:34  −37.48  179.46  10.0  7.3  14:53:03  387  858 
12  2016-09-01 16:37:57  −37.36  179.15  19.0  7.0  18:03:10  392  891 
13  2014-11-16 22:33:20  −37.65  179.66  22.0  6.7  23:58:14  374  864 
a  15  2014-06-23 19:21:45  −29.94  −177.52  10.0  6.5  —  —  — 
Kermadec Arc earthquakes Hydroacoustic arrivals at H03S LSC
Group No. Date and time (UTC) Latitude (°) Longitude (°) Depth (km) Mw Arrival time (UTC) Duration (s) Conversion zone length (km)
2021-03-04 23:12:59  −28.52  −176.68  24.0  6.5  00:49:00  128  343 
2022-03-02 12:52:07  −30.08  −177.73  24.0  6.6  14:22:09  429  1147 
2022-01-29 02:46:39  −29.56  −176.72  8.0  6.5  04:19:27  282  819 
2021-06-20 17:05:50  −30.22  −177.84  25.0  6.5  18:36:26  404  1120 
2021-03-04 19:28:33  −29.72  −177.28  28.9  8.1  20:59:35  532  1369 
2021-03-04 17:41:23  −29.68  −177.84  43.0  7.4  19:12:07  510  1327 
10  2019-06-15 22:55:04  −30.64  −178.10  46.0  7.3  00:25:25  422  1107 
14  2014-06-23 20:06:20  −29.94  −177.61  26.6  6.7  21:39:16  275  800 
16  2014-06-23 19:19:15  −29.98  −177.72  20.0  6.9  20:50:14  373  949 
2022-08-14 13:44:19  −32.74  −179.01  30.0  6.6  15:14:29  433  1070 
2020-06-18 12:49:53  −33.29  −177.86  10.0  7.4  14:13:39  786  1880 
11  2018-09-10 04:19:02  −31.74  −179.37  115.0  6.9  05:45:54  637  1615 
2021-03-04 13:27:34  −37.48  179.46  10.0  7.3  14:53:03  387  858 
12  2016-09-01 16:37:57  −37.36  179.15  19.0  7.0  18:03:10  392  891 
13  2014-11-16 22:33:20  −37.65  179.66  22.0  6.7  23:58:14  374  864 
a  15  2014-06-23 19:21:45  −29.94  −177.52  10.0  6.5  —  —  — 
a

Not used in this study.

Station HA03 of the IMS network consists of two bottom-moored hydrophone triplets, located approximately 15 km north (H03N) and south (H03S) of Isla Robinson Crusoe, the easternmost island in the Juan Fernández archipelago (Fig. 1). Initially, the southern triplet was in operation from January 2003 to March 2004, when data transmission ceased due to a fault most likely located in the deep-water segment of the trunk cable. After the loss of the remaining installation due to the 2010 Chile earthquake and tsunami (Fritz , 2011), both triplets were reestablished in April 2014. The three hydrophones of the southern triplet H03S1, H03S2, and H03S3 are moored near the axis of the Sound Fixing and Ranging (SOFAR) channel at approximately 830 m water depth. The hydrophones are organized in a tripartite configuration and at an equidistant spacing of ∼2 km. Acoustic measurements are made at a sampling rate of 250 Hz and transmitted in near real-time to the International Data Centre in Vienna for automatic processing and analyst review (Hanson , 2001).

One-hour-long segments recorded at H03S1, encompassing the hydroacoustic arrivals from the 16 earthquakes, are presented in Fig. 3. The signals are centered on the theoretical T-wave arrival time, which was calculated considering a constant propagation speed of 1.48 km/s over the geodesic distance between the reported epicenter and H03S1. The signal-to-noise ratio is sufficient to visually distinguish the T-wave arrivals of all 16 events. Although not shown here, a strong signal correlation was observed between the signals at all three hydrophones of H03S.

FIG. 3.

Signals recorded at hydrophone H03S1 for the 16 earthquakes (sorted from north to south and titled according to the group and number in Table I) are presented where the 1-hour time series are centered on the theoretical T wave arrival time. Signals are demeaned and normalized to the absolute maximum. All plots are presented with y axis limits set between −1 and 1.

