Tertiary Waves Measured During 2017 Pohang Earthquake Using an Underwater Glider

applied sciences

Article Tertiary Waves Measured during 2017 Pohang Earthquake Using an Underwater Glider

Jung-Han Lee 1,2 , Sung-Hyub Ko 2, Seom-Kyu Jung 2 and Jong-Wu Hyeon 2,* 1 Department of Electronic Engineering, Sogang University, Seoul 04107, Korea; [email protected] 2 Marine Security and Safety Research Center, Korea Institute of Ocean Science & Technology, Busan 49111, Korea; [email protected] (S.-H.K.); [email protected] (S.-K.J.) * Correspondence: [email protected]

Received: 18 August 2019; Accepted: 10 September 2019; Published: 14 September 2019

Abstract: An underwater glider equipped with a hydrophone observed the acoustic sounds of an earthquake that occurred on 15 November 2017 05:29:32 (UTC) in the Pohang area. The underwater glider observed the earthquake sounds after 19 s (05:29:51) at approximately 140 km from the Pohang epicenter. In order to distinguish the earthquake sound from the glider’s operation noise, the noise sources and Sound Pressure Level (SPL) of the underwater glider were analyzed and measured at laboratory tank and sea. The earthquake acoustic signal was distinguished from glider’s self-noises of fin, pumped Conductivity-Temperature-Depth profiler (CTD) and altimeter which exist over 100 Hz. The dominant frequencies of the earthquake acoustic signals due to the earthquake were 10 Hz. Frequencies at which the spectra had dropped 60 dB were 50 Hz. By analysis of time correlation with seismic waves detected by five seismic land stations and the earthquake acoustic signal, it is clearly shown that the seismic waves converted to Tertiary waves and then detected by the underwater glider. The results allow constraining the acoustic sound level of the earthquake and suggest that the glider provides an effective platform for enhancing the earth seismic observation systems and monitoring natural and anthropogenic ocean sounds.

Keywords: underwater glider; Pohang earthquake; tertiary waves (T-Waves); acoustic signal

1. Introduction Seismic energy from submarine earthquakes is converted into acoustic energy at the seafloor-water boundary. A Tertiary wave (or T-wave) is the acoustic signal from these earthquakes. A T-wave typically has frequencies ranging from 4 to 50 Hz. T-waves propagate efficiently in the ocean compared to seismic waves through the earth and can be detected at great distances. Underwater sound observation techniques can directly observe oceanic earthquakes, landslides, volcanic eruptions, and corresponding underwater explosions in the ocean. Although not many, there have been some reports of direct observations of hydro-acoustic waves due to underwater seismic activity from an offshore observatory [1–4]. These are the first direct measurements ever performed in a tsunami generation area. It is becoming increasingly important to acquire low-cost, high-efficiency, small-scale, unmanned, direct observation technology to replace existing high-cost, low-efficiency, large-scale, manned exploration platforms such as marine research vessels and stations in [1–6]. Unmanned observation platforms such as Argo floats, Underwater Gliders, WaveGliders, Autonomous Underwater Vehicles and Autonomous Surface Vehicles have been widely used in marine surveys and military defense operations. Among these platforms, underwater gliders have been actively used for measurement of environmental underwater ambient noise. The addition of a single hydrophone sensor to an underwater glider provides a capability for measurement of underwater noise level [7–14] and the addition of a hydrophone array enables surveillance of the underwater acoustic environment [15,16].

Appl. Sci. 2019, 9, 3860; doi:10.3390/app9183860 www.mdpi.com/journal/applsci Appl. Sci. 2019, 9, 3860 2 of 13

Research programs using these technologies have been conducted to investigate geophysical causes of seaﬂoor earthquakes, landslides, volcanic eruptions and corresponding underwater explosions using hydrophone recorders [17,18], an underwater glider [19]. Reference [19] observed underwater sounds at depths of 200 to 600 m using an underwater glider at a distance of 10 km from the Lau Basin submarine volcano. The sound frequency due to the volcano was 10–300 Hz, and the sound level was 25 dB higher than the ambient noise level. The prototype and concept as ocean surface − gateway (OSG) using WaverGliders for retransmitting data of ocean bottom package to shore were demonstrated in [20]. In this paper we report the analysis of earthquake sounds recorded by an underwater glider equipped with a single hydrophone during the magnitude 5.5 earthquake of 15 November 2017, at Pohang in the Republic of Korea. The underwater glider measured the earthquake sounds at about 140 km from the epicenter, in a water depth of about 200 m during an underwater noise measurement survey from Uljin to Geoseong along the east coast of South KOREA. In order to distinguish the earthquake sound from the glider’s operation noise, the noise sources and Sound Pressure Level (SPL) of the underwater glider were analyzed and measured. Next, the earthquake sounds measured by the gilder was analyzed in time and frequency domain. Finally, the time correlation of the seismic waves and earthquake sounds observed by the seismic land station and the glider were described.

2. Materials and Methods

2.1. Noise Sources of the Underwater Glider Underwater gliders are propelled by adjusting the buoyancy, thereby using the resulting vertical motion to drive the glider. Underwater gliders have no propeller, so there is little operating noise. Its possible noise sources, such as the air bladder, fin steering, battery movement and volume piston are described in [21]. Chandrayadula et al. [11] showed that high-intensity sounds such as the sloshing of the glider at the sea surface, the bladder pump, the ballast pump, and the battery packs moving around are all broadband in nature. The bulk of the broadband noises were concentrated at the lower frequencies 0–3 kHz. However, these only describe possible noise sources and no specific frequency and noise-level analysis of each of the noise sources, including the driving and sensor component parts, has yet been studied. Recently, An analysis has been reported on the self-noise of only the driving part of the Petrel II underwater glider developed in China [22]. In this paper, the noise sources of the driving and sensor parts of the underwater glider observed the earthquake sounds are described in detail. A Slocum underwater glider (200 m class) manufactured by Teledyne-Webb Research is used in this study (Figure1). The glider has an overall length of 1.5 m and a mass of 60 kg. The buoyancy engine is an electrically powered piston drive, located in the nose section of the glider. The drive allows the glider to take in and expel water, thereby changing its overall buoyancy. The mechanism allows a nearly neutrally buoyant trimmed glider to change its displacement in water by 250 cc which corresponds to approximately 0.5% of the total volume ± ± displaced. This change in buoyancy generates a vertical force which is translated through two swept wings into a combined forward and up/downward motion. Due to the location of the piston drive, also called the buoyancy engine, the change in direction of the buoyant force also creates the main pitching moment for the glider. Besides the buoyancy engine the glider possesses two more control actuators, a 9.1 kg battery pack, referred to as the sliding mass, that can be linearly translated along the main axis of the glider and a rudder attached to the vehicle tail fin structure. The sliding mass is used for fine tuning the pitch angle. A pumped Conductivity-Temperature-Depth profiler (CTD) sensor is contained in the payload bay of the glider. When the underwater glider moves vertically and horizontally, the driving and sensor parts generate noise. The noise sources associated with the driving parts include a piston, pitch controller, fin, and air bladder. The sensor noise sources are a pumped CTD and an altimeter. Appl.Appl. Sci. Sci. 20192019, 9,, x9 ,FOR 3860 PEER REVIEW 3 of3 13 of 13