FIG. 3.

Signals recorded at hydrophone H03S1 for the 16 earthquakes (sorted from north to south and titled according to the group and number in Table I) are presented where the 1-hour time series are centered on the theoretical T wave arrival time. Signals are demeaned and normalized to the absolute maximum. All plots are presented with y axis limits set between −1 and 1.

Close modal

The DASE Tool Kit–Graphical Progressive Multi-Channel Correlation (DTK-GPMCC) program (Cansi, 1995; Cansi and Klinger, 1997; Vergoz , 2021) is used here to process the recorded signals at H03S. The base algorithm of DTK-GPMCC was originally developed for infrasonic data processing and later adapted (Vergoz , 2021) and successfully applied for IMS hydroacoustic data processing (Donner , 2023; Nielsen , 2020; Oliveira , 2022; Vergoz , 2021). DTK-GPMCC can then be used, for instance, for the interpretation of the sequences of several hours of arrivals at hydrophone stations due to underwater seismicity/volcanism activity, seismic airgun surveys, iceberg activity, and marine mammals.

A bandpass filter between 0.8 and 4.5 Hz was applied to the recorded waveforms to enhance the stability of the signal, which was expected to correlate better than the original wider 100-Hz band. Detections were computed with DTK-GPMCC version 6.11.0 using ten bands (frequency cells) with equal bandwidths between 0.8 and 4.5 Hz. Detection pixels (time-frequency cells) with window lengths of 15 s and a time-series overlap of 90% were considered. The consistency threshold (criterion between the correlation in the three hydrophones) in all the frequency bands was the default value of 0.05 under which the direction of arrival is well determined. From this process, an arrival time and direction of arrival are obtained for each detection pixel.

Oliveira (2022) showed that signals from strong Kermadec Arc earthquakes at H03S contain signatures of “earlier” T waves converted from seismic waves in a generation zone east of the epicenter, which could include the LSC. These signals, recorded at H03S, were typically followed by signals of the main T waves generated near the epicenter and later by signals from reflections of the main T waves at the shelf-break off the Chilean coast and the Antarctic Peninsula. To identify the origin of the earlier T waves, which correspond to the seismic to acoustic conversion, all the earlier detection pixels were back-projected following a procedure described next.

The observables for the back projection to determine the locations of the conversion points were the arrival time and azimuth of the T waves at H03S. From the arrival azimuth, we can determine all possible conversion points along the back azimuth geodesic line from H03S. Then, with the T-wave arrival time information, we can then determine the most likely location of the conversion point along the geodesic line by computing the theoretical arrival time and comparing it with the measurements. The theoretical arrival time calculation is broken into two steps. First, the arrival time from the epicenter to potential locations of conversion points is computed with an apparent seismic wave velocity. Then, this solid wave arrival time is augmented with the travel time from the conversion point to H03S with an underwater sound speed.

Using 4.4 km/s for solid Earth propagation between the epicenter and the conversion point and a nominal value of 1.48 km/s for in-water propagation along the geodesic path from the conversion point to H03S — a reasonable first-order approximation for deep sound channel propagation in the mid-latitudinal South Pacific Ocean (e.g., Metz , 2018; Munk and Forbes, 1989) — the back-projection procedure indeed pinned down the conversion points along the LSC. As shown next, the 4.4 km/s used in the procedure represents the apparent velocity of S phases for the area under study. We found that using a representative P-phase apparent velocity (7.87 km/s) for back-projection leads to conversion points in benign seafloor areas where the conversion process is unlikely to occur.

P- and S-phase ray paths and travel times from the epicenters of earthquakes 6 (group B), 9 (group C), and 12 (group D) to different locations along the LSC are computed using the SAndia LoS Alamos 3D (SALSA3D) Earth model and PCalc. SALSA3D is a three-dimensional global seismic velocity model of the Earth's mantle that captures spatial variations of the solid earth's properties and was developed using 3D tomographic inversion of an extensive seismic travel-time dataset (Ballard , 2016b; Begnaud , 2015). This velocity model is stored in GeoTess format (Ballard , 2016a) and has been employed by PCalc (Ballard , 2009; Conley , 2021), a 3D ray tracer based on the pseudo-bending algorithm of Um and Thurber (1987) and Zhao and Lei (2004). The lateral heterogeneity of the Earth, which mostly affects propagation at local and regional seismic distances (less than about 2000 km) where waves spend most of their time in the highly heterogeneous shallow parts of the Earth, is then taken into account by the 3D velocity model and ray tracer.

The ray paths and travel times results for earthquakes 6, 9, and 12 are shown in Fig. 4 and Table II. Three locations along LSC are considered for earthquakes 6 and 9, and two locations for earthquake 12, resulting in eight seismic travel paths being computed. The P and S phase apparent velocities (calculated as geodesic distance over travel time) are shown in Fig. 5.

FIG. 4.

(Color online) Left, Bathymetric maps with the location of the epicenters (black star) and the geodesic paths (black lines) to selected seamounts in the LSC. Right, Computed P (black) and S (red) phase paths from the epicenter to the seamounts using the SALSA3D Earth model.

FIG. 4.

(Color online) Left, Bathymetric maps with the location of the epicenters (black star) and the geodesic paths (black lines) to selected seamounts in the LSC. Right, Computed P (black) and S (red) phase paths from the epicenter to the seamounts using the SALSA3D Earth model.

Close modal
TABLE II.

As a complement of Fig. 4, the table provides, for each seamount, the latitude and longitude of its location and the travel time, T, and uncertainty, δT, from the epicenter for P and S phases.

Seamount P phase S phase
Latitude Longitude T (s) δT (s) T (s) δT (s)
P06_01  −28.58  −173.3  53.982  1.3284  99.116  2.6782 
P06_02  −35.40  −170.4  116.416  1.8362  209.206  3.1561 
P06_03  −40.71  −165.4  204.377  6.4182  361.561  3.9394 
P09_01  −32.7  −171.6  77.007  1.7202  139.673  2.9232 
P09_02  −38.44  −168.0  135.477  1.7995  241.375  3.3905 
P09_03  −45.35  −157.5  269.902  1.8537  479.116  3.8359 
P12_01  −39.15  −167.4  149.687  1.7181  265.199  3.4843 
P12_02  −45.35  −157.5  261.763  1.6557  461.571  3.8156 
Seamount P phase S phase
Latitude Longitude T (s) δT (s) T (s) δT (s)
P06_01  −28.58  −173.3  53.982  1.3284  99.116  2.6782 
P06_02  −35.40  −170.4  116.416  1.8362  209.206  3.1561 
P06_03  −40.71  −165.4  204.377  6.4182  361.561  3.9394 
P09_01  −32.7  −171.6  77.007  1.7202  139.673  2.9232 
P09_02  −38.44  −168.0  135.477  1.7995  241.375  3.3905 
P09_03  −45.35  −157.5  269.902  1.8537  479.116  3.8359 
P12_01  −39.15  −167.4  149.687  1.7181  265.199  3.4843 
P12_02  −45.35  −157.5  261.763  1.6557  461.571  3.8156 
FIG. 5.

(Color online) P and S phase apparent velocity for the eight seismic travel paths presented in Fig. 4. The apparent velocity is presented as a function of the geodesic distance between the epicenter and seamount (top), and the location of the seamount (bottom).

FIG. 5.

(Color online) P and S phase apparent velocity for the eight seismic travel paths presented in Fig. 4. The apparent velocity is presented as a function of the geodesic distance between the epicenter and seamount (top), and the location of the seamount (bottom).

Close modal

The apparent velocities of P and S phases for the eight propagation paths increase with latitude along LSC. For the eight propagation paths, the P phase apparent velocity ranges from 7.55 to 8.16 km/s with an average of 7.87 km/s. Similarly, S phase apparent velocity varies from 4.11 to 4.63 km/s with an average of 4.40 km/s.

Even though we used a 3D model for ray tracing and obtained full 3D seismic travel time estimates for the eight seismic travel paths, we simplified the back-projection procedure by using a single, averaged apparent seismic velocity for all 15 earthquakes.