Figure 1. Components of the Slocum Underwater Glider. 2.2. Driving-Part NoiseFigure Sources 1. Components of the Slocum Underwater Glider.

2.2. DrivingTo change-Part N itsoise direction Sources of vertical motion the underwater glider piston pulls sea water to change the effective hull volume. This produces a combined noise from the dc motor, reducer, ball screw (linear To change its direction of vertical motion the underwater glider piston pulls sea water to change motion) and piston. When the glider’s vertical motion changes, the glider moves the battery pack back the effective hull volume. This produces a combined noise from the dc motor, reducer, ball screw and forth to adjust the angle of the bow to match the hull glide angle. This produces a combined noise (linear motion) and piston. When the glider’s vertical motion changes, the glider moves the battery of dc motor, reducer, lead screw and battery pack sliding. The underwater glider moves the tail fin to pack back and forth to adjust the angle of the bow to match the hull glide angle. This produces a control the direction of movement every 3 to 4 s. This generates a noise from a small stepper-motor combined noise of dc motor, reducer, lead screw and battery pack sliding. The underwater glider drive. An air bladder operating under the tail antenna prevents the tail antenna from sinking into the moves the tail fin to control the direction of movement every 3 to 4 seconds. This generates a noise water when the underwater glider is surfacing for Iridium communication. An air pump drive starts from a small stepper-motor drive. An air bladder operating under the tail antenna prevents the tail as the glider ascends through the uppermost 10 m water depth; this pump generates noise. antenna from sinking into the water when the underwater glider is surfacing for Iridium communication.2.3. Sensor-Part An Noise air Sourcespump drive starts as the glider ascends through the uppermost 10 m water depth; this pump generates noise. For stable CTD measurement, water is pumped through the CTD sensor and this generates noise. 2.3The. Sensor pumped-Part N CTDoise operatesSources continuously. The rate of pumping can be changed according to whether the glider is descending or ascending. The altimeter is normally used only while diving to prevent the gliderFor from stable hitting CTD themeasurement, sea floor. This water is yet is another pumped source through of noise. the CTD sensor and this generates noise. The pumped CTD operates continuously. The rate of pumping can be changed according to whether3. Results the glider is descending or ascending. The altimeter is normally used only while diving to prevent the glider from hitting the sea floor. This is yet another source of noise. 3.1. Noise Measurement of the Underwater Glider 3. Results 3.1.1. Glider’s Operation Noise Measurement at Laboratory Tank 3.1. NoiseSound Measurement pressure of level the U (SPL)nderwater is usually Glider expressed as twenty times the logarithm to base 10 of the ratio of the root-mean-square sound pressure to the reference sound pressure: SPL = 20 log (pˆ/p0) 3.1.1.where Glider’spˆ and Opperation0 are the N root-mean-squareoise Measurement sound at Laboratory pressure andTank reference sound pressure, respectively. 2 SPLSound has units pressure of dB level re 1 µ(SPL)Pa . SPLis usually values expressed of the driving- as twenty and sensor-parttimes the logarithm noise sources to base are 10 discussed. of the Noises from both the driving and sensor parts of the underwater glider were measured in a ratio of the root-mean-square sound pressure to the reference sound pressure: SPL = 20 log (푝̂/p0) laboratory tank (Figure2). The size of the tank is 20 m (W) 50 m (L) 10 m (D) and water depth in where 푝̂ and p0 are the root-mean-square sound pressure and× reference sound× pressure, respectively. SPLthe has tank units was of about dB re 81 m.µPa Since2. SPL the values water of tank the driving contained- and fresh sensor water,-part the noise buoyancy sources of are the discussed. underwater gliderNoises was from adjusted both in the advance driving to and be neutrally sensor parts buoyant of the in freshunderwater water. Inglider order were to record measured the noise in a of laboratorythe underwater tank (Fig glider,ure 2). a hydrophone The size of the was tank mounted is 20 m on (W) the underwater× 50 m (L) ×glider 10 m (D) body. and The water reference depth value in was 168 dB re 1 µPa2. The gain and resolution of the recorder were 30 dB and 16 bits, respectively. the tank− was about 8 m. Since the water tank contained fresh water, the buoyancy of the underwater glider was adjusted in advance to be neutrally buoyant in fresh water. In order to record the noise of the underwater glider, a hydrophone was mounted on the underwater glider body. The reference value was −168 dB re 1 µPa2. The gain and resolution of the recorder were 30 dB and 16 bits, respectively. Appl. Sci. 2019, 9, x FOR PEER REVIEW 4 of 13