Figure 6 shows the spectrograms and waveforms of the hydroacoustic arrivals associated with four earthquakes (5, 6, 9, and 12) recorded at H03S1, each representing one of the four groups (A, B, C, and D). In the same figure, the direction of arrival (back azimuth) of each detection pixel is plotted. The time window that contains the arrivals identified as being from conversion at the LSC is highlighted in the waveform plot. This window precedes the main energy arrival, which represents the T wave resulting from seismic-to-acoustic conversion most closely to the epicenter. The arrival time and duration of the LSC signals for all the analyzed earthquakes are provided in Table I. The same table provides the lengths (northwest-southeast extension along the LSC) of the conversion zones obtained after the pixels are back-projected onto a map (Fig. 7). The pixels' back projections for the 15 earthquakes agree well with the northwest-southeast trending extension of the LSC. Different velocity models were tested considering other realistic values for the propagation in the solid Earth (e.g., 7.87 km/s) and in the ocean (e.g., 1.5 km/s). In general, all tested velocity models allowed us to identify the LSC as the most likely location where conversion of seismic energy into the oceanic water column takes place, representing well the orientation of the seamount chain but with pixel projections more distant from the seamount chain.

FIG. 6.

(Color online) Spectrograms and waveform data recorded at H03S1 for earthquakes 5 (Group A), 6 (Group B), 9 (Group C), and 12 (Group D). The back azimuth is presented for the LSC arrivals. Horizontal red line indicates the azimuth to the epicenter. Rectangles in the waveforms identify the time interval with detected pixels with origin in the LSC.

FIG. 6.

(Color online) Spectrograms and waveform data recorded at H03S1 for earthquakes 5 (Group A), 6 (Group B), 9 (Group C), and 12 (Group D). The back azimuth is presented for the LSC arrivals. Horizontal red line indicates the azimuth to the epicenter. Rectangles in the waveforms identify the time interval with detected pixels with origin in the LSC.

Close modal
FIG. 7.

(Color online) Locations of the conversion from seismic to underwater sound waves along the LSC for the four earthquake groups. Each blue circle represents a detection pixel at H03S back-projected onto the map. Red circles represent earthquake epicenters.

FIG. 7.

(Color online) Locations of the conversion from seismic to underwater sound waves along the LSC for the four earthquake groups. Each blue circle represents a detection pixel at H03S back-projected onto the map. Red circles represent earthquake epicenters.

Close modal

The duration of the hydroacoustic signals recorded at H03S corresponding to conversion at the LSC ranged between 128 and 786 s with an average duration of 424 s and tended to increase with the earthquake magnitude (Fig. 8). The average duration was higher for Group C (619 s) than for B (403 s) and D (385 s). The length of the conversion zone ranged between 343 and 1880 km and on average is higher for Group C (1522 km) than D, B, and A, 871, 1080, and 1343 km, respectively. There is a strong linear correlation (R2 = 0.96) between the duration of the signal at H03S and the length of the conversion zone (Fig. 9).

FIG. 8.

(Color online) Duration of the signals recorded at H03S corresponding to conversion at the LSC as a function of earthquake magnitude (Mw). Each point corresponds to one of the 15 earthquakes analyzed.

FIG. 8.

(Color online) Duration of the signals recorded at H03S corresponding to conversion at the LSC as a function of earthquake magnitude (Mw). Each point corresponds to one of the 15 earthquakes analyzed.

Close modal
FIG. 9.

(Color online) Duration of the signals recorded at H03S corresponding to conversion at the LSC as a function of the length of the coupling zone. Each point corresponds to one of the 15 earthquakes analyzed.

FIG. 9.

(Color online) Duration of the signals recorded at H03S corresponding to conversion at the LSC as a function of the length of the coupling zone. Each point corresponds to one of the 15 earthquakes analyzed.

Close modal

Due to the northwest-southeast orientation of the LSC relative to the north-south extension of the Kermadec Arc and the location of the hydrophone triplet, acoustic waves triggered at the southeastern part of the chain arrive earlier at H03S than those triggered at the northwestern part. Note that the southeastern part of the chain is approximately 2000 km closer to H03S than the northwestern part. Consequently, for the 15 earthquakes analyzed, the hydroacoustic signal at H03S from the LSC exhibits a back azimuth variation, moving from south to north between 233°and 243°. Indeed, there is an apparent move-out of the coupling points due to the epicenter-conversion-point geometry because seismic energy reaches seamounts at the northwestern end of the conversion zone first. In-water travel time is lower than seismic travel time, which facilitates the south-to-north move-out of the back azimuth.

The time interval between the end of the LSC arrivals (i.e., the end of the highlighted window in Fig. 6) and the main arrivals related to the T waves triggered near the epicenter is larger for group D. This can be due to the insufficiency of these earthquakes to trigger acoustic waves in the northwestern part of the LSC, which arrive to H03S later than the waves triggered in the south part of the seamount chain.

The signals recorded at the IMS hydrophone triplet H03S (Juan Fernández, Chile) for the 15 Kermadec Arc earthquakes (Mw > 6.5 from 2014 to 2022), confirm conversion from seismic to underwater sound waves along the LSC. Results suggest that the seismic to acoustic conversion moves over time along the seamount chain, starting at seamounts closer to the epicenter followed by seamounts further away. As the Kermadec Arc follows a north-south orientation and the LSC follows a northwest-southeast orientation, northern earthquakes are geographically closer to the seamount chain. Observations at H03S suggest that the conversion at the seamounts occurs between approximately 300 and 2000 km from the epicenter. This distance depends on how far the epicenter is from the seamount chain and on the ability of the earthquake to trigger the conversion on different parts of the seamount chain. The extent of the conversion zone along the seamount chain can approximately range from 300 to 1800 km, and its location depends on the latitude and longitude of the earthquake epicenter.

T waves triggered at the LSC arrive at H03S earlier and are less energetic than the T waves triggered near the epicenter (compare Fig. 6). We interpret this as an effect of increased attenuation due to the prolonged travel path in the ground. However, for some selected earthquakes, such as earthquake 9 with Mw = 7.4, T waves from the LSC can exhibit energy levels comparable to those of T waves triggered at the epicenter itself. This poses a significant challenge for automated event detection, association, and location processors that typically rely on the calculation of energy ratios to determine whether an arrival is present in the data and, if so, to subsequently define an onset time. Further studies are essential to better understand the T-wave conversion mechanism and propagation of these particular earthquakes. This should be supported by underwater acoustic and seismic field measurements along the Kermadec Arc and the LSC. Such measurements will also further elucidate the influence of LCS-induced three-dimensional underwater sound propagation effects on T waves triggered at the epicenter and propagating within the SOFAR channel, leading to defocusing shadow zones behind the highest seamounts.

Y.-T.L. was supported by Office of Naval Research Grant No. N00014-21-1-2416.

The authors report no conflict of interest. The views expressed by the authors of this study do not necessarily reflect those of the Preparatory Commission for the CTBTO.

The Preparatory Commission for the CTBTO provides access to IMS data and IDC data products through the virtual Data Exploitation Centre. Instructions on how to request access to the virtual Data Exploitation Centre platform are provided under www.ctbto.org/resources/for-researchers-experts/vdec. The PCalc ray tracer and SALSA3D velocity model and software are available freely at https://github.com/sandialabs/Salsa3DSoftware.