Appl. Sci. 2019, 9, 3860 4 of 13 Appl. Sci. 2019, 9, x FOR PEER REVIEW 4 of 13

Figure 2. A underwater glider being prepared for immersion into a tank and diving. FigureFigure 2. A underwater2. A underwater glider glider being being preparedprepared for for immersion immersion into into a tank a tankand diving. and diving. Except for the tail, which was used for radio frequency communication with the operator, the ExceptExcept for the for tail,the tail, which which was was usedused for for radio radio frequency frequency communication communication with the withoperator, the the operator, underwater glider was fully submerged. Noise was generated by inputting an operation command the underwaterunderwater glider glider was was fully fully submerged. submerged. Noise was Noise generated was by generated inputting byan operation inputting command an operation arbitrarily to measure the SPL of each driving and sensor component (Figure 3). Fin noise commandarbitrarily arbitrarily to measure to measure the SPL the of SPL each of driving each driving and sensor and component sensor component (Figure 3). Fin(Figure noise3 ). Fin measurement,measurement, however, however, was was not not possible possible because because the the tail tail was was not not submerged, submerged, to allow to allowfor radio for radio noise measurement, however, was not possible because the tail was not submerged, to allow for radio frequencyfrequency communication. communication. The The noise noise analysis analysis ofof fin operation operation was was obtained obtained from fromthe reco therded reco datarded data frequency communication. The noise analysis of fin operation was obtained from the recorded data at sea, at as sea,described as described in the in thenext next section. section. The The pumpedpumped CTD CTD did did not not work work due todue the to low the electrical low electrical at sea, asconductivity described value inthe of the next fresh section. water used The in pumped the tank; CTDso, the did pumped notwork CTD noise due analysis to thelow is also electrical conductivity value of the fresh water used in the tank; so, the pumped CTD noise analysis is also conductivitydescribed value in the of next the section. fresh water The Piston used noise in the leve tank;l spectrum so, the rises pumped above the CTD background noise analysis level by is also described in the next section. The Piston noise level spectrum rises above the background level by describedmore in thethan next 5 dB section.in all frequency The Piston bands noise up to level 10 kHz spectrum (Figure 4a). rises The above Pitch thenoise background level likewise level rises by more more than 5 dB in all frequency bands up to 10 kHz (Figure 4a). The Pitch noise level likewise rises than 5 dBby inmore all than frequency 5 dB in frequency bands up bands to 10 above kHz 1 (Figure kHz (Fig4a).ure The4b). The Pitch Air noise bladder level noise likewise level rises rises more by more by morethan than 10 5dB dB at in155, frequency 310, 463, 522, bands 552, above618 Hz 1and kHz to a(Fig lesserure extent 4b). Thein a broadAir bladder frequency noise band level (Fig urerises more than 5 dB in frequency bands above 1 kHz (Figure4b). The Air bladder noise level rises more than than 104c). dB The at 155, Altimeter 310, noise463, 522,level 552,rises 618more Hz than and 5 dB to at a frequencies lesser extent above in 100 a broad Hz (Fig frequencyure 4d). To recordband (Figure 10 dB at 155, 310, 463, 522, 552, 618 Hz and to a lesser extent in a broad frequency band (Figure4c). The 4c). Thenoise Altimeter from the noise flow betweenlevel rises the moreunderwater than glider5 dB at and frequencies the hydrophone, above the 100 glider Hz performed (Figure 4d). a “yo” To record Altimeter(descent noise and level ascent) rises moremission. than By 5measuring dB at frequencies the noise from above repeated 100 Hz yo (Figure missions,4d). it Towas record confirmed noise from noise from the flow between the underwater glider and the hydrophone, the glider performed a “yo” the flowthat between there was the no underwater flow noise caused glider by and turbulence the hydrophone, between the the underwater glider performed glider body a and “yo” the (descent (descent and ascent) mission. By measuring the noise from repeated yo missions, it was confirmed and ascent)hydrophone. mission. By measuring the noise from repeated yo missions, it was confirmed that there that there was no flow noise caused by turbulence between the underwater glider body and the was no flow noise caused by turbulence between the underwater glider body and the hydrophone. hydrophone.

Figure 3. Recorded noise signal when the underwater glider operates in the laboratory tank.

Figure 3.3. Recorded noise signal when the underwater glider operates inin thethe laboratorylaboratory tank.tank. Appl.Appl. Sci. Sci.2019 2019, 9, ,9 3860, x FOR PEER REVIEW 55 of of 13 13

(a) (b)

FigureFigure 4. 4. MeasurementMeasurement noise level when when the the underwater underwater glider glider operates operates in in the the laboratory laboratory tank: tank: (a) (aPiston,) Piston, (b ()b Pitch,) Pitch, (c ()c Air) Air Bladder, Bladder, and and (d ()d Altimeter.) Altimeter.

3.1.2.3.1.2. Glider’s Glider’s Operation Operation Noise Noise Measurement Measurement at at Sea Sea TheThe underwater underwater glider glider dived dived to to a deptha depth of 10–200of 10–200 m andm and repeated repeated six yosix missions yo missions during during about about 6 h. It6 took hours. about It took one about hour forone each hour yo. for Figure each 5yo. shows Figure the 5 commandshows the values command of the values driving of partthe driving according part toaccording the water to depth the when water the depth underwater when the glider underwater is operated glider at sea. is The operated driving-unit at sea. command The driving value-unit is Appl.command1 (low) Sci. 2019 to +,value 91, x (high). FOR is PEER−1 (low) REVIEW to +1 (high). 6 of 13 − When the underwater glider changes from descent to ascent, the values of the buoyancy control (Vol Piston) and the Pitch control (Pitch) change from −1 to +1; these values are negative during descent and positive during ascent. The Fin controller operated once every 3 to 4 seconds and the command value changed fast, regardless of descent or ascent. It can be seen that the Air Bladder command value is 0 except near the end when it is 1 as the bladder inflates raising the glider to the surface thereby allowing communication with the operator via satellite. The Fin and pumped CTD noise levels were calculated from recorded data at sea. The same hydrophone was mounted on the same position of the underwater glider body as at tank experiment. Due to the ambient noise contained in the recordings at sea, the reference sound level of Fin and pumped CTD is greater than that in the laboratory tank. The Fin noise level rose 4–5 dB at frequencies of 800–950 Hz (Figure 6a). The pumped CTD noise level rose by more than 10 dB at discrete frequencies of 357, 410, 710, and 1450 Hz (Figure 6b).

Figure 5. Command values when the glider is operating at sea. Figure 5. Command values when the glider is operating at sea.

(a) (b)

Figure 6. Measurement noise levels during underwater glider operation at sea: (a) Fin, and (b) Pumped Conductivity-Temperature-Depth profiler (CTD).

3.2. Pohang 2017 Earthquake: Analysis of in Situ Observations

3.2.1. In Situ Underwater Ambient Noise Measurement The measurement survey of underwater ambient noise was conducted using an underwater glider with a single hydrophone from 13 to 17 November 2017 from Ulgin to the Goseong area, along the east coast of South Korea in Figure 7. The underwater glider was launched at Ulgin 13 November 05:04 (UTC) and recovered at Goesong 17 November 06:20 (UTC) using the R/V Eodo of Korea Institute Ocean Science and Technology. The underwater glider moved vertically between depths of about 10 m and 200 m. During the survey, the glider advanced at an average horizontal speed of 0.02 m/s, logging data while repeatedly taking profiles between 10 and 200 m depth. The glider drifted when it surfaced between dives to transmit CTD data and Global Positioning System (GPS) data to the shore station through a satellite connection. The average surface current calculated from the glider’s GPS data was northward with a velocity of 0.35 m/s. A single omni-directional hydrophone (HTI92MIN from High Tech with a sensitivity of −168 dB re 1V/1μPa) was mounted on the underwater glider. We synchronized the time of the hydrophone with the underwater glider before launched the glider. The signal was digitized by the internal acoustic data logger at a 24 kHz sample rate with 16-bit resolution and 30 dB gain. Each file was approximately 255 MB in size, 60 minutes long; a total of 18.3 GB was stored on the 128 GB internal Appl. Sci. 2019, 9, x FOR PEER REVIEW 6 of 13

Appl. Sci. 2019, 9, 3860 6 of 13

When the underwater glider changes from descent to ascent, the values of the buoyancy control (Vol Piston) and the Pitch control (Pitch) change from 1 to +1; these values are negative during descent − and positive during ascent. The Fin controller operated once every 3 to 4 s and the command value changed fast, regardless of descent or ascent. It can be seen that the Air Bladder command value is 0 except near the end when it is 1 as the bladder inﬂates raising the glider to the surface thereby allowing communication with the operator via satellite. The Fin and pumped CTD noise levels were calculated from recorded data at sea. The same hydrophone was mounted on the same position of the underwater glider body as at tank experiment. Due to the ambient noise contained in the recordings at sea, the reference sound level of Fin and pumped CTD is greater than that in the laboratory tank. The Fin noise level rose 4–5 dB at frequencies of 800–950 Hz (Figure6a). The pumped CTD noise level rose by more than 10 dB at discrete frequencies Figure 5. Command values when the glider is operating at sea. of 357, 410, 710, and 1450 Hz (Figure6b).

(a) (b)

Figure 6.6. MeasurementMeasurement noise noise levels levels during during underwater underwater glider glider operation operation at sea: at (a ) sea: Fin, ( anda) Fin, (b) Pumped and (b) PumpedConductivity-Temperature-Depth Conductivity-Temperature proﬁler-Depth (CTD). profiler (CTD).

3.2. P Pohangohang 2017 2017 E Earthquake:arthquake: A Analysisnalysis of of in in S Situitu O Observationsbservations

3.3.2.1.2.1. In In S Situitu U Underwaternderwater A Ambientmbient N Noiseoise M Measurementeasurement The measurement measurement survey survey of of underwater underwater ambient ambient noise noise was was conducted conducted using using an underwater an underwater glider gliderwith a with single a hydrophonesingle hydrophone from 13 from to 17 13 November to 17 November 2017 from 2017 Ulgin from to Ulgin the Goseong to the Goseong area, along area, the along east thecoast east of coast South of Korea South in Korea Figure in7 Fig. Theure underwater 7. The underwater glider wasglider launched was launched at Ulgin at 13Ulgin November 13 November 05:04 05:04(UTC) (UTC) and recovered and recovered at Goesong at Goesong 17 November 17 November 06:20 (UTC) 06:20 using (UTC) the using R/V the Eodo R/V of EodoKorea of Institute Korea InstituteOcean Science Ocean and Science Technology. and Technology. The underwater The underwater glider moved glider vertically movedbetween vertically depths between of about depths 10 of m aboutand 200 10 m.m and During 200 them. During survey, the glidersurvey, advanced the glider at advanced an average at horizontalan average speed horizontal of 0.02 spe m/eds, logging of 0.02 m/s,data logging while repeatedly data while taking repeatedly profiles taking between profiles 10 and between 200 m depth. 10 and The 200 glider m depth. drifted The when glider it surfaced drifted whenbetween it surfaced dives to between transmit dives CTD datato transmit and Global CTD Positioningdata and Global System Positioning (GPS) data System to the (GPS) shore data station to thethrough shore a satellitestation through connection. a satellite The average connection. surface The current average calculated surface from current the glider’s calculated GPS from data was the glider’snorthward GPS with data a was velocity northward of 0.35 with m/s. a velocity of 0.35 m/s. A single omni-directional hydrophone (HTI92MIN from High Tech with a sensitivity of 168 dB A single omni-directional hydrophone (HTI92MIN from High Tech with a sensitivity of −168 dB µ rere 1V1V/1μPa)/1 Pa) waswas mountedmounted on on the the underwater underwater glider. glider. We We synchronized synchronized the the time time of the of hydrophonethe hydrophone with withthe underwater the underwater glider glider before before launched launched the glider. the glider. The signal The was signal digitized was digitized by the internal by the acousticinternal acousticdata logger data at logger a 24 kHz at samplea 24 kHz rate sample with 16-bit rate resolutionwith 16-bit and resolution 30 dB gain. and Each 30 dB file gain. was approximatelyEach file was approximately255 MB in size, 255 60 min MB long;in size, a total 60 minutes of 18.3 GBlong; was a total stored of on 18.3 the GB 128 was GB stored internal on flash the 128 memory GB internal during the 4-day survey. No acoustic data were transmitted in real time due to the large size of the files. All recorded data were processed using MATLAB software after the recovery of the glider. Recordings from the survey produced a one-dimensional time series of sound level spanning approximately 150 km from Ulgin to Geseong. During the survey, on 15 November 05:29:32 (UTC), a magnitude Appl. Sci. 2019, 9, x FOR PEER REVIEW 7 of 13 flash memory during the 4-day survey. No acoustic data were transmitted in real time due to the large size of the files. All recorded data were processed using MATLAB software after the recovery of Appl.the glider. Sci. 2019 ,Recordings9, 3860 from the survey produced a one-dimensional time series of sound level7 of 13 spanning approximately 150 km from Ulgin to Geseong. During the survey, on 15 November 05:29:32 (UTC), a magnitude 5.5 earthquake occurred at Pohang (epicenter 36.12°N, 129.36°E) [23,24]. 5.5 earthquake occurred at Pohang (epicenter 36.12 N, 129.36 E) [23,24]. Underwater sounds resulting Underwater sounds resulting from the momentary earthquake◦ ◦ were observed in the glider’s acoustic from the momentary earthquake were observed in the glider’s acoustic record. record.