1.
Ballard
,
S.
,
Hipp
,
J.
,
Kraus
,
B.
,
Encarnacao
,
A.
, and
Young
,
C.
(
2016a
). “
GeoTess: A generalized Earth model software utility
,”
Seismol. Res. Lett.
87
(
3
),
719
725
.
2.
Ballard
,
S.
,
Hipp
,
J. R.
,
Begnaud
,
M. L.
,
Young
,
C. J.
,
Encarnacao
,
A. V.
,
Chael
,
E. P.
, and
Phillips
,
W. S.
(
2016b
). “
SALSA3D: A tomographic model of compressional wave slowness in the Earth's mantle for improved travel–time prediction and travel–time prediction uncertainty
,”
Bull. Seismol. Soc. Am.
106
,
2900
2916
.
3.
Ballard
,
S.
,
Hipp
,
J. R.
, and
Young
,
C. J.
(
2009
). “
Efficient and accurate calculation of ray theory seismic travel time through variable resolution 3D Earth models
,”
Seismol. Res. Lett.
80
(
6
),
989
999
.
4.
Begnaud
,
M. L.
,
Ballard
,
S.
,
Young
,
C. J.
,
Hipp
,
J. R.
,
Encarnacao
,
A. V.
,
Maceira
,
M.
,
Phillips
,
W. S.
,
Chael
,
E. P.
, and
Rowe
,
C. A.
(
2015
). “
Extending SALSA3D: Adding secondary phases to a global 3D model for improved seismic event location
,” in
Seismology Society of America Meeting
, Pasadena, California.
5.
Bottero
,
A.
,
Cristini
,
P.
, and
Komatitsch
,
D.
(
2020
). “
On the influence of slopes, source, seabed and water column properties on T waves: Generation at shore
,”
Pure Appl. Geophys.
177
,
5695
5711
.
6.
Cansi
,
Y.
(
1995
). “
An automatic seismic event processing for detection and location: The P.M.C.C. method
,”
Geophys. Res. Lett.
22
,
1021
1024
, https://doi.org/10.1029/95GL00468.
7.
Cansi
,
Y.
, and
Klinger
,
Y.
(
1997
). “
An automated data processing method for mini-arrays
,”
Newsl. Eur. Mediterranean Seismol. Centre
11
,
2
4
.
8.
Conley
,
A.
,
Downey
,
N. J.
,
Ballard
,
S.
,
Hipp
,
J. R.
,
Hammond
,
P.
,
Davenport
,
K.
, and
Begnaud
,
M. E.
(
2021
). “
PCalc user's manual
,” Technical Report (
Sandia National Laboratories
,
Albuquerque, NM
).
9.
de Groot-Hedlin
,
C. D.
, and
Orcutt
,
J. A.
(
2001
). “
Excitation of T-phases by seafloor scattering
,”
J. Acoust. Soc. Am.
109
,
1944
1954
.
10.
Donner
,
S.
,
Steinberg
,
A.
,
Lehr
,
J.
,
Pilger
,
C.
,
Hupe
,
P.
,
Gaebler
,
P.
,
Ross
,
J. O.
,
Eibl
,
E. P. S.
,
Heimann
,
S.
,
Rebscher
,
D.
,
Plenefisch
,
T.
, and
Ceranna
,
L.
(
2023
). “
The January 2022 Hunga Volcano explosive eruption from the multitechnological perspective of CTBT monitoring
,”
Geophys. J. Int.
235
,
48
73
.
11.
Freeman
,
S. E.
,
D'Spain
,
G. L.
,
Lynch
,
S. D.
,
Stephen
,
R. A.
,
Heaney
,
K. D.
,
Murray
,
J. J.
,
Baggeroer
,
A. B.
,
Worcester
,
P. F.
,
Dzieciuch
,
M. A.
, and
Mercer
,
J. A.
(
2013
). “
Estimating the horizontal and vertical direction-of-arrival of water-borne seismic signals in the Northern Philippine Sea
,”
J. Acoust. Soc. Am.
134
(
4
),
3282
3298
.
12.
Fritz
,
H. M.
,
Petroff
,
C. M.
,
Catalán
,
P. A.
,
Cienfuegos
,
R.
,
Winckler
,
P.
,
Kalligeris
,
N.