Figure 7. Underwater glider survey track (red line) map (seismic station: (a) Busan (BUS), (b) Chungju Figure 7. Underwater glider survey track (red line) map (seismic station: (a) Busan (BUS), (b) (CHJ), (c) Namwon (NAWB), (d) Seawha (SEHB), (e) Seoul (SEO)). Chungju (CHJ), (c) Namwon (NAWB), (d) Seawha (SEHB), (e) Seoul (SEO)). 3.2.2. Underwater Position of Underwater Glider 3.2.2. Underwater Position of Underwater Glider The glider surfaced every 5 h and sent the GPS location information during the survey. The underwaterThe glider trajectory surfaced ofevery the glider 5 hours was and calculated sent the byGPS a linearlocation interpolation information between during every the survey. two surfaced The underwaterGPS points. trajectory When the of the earthquake glider was occurred calculated at 05:29:32 by a linear (UTC), interpolation the surfaced between GPS positions every two were surfaced37.3482 GPS◦N, 129.6948 points. ◦WhenE at 00:55:19 the earthqu (UTC)ake and occurred 37.3806◦ N,at 05:29:32 129.6994 ◦(UTC),E at 05:52:46 the surfaced (UTC), respectively.GPS positions The werecalculated 37.3482°N, GPS position 129.6948°E of the at glider 00:55:19 is at 37.378 (UTC)◦ N, and 129.699 37.3806°N,◦E, approximately 129.6994°E 140 at km 05:52:46 from the (UTC), Pohang respectively.epicenter when The calculated the glider recordedGPS position acoustic of the sounds glider from is at the 37.378°N, earthquake 129.699°E, at 05:29:51 approximately (UTC). The depth140 kmof fro them seafloor the Pohang bottom epicenter and the when glider the at the glider GPS recorded position wasacoustic at about sounds 430 mfrom and the 200 earthquake m, respectively. at 05:29:51 (UTC). The depth of the seafloor bottom and the glider at the GPS position was at about 430 m 3.2.3.and 200 Time m, andrespectively. Frequency Analysis of Earthquake Underwater Sound It is most common for noise signals to be represented in the frequency domain as noise spectral 3.2.3. Time and Frequency Analysis of Earthquake Underwater Sound levels plotted as a function of frequency [25,26]. The data for ambient noise is usually expressed as spectralIt is most density common (SD) levelsfor noise or powersignals spectral to be represented density (PSD), in the where frequency the data domain in each as noise frequency spectral band levelshas beenplotted normalized as a function by dividing of frequency by the [25,26] bandwidth. The data of the for frequency ambient noise band. is The usually units expressed of the levels as in spectraleach band density are then(SD) dBlevels re 1 orµPa power2/Hz. spectral We used density the following (PSD), analysiswhere the method data in to each distinguish frequency earthquake band hasunderwater been normalized sounds. by Extraction dividing ofby necessary the bandwidth timing of data the fromfrequency the recording band. The files; units Calculation of the levels of Power in eachSpectrum band are Density then (PSD) dB re using 1 μPa a2 time/Hz. window We used of the chosen following length analysisin each recording. method to The distinguish length and earthquakeoverlap of underwater FFT window sounds. for 1s and Extraction 50 %, respectively; of necessary Correction timing of dataPSD for from recorder’s the recording sensitivity, files; gain Calculationand, when of possible, Power frequency Spectrum response Density to (PSD) get the using output a time data window in sound ofpressure chosen units length (µPa). in each Figure8 shows PSD of a 1 h recording started at 05:29:35 (UTC) collected by the glider when the earthquake occurred at 05:29:32 (UTC). It contains underwater earthquake sound as well as the Appl. Sci. 2019, 9, x FOR PEER REVIEW 8 of 13 recording. The length and overlap of FFT window for 1s and 50 %, respectively; Correction of PSD for recorder’s sensitivity, gain and, when possible, frequency response to get the output data in sound pressureAppl. Sci. units2019 ,(μPa).9, 3860 8 of 13 Figure 8 shows PSD of a 1 hour recording started at 05:29:35 (UTC) collected by the glider when the earthquake occurred at 05:29:32 (UTC). It contains underwater earthquake sound as well as the operating noise of the glider’s machinery mentioned in Section3, e.g., piston and pitch motors, fin, operating noise of the glider’s machinery mentioned in section III e.g., piston and pitch motors, fin, airbladder, altimeter, pumped CTD. The glider descended with pumped CTD and altimeter turned airbladder, altimeter, pumped CTD. The glider descended with pumped CTD and altimeter turned on until it reached the turning depth (200 m), and it then turned on the piston and pitch motor for on until it reached the turning depth (200 m), and it then turned on the piston and pitch motor for ascending and inflated the external air bladder to begin surfacing for communication. The pumped ascending and inflated the external air bladder to begin surfacing for communication. The pumped CTD and altimeter were turned off when the glider was ascending. The earthquake sounds were CTD and altimeter were turned off when the glider was ascending. The earthquake sounds were dominated below about 50 Hz. dominated below about 50 Hz.

FigureFigure 8. Spectrogram 8. Spectrogram of earthquake of earthquake underwater underwater sound sound recording recording (started (started at 05:29:35 at 05:29:35 UTC). UTC).

TheThe time time sequence sequence of of the the acoustic signal signal has has been been reconstructed, reconstructed, from from the wav-filethe wav- recordsfile records starting startingat 05:29:35 at 05:29:35 (UTC) (UTC) in Figure in Fig9a.ure This 9a time. This is time the start is the reference start reference time t0 time= 0s t (i.e.,0 = 0 whens (i.e., the when recording the recordingstarts at starts 05:29:35); at 05:29:35); 16 s later 16 (i.e., s latert1 = (i.e.,05:29:51), t1 = 05:29:51), is the time is ofthe first time arrival of first of thearrival earthquake of the earthquake sounds at the soundsglider’s at position, the glider’s at about position, 200m at depth. about The 200m arrival depth. time The (t1) at arrival about time 19 s from (t1) atearthquake about 19 correspond s from earthquaketo about 7correspond km/s which to may about be 7 the km/s speed which of seismic may be waves the speed in the of crust seismic just waves up to the in bottomthe crust of just detection up to thepoint, bottom then of in thedetection water point, column then and in finally the water detected column by theand underwater finally detected glider. by Thesethe underwater sounds were glider.observed These for sounds over 30were s. observed for over 30 seconds. SpectrumSpectrum and and spectrogram spectrogram analyses analyses of the of the earthquake earthquake sound sound have have been been carried carried out. out. By Bycarefully carefully examiningexamining the the frequenc frequencyy spectrum spectrum of theof the record record in inFig Figureure 9b,9 b, it it can can be be seen seen that that the the dominant dominant frequencyfrequency of theof the earthquake earthquake sounds sounds was was about about 10 10 Hz. Hz. Fig Figureure 9c 9 showsc shows that that after after the the earthquake earthquake occurred,occurred, the the average average ambient ambient noise noise level level at frequencies at frequencies below below 50 Hz 50 Hzwas was appr approximatelyoximately 60 dB 60 dBhigher higher thanthan the the ambient ambient noise noise level level before before the the earthquake earthquake occurred. occurred. The The earthquake earthquake acoustic acoustic signal signal can can be be distinguisheddistinguished from from glider’s glider’s self self-noises-noises of fin,of fin, pumped pumped CTD CTD and and altimeter. altimeter. The The fin fin noise noise exists exists at at frequenciesfrequencies of of800 800–950–950 Hz Hz every every 3 3to to 4 4 seconds. s. The pumped The pumped CTD sensor CTD noisesensor exists noise at exists discrete at frequenciesdiscrete frequenciesof 357, 410, of 710,357, 1450410, Hz.710, The1450 dominant Hz. The dominant CTD noise CTD frequency noise isfrequency 710 Hz. Theis 710 altimeter Hz. The noise altimeter exists at noisefrequencies exists at frequencies above 100 Hz. above 100 Hz. Appl. Sci. 2019, 9, 3860 9 of 13 Appl. Sci. 2019, 9, x FOR PEER REVIEW 9 of 13

(a) (b)

(c)

FigureFigure 9.9. TheThe earthquakeearthquake acousticacoustic signal:signal : (a(a)) Time Time reconstructed reconstructed signal, signal, ( b(b)) frequency frequency spectrum spectrum andand (c(c)) Spectrogram. Spectrogram.

3.2.4.3.2.4. CorrelationCorrelation withwith SituSitu SeismicSeismic WavesWaves andand EarthquakeEarthquake Sounds Sounds EarthquakeEarthquake generatedgenerated acousticacoustic waveswaves inin thethe oceanocean areare referredreferred toto asas T-waveT-wave becausebecause inin certaincertain conditions,conditions, whenwhen theythey reachreach thethe shore,shore, theythey maymay convertconvert backback toto seismicseismic waveswaves andand arrivearrive thirdthird afterafter thethe P-P- andand S-seismicS-seismic waveswaves onon near-shorenear-shore seismological seismological stations. stations. IRISIRIS (Incorporated(Incorporated ResearchResearch InstitutionsInstitutions forfor Seismology)Seismology) archivesarchives waveformwaveform (time-series)(time-series) datadata fromfrom stations stations around around the the world. world. Seismic Seismic wave wave data data of of the the Pohang Pohang Mw Mw 5.5 5.5 earthquake earthquake was was collected collected by theby the Busan Busan (35.25 (35.25°N,◦N, 129.11 129.11°E),◦E), Chungju Chungju (36.87 (36.87°N,◦N, 127.97 127.97°E),◦E), Namwon Namwon (35.42 (35.42°N,◦N, 127.40 127.40°E),◦E), SeawhaSeawha (38.27(38.27°N,◦N, 128.25128.25°E),◦E), SeoulSeoul (37.49(37.49°N,◦N, 126.92126.92°E)◦E) stationstation inin FigureFigure7 .7. The The Busan Busan (BUS) (BUS) station station was was about about 9292 kmkm awayaway fromfrom thethe PohangPohang epicenter;epicenter; thethe ChungjuChungju (CHJ)(CHJ) stationstation waswas 147147 km,km, NamwonNamwon (NAWB)(NAWB) stationstation waswas 183183 km,km, SeawhaSeawha (SEHB)(SEHB)station station andand SeoulSeoul (SEO)(SEO) station station were were 255 255 km km same same distance. distance. TheThe underwaterunderwater gliderglider waswas aboutabout 140140 kmkm awayaway fromfrom the the Pohang Pohang epicenter. epicenter. TheThe timetime series series of of the the seismic seismic and and earthquake earthquake acoustic acoustic signal signal detected detected at theat the five five stations stations and and the underwaterthe underwater glider glider are presentedare presented in Figure in Fig 10ure. The10. The time time of arrival of arrival of theof the seismic seismic waves waves observed observed at theat the station station is proportionalis proportional to to the the distance distance from from the the epicenter. epicenter. Since Since the the underwater underwater glider glider was was closer closer thanthan ChungjuChungju stationstation toto thethe epicenter,epicenter, thethe arrivalarrival timetime ofof thethe T-waveT-wave (i.e.,(i.e., earthquakeearthquake acousticacoustic signal)signal) observedobserved by by thethe underwaterunderwater glider glider was was earlierearlier thanthan thatthat ofof thethe seismicseismic waveswaves observedobserved atat thethe ChungjuChungju station.station. TheThe seismicseismic P- P- and and S-waves S-waves travel travel at at velocities velocities from from 2000 2000 to 7000to 7000 m/ sm/s in thein the crust crust just just up toup the to seafloorthe seafloor bottom bottom (about (about 430 m) 430 of m) detection of detection point, point, then in then the in water the columnwater column and finally and finally detected detected by the underwaterby the underwater glider (aboutglider 200(about m). 200 It is m). clearly It is clearly shown shown that the that acoustic the acoustic signal detectedsignal detected by the gliderby the fromglider the from Pohang the Pohang earthquake earthquake is a Tertiary is a Tertiary wave (T-wave) wave (T in-wave) Figure in 11 Fig. ure 11. Appl. Sci. 2019, 9, x FOR PEER REVIEW 10 of 13