,
Weiss
,
R.
,
Barrientos
,
S. E.
,
Meneses
,
G.
,
Valderas-Bermejo
,
C.
,
Ebeling
,
C.
,
Papadopoulos
,
A.
,
Contreras
,
M.
,
Almar
,
R.
,
Dominguez
,
J. C.
, and
Synolakis
,
C. E.
(
2011
). “
Field survey of the 27 February 2010 Chile tsunami
,”
Pure Appl. Geophys.
168
,
1989
2010
.
13.
GEBCO Compilation Group.
(
2023
). “
GEBCO_2023 grid
,” https://doi.org/10.5285/f98b053b-0cbc-6c23-e053-6c86abc0af7b.
14.
Graeber
,
F. M.
, and
Piserchia
,
P.-F.
(
2004
). “
Zones of T-wave excitation in the NE Indian Ocean mapped using variations in back azimuth over time obtained from multi-channel correlation of IMS hydrophone triplet data
,”
Geophys. J. Int.
158
,
239
256
.
15.
Hanson
,
J.
,
Le Bras
,
R.
,
Dysart
,
P.
,
Brumbaugh
,
D.
,
Gault
,
A.
, and
Guern
,
J.
(
2001
). “
Operational processing of hydroacoustics at the Prototype International Data Center
,”
Pure Appl. Geophys.
158
,
425
456
.
16.
Lawrence
,
M. W.
, and
Grenard
,
P.
(
1998
). “
Hydroacoustic monitoring system for the Comprehensive Nuclear-Test-Ban Treaty
,” in
IEEE Oceanic Engineering Society. OCEANS'98. Conference Proceedings (Catalog No. 98CH36259), Nice, France
, Vol. 2, pp.
694
697
.
17.
Lin
,
C.-H.
(
2001
). “
T–waves excited by S–waves and oscillated within the ocean above the Southeastern Taiwan Forearc
,”
Geophys. Res. Lett.
28
,
3297
3300
, https://doi.org/10.1029/2001GL013152.
18.
Matoza
,
R. S.
,
Fee
,
D.
,
Assink
,
J. D.
,
Iezzi
,
A. M.
,
Green
,
D. N.
,
Kim
,
K.
,
Toney
,
L.
,
Lecocq
,
T.
,
Krishnamoorthy
,
S.
,
Jean-Lalande
,
M.
,
Nishida
,
K.
,
Gee
,
K. L.
,
Haney
,
M. M.
,
Ortiz
,
H. D.
,
Brissaud
,
Q.
,
Martire
,
L.
,
Rolland
,
L.
,
Vergados
,
P.
,
Nippress
,
A.
,
Park
,
J.
,
Shani-Kadmiel
,
S.
,
Witsil
,
A.
,
Arrowsmith
,
S.
,
Caudron
,
C.
,
Watada
,
S.
,
Perttu
,
A. B.
,
Taisne
,
B.
,
Mialle
,
P.
,
Le Pichon
,
A.
,
Vergoz
,
J.
,
Hupe
,
P.
,
Blom
,
P. S.
,
Waxler
,
R.
,
De Angelis
,
S.
,
Snively
,
J. B.
,
Ringler
,
A. T.
,
Anthony
,
R. E.
,
Jolly
,
A. D.
,
Kilgour
,
G.
,
Averbuch
,
G.
,
Ripepe
,
M.
,
Ichihara
,
M.
,
Arciniega-Ceballos
,
A.
,
Astafyeva
,
E.
,
Ceranna
,
L.
,
Cevuard
,
S.
,
Che
,
I.-Y.
,
De Negri
,
R.
,
Ebeling
,
C. W.
,
Evers
,
L. G.
,
Franco-Marin
,
L. E.
,
Gabrielson
,
T. B.
,
Hafner
,
K.
,
Harrison
,
R. G.
,
Komjathy
,
A.
,
Lacanna
,
G.
,
Lyons
,
J.
,
Macpherson
,
K. A.
,
Marchetti
,
E.
,
McKee
,
K. F.
,
Mellors
,
R. J.
,
Mendo-Pérez
,
G.
,
Mikesell
,
T. D.
,
Munaibari
,
E.
,
Oyola-Merced
,
M.
,
Park
,
I.
,
Pilger
,
C.
,
Ramos
,
C.
,
Ruiz
,
M. C.
,
Sabatini
,
R.
,
Schwaiger
,
H. F.
,
Tailpied
,
D.
,
Talmadge
,
C.
,
Vidot
,
J.
,
Webster
,
J.
, and
Wilson
,
D. C.
(
2022
). “
Atmospheric waves and global seismoacoustic observations of the January 2022 Hunga eruption, Tonga
,”
Science
377
,
95
100
.
19.
Matsumoto
,
H.
,
Haralabus
,
G.
,
Zampolli
,
M.
, and
Özel
,
N. M.
(