Appl. Sci. 2019, 9, 3860 10 of 13 Appl. Sci. 2019, 9, x FOR PEER REVIEW 10 of 13

Figure 10. Time signal of the seismic and T-waves observed by the seismic observatory and Figure 10. Time signal of the seismic and T-waves observed by the seismic observatory and underwater underwaterFigure glider 10. Timeposition signal from of the the Pohang seismic andearthquake T-waves epicenter. observed by the seismic observatory and glider position from the Pohang earthquake epicenter. underwater glider position from the Pohang earthquake epicenter.

FigureFigure 11. 11.The The creation creation of of T-waves T-waves from from seismic seismic waves waves and and detection detection by by the the underwater underwater glider. glider. Figure 11. The creation of T-waves from seismic waves and detection by the underwater glider. 4. Discussion 4. Discussion Early models of the Slocum underwater glider were equipped with a non-pumped CTD. The Earlynon models-pumped of CTD the has Slocum a large underwater salinity uncertainty glider because were equippedthe flow rate with in the a conductivitynon-pumped cell CTD.is not The non-pumped CTD has a large salinity uncertainty because the flow rate in the conductivity cell is not Appl. Sci. 2019, 9, 3860 11 of 13

4. Discussion Early models of the Slocum underwater glider were equipped with a non-pumped CTD. The non-pumped CTD has a large salinity uncertainty because the flow rate in the conductivity cell is not constant. The salinity measurement uncertainty of a non-pumped CTD was about 30 times greater than that of a pumped CTD [27]. To obtain good quality ocean observation data, recently-produced Slocum underwater gliders are equipped only with a pumped CTD. In terms of the overall operating time of the underwater glider, the pumped CTD noise is higher and it occurs for a longer time than noise from other sources. It is the main noise source in underwater glider operation. When an acoustic source emits in a test tank, acoustic waves are reflected by the walls, the surface, and the bottom. The multiple reflections give rise to a reverberated field. Measuring underwater sound power in reverberant environment such as a tank has advantages of speed but is limited in accuracy and frequency range, but it has been used to good effect in specific applications. For small sources (or small noisy components) like underwater glider, it is possible to make measurements of radiated noise in reverberant tanks [28]. For the seismic observation of seafloor and coastal earthquake disaster risk assessment, a comprehensive analysis of seismic sources is needed. This requires simultaneously observing underwater phenomena caused by seafloor earthquakes. Moreover, submarine earthquake underwater acoustical observation technology using underwater gliders is very synergistic because it can provide basic research data on natural, anthropogenic, and ecological underwater sources in the ocean. In addition, underwater sound observations using underwater gliders can be of great assistance in developing earth observation systems by providing submarine platforms ensuring stable operational technology and linking with existing earthquake observation networks. To this end, improving the method for estimating a glider’s underwater trajectory is important because it is a challenging task due to several errors in the linear interpolation method. It is not realistic because the position is estimated by dividing the distance between the real GPS fixed position by transit time, and the glider’s direction and speed remain constant. In other words, the trajectory looks like a straight line. Our future work aims to provide a better method for estimating a glider’s optimal underwater trajectory, focusing on the importance of the ocean currents and underwater wireless communication technology to transmit the acoustic data observed by the underwater glider to the shore in real time through the OSG such as WaveGliders in [20].

5. Conclusions In this study, the acoustic sounds of earthquake occurred on 15 November 2017 in the Pohang area of South Korea observed by underwater glider were analyzed. By analysis of time correlation with seismic waves detected by ﬁve seismic land stations and the earthquake acoustic signal, it is clearly shown that the seismic waves converted to Tertiary waves (i.e., earthquake acoustic signal) and then it detected by the underwater glider. In order to distinguish the earthquake sound from the glider’s operation noise, Sound Pressure Level (SPL) of the noise sources associated with the driving and sensor parts of the underwater glider were analyzed and measured at laboratory tank and sea. The earthquake acoustic signal in the range of 10–50 Hz was distinguished from glider’s self-noises which are exist over 100 Hz. The results allow constraining the acoustic sound level of the earthquake and suggest that the glider provides an eﬀective platform for enhancing the earth seismic observation systems and monitoring natural and anthropogenic ocean sounds.

Author Contributions: Conceptualization, methodology, investigation, writing—original draft preparation, writing—review and editing, J.-H.L., S.-H.K. and J.-W.H.; project administration, S.-K.J. Funding: This research was funded by Korea Institute of Ocean Science & Technology, grant number PE99741. Acknowledgments: This research was part of the project titled “Development of maritime defense and security technology [PE99741]” funded by Korea Institute of Ocean Science & Technology in 2019. Conﬂicts of Interest: The authors declare no conﬂict of interest. Appl. Sci. 2019, 9, 3860 12 of 13

References

1. Momma, H.; Fujiwara, N.; Kawaguchi, K.; Iwase, R.; Suzuki, S.; Kinoshita, H. Monitoring system for submarine earthquakes and deep sea environment. In Proceedings of the OCEANS’97. MTS/IEEE Conference Proceedings, Halifax, NS, Canada, 6–9 October 1997; pp. 1453–1459. 2. Baba, T.; Hirata, K.; Kaneda, Y. Tsunami magnitudes determined from ocean-bottom pressure gauge data around Japan. Geophys. Res. Lett. 2004, 31.[CrossRef] 3. Nosov, M.; Kolesov, S.; Denisova, A.; Alekseev, A.; Levin, B. On the near-bottom pressure variations in the region of the 2003 Tokachi-Oki tsunami source. Oceanology 2007, 47, 26–32. [CrossRef] 4. Bolshakova, A.; Inoue, S.; Kolesov, S.; Matsumoto, H.; Nosov, M.; Ohmachi, T. Hydroacoustic eﬀects in the 2003 Tokachi-oki tsunami source. Russ. J. Earth Sci. 2011, 12, 1–14. [CrossRef] 5. Cecioni, C.; Romano, A.; Bellotti, G.; De Girolamo, P. Hydroacoustic waves measured during the 2012 Negros-Cebu earthquake. J. Waterw. Port Coast. Ocean Eng. 2018, 144.[CrossRef] 6. Abdolali, A.; Cecioni, C.; Bellotti, G.; Kirby, J.T. Hydro-acoustic and tsunami waves generated by the 2012 Haida Gwaii earthquake: Modeling and in situ measurements. J. Geophys. Res. Oceans 2015, 120, 958–971. [CrossRef] 7. Holmes, J.D.; Carey, W.M.; Lynch, J.F. An overview of unmanned underwater vehicle noise in the low to mid frequencies bands. Proc. Meet. Acoust. 2010, 9.[CrossRef] 8. Silva, A.; Matos, A.; Soares, C.; Alves, J.C.; Valente, J.; Zabel, F.; Cabral, H.; Abreu, N.; Cruz, N.; Almeida, R. Measuring underwater noise with high endurance surface and underwater autonomous vehicles. In Proceedings of the 2013 OCEANS-San Diego, San Diego, CA, USA, 23–26 September 2013; pp. 1–6. 9. Jiang, Y.-M.; Alvarez, A.; Cecchi, D.; Garau, B.; Micheli, M. In-situ acoustic received level measurements with glider based reactive behaviour. In Proceedings of the OCEANS 2015-Genova, Genova, Italy, 18–21 May 2015; pp. 1–6. 10. Takahashi, R.; Akamatsu, T.; Hasegawa, D.; Okunishi, T.; Nishida, Y.; Ura, T.; Takahashi, H.; Katsuragawa, M. Evaluation of availability on passive acoustic devices on underwater platforms. In Proceedings of the 2016 Techno-Ocean (Techno-Ocean), Kobe, Japan, 6–8 October 2016; pp. 317–320. 11. Chandrayadula, T.K.; Miller, C.W.; Joseph, J. Monterey Bay ambient noise proﬁles using underwater gliders. J. Acoust. Soc. Am. 2013, 133, 3395. [CrossRef] 12. Cauchy, P.; Testor, P.; Mortier, L.; Beguery, L.; Bouin, M.-N. Passive acoustics embedded on gliders—Weather observation through ambient noise. J. Acoust. Soc. Am. 2014, 135, 2306–2307. [CrossRef] 13. Ferguson, B.G.; Lo, K.W.; Rodgers, J.D. Sensing the underwater acoustic environment with a single hydrophone onboard an undersea glider. In Proceedings of the OCEANS 2010 IEEE-Sydney, Sydney, Australia, 24–27 May 2010; pp. 1–5. 14. Maguer, A.; Dymond, R.; Grati, A.; Stoner, R.; Guerrini, P.; Troiano, L.; Alvarez, A. Ocean gliders payloads for persistent maritime surveillance and monitoring. In Proceedings of the 2013 OCEANS-San Diego, San Diego, CA, USA, 23–26 September 2013; pp. 1–8. 15. Nielsen, P.L.; Siderius, M.; Muzi, L. Glider-based seabed characterization using natural-made ambient noise. In Proceedings of the OCEANS 2015-Genova, Genova, Italy, 18–21 May 2015; pp. 1–7. 16. Tesei, A.; Been, R.; Williams, D.; Cardeira, B.; Galletti, D.; Cecchi, D.; Garau, B.; Maguer, A. Passive acoustic surveillance of surface vessels using tridimensional array on an underwater glider. In Proceedings of the OCEANS 2015-Genova, Genova, Italy, 18–21 May 2015; pp. 1–8. 17. Roth, E.H.; Hildebrand, J.A.; Wiggins, S.M.; Ross, D. Underwater ambient noise on the Chukchi Sea continental slope from 2006–2009. J. Acoust. Soc. Am. 2012, 131, 104–110. [CrossRef][PubMed] 18. Spindel, R.C.; Na, J.; Dahl, P.H.; Oh, S.; Eggen, C.; Kim, Y.G.; Akulichev, V.A.; Morgunov, Y.N. Acoustic tomography for monitoring the Sea of Japan: A pilot experiment. IEEE J. Ocean. Eng. 2003, 28, 297–302. [CrossRef] 19. Matsumoto, H.; Stalin, S.; Embley, R.; Haxel, J.; Bohnenstiehl, D.; Dziak, R.; Meinig, C.; Resing, J.; Delich, N. Hydroacoustics of a submarine eruption in the Northeast Lau Basin using an acoustic glider. In Proceedings of the OCEANS 2010 IEEE-Sydney, Sydney, Australia, 24–27 May 2010; pp. 1–6. 20. Berger, J.; Laske, G.; Babcock, J.; Orcutt, J. An ocean bottom seismic observatory with near real-time telemetry. Earth Space Sci. 2016, 3, 68–77. [CrossRef] Appl. Sci. 2019, 9, 3860 13 of 13

21. Miller, J.H.; Potty, G.R. Modeling and Measuring Variability in 3-D Acoustic Normal Mode Propagation in Shallow Water Near Ocean Fronts Using Fixed and Moving Sources and Receivers; ADA573346; Rhode Island University Narragansett Graduate School of Oceanography: Narragansett, RI, USA, 2007. 22. Liu, L.; Xiao, L.; Lan, S.-Q.; Liu