2016
). “
T-phase and tsunami pressure waveforms recorded by near-source IMS water-column hydrophone triplets during the 2015 Chile earthquake
,”
Geophys. Res. Lett.
43
,
12511
12519
, https://doi.org/10.1002/2016GL071425.
20.
Metz
,
D.
,
Watts
,
A. B.
,
Grevemeyer
,
I.
, and
Rodgers
,
M.
(
2018
). “
Tracking submarine volcanic activity at Monowai: Constraints from long-range hydroacoustic measurements
,”
JGR Solid Earth
123
,
7877
7895
.
21.
Munk
,
W. H.
, and
Forbes
,
A. M.
(
1989
). “
Global ocean warming: An acoustic measure?
,”
J. Phys. Oceanogr.
19
(
11
),
1765
1778
.
22.
Nielsen
,
P. L.
,
Zampolli
,
M.
,
Le Bras
,
R.
,
Mialle
,
P.
,
Bittner
,
P.
,
Poplavskiy
,
A.
,
Rozhkov
,
M.
,
Haralabus
,
G.
,
Tomuta
,
E.
,
Bell
,
R.
,
Grenard
,
P.
,
Taylor
,
T.
, and
Meral Özel
,
N.
(
2020
). “
CTBTO's data and analysis pertaining to the search for the missing Argentine submarine ARA San Juan
,”
Pure Appl. Geophys.
178
,
2557
2577
.
23.
Okal
,
E. A.
(
2008
). “
The generation of T waves by earthquakes
,”
Adv. Geophys.
49
,
1
65
.
24.
Oliveira
,
T. C. A.
,
Nielsen
,
P.
,
Lin
,
Y.-T.
,
Kushida
,
N.
, and
Jesus
,
S. M.
(
2022
). “
Megameter propagation and correlation of T-waves from Kermadec Trench and Islands
,”
Front. Mar. Sci.
9
,
1009013
.
25.
Romano
,
F.
,
Gusman
,
A. R.
,
Power
,
W.
,
Piatanesi
,
A.
,
Volpe
,
M.
,
Scala
,
A.
, and
Lorito
,
S.
(
2021
). “
Tsunami source of the 2021 MW 8.1 Raoul Island earthquake from DART and tide-gauge data inversion
,”
Geophys. Res. Lett.
48
,
e2021GL094449
, https://doi.org/10.1029/2021GL094449.
26.
Talandier
,
J.
, and
Okal
,
E. A.
(
1979
). “
Human perception of T waves: The June 22, 1977 Tonga earthquake felt on Tahiti
,”
Bull. Seismol. Soc. Am.
69
,
1475
1486
.
27.
Todd
,
E. K.
, and
Lay
,
T.
(
2013
). “
The 2011 Northern Kermadec earthquake doublet and subduction zone faulting interactions
,”
JGR Solid Earth
118
,
249
261
.
28.
Um
,
J.
, and
Thurber
,
C.
(
1987
). “
A fast algorithm for two-point seismic ray tracing
,”
Bull. Seismol. Soc. Am.
77
,
972
986
.
29.
U.S. Geological Survey
(
2023
). “
Earthquake hazards program
,” http://earthquake.usgs.gov.
30.
Vanderkluysen
,
L.
,
Mahoney
,
J. J.
,
Koppers
,
A. A. P.
,
Beier
,
C.
,
Regelous
,
M.
,
Gee
,
J. S.
, and
Lonsdale
,
P. F.
(
2014
). “
Louisville Seamount Chain: Petrogenetic processes and geochemical evolution of the mantle source
,”
Geochem. Geophys. Geosyst.
15
,
2380
2400
, https://doi.org/10.1002/2014GC005288.
31.
Vergoz
,
J.
,
Cansi
,
Y.
,
Cano
,
Y.
, and
Gaillard
,
P.
(
2021
). “
Analysis of hydroacoustic signals associated to the loss of the Argentinian ARA San Juan submarine
,”
Pure Appl. Geophys.
178
,
2527
2556
.
32.
Zhao
,
D.
, and
Lei
,
J.
(
2004
). “
Seismic ray path variations in a 3D global velocity model
,”
Phys. Earth Planet. Inter.
141
(
3
),
153
166
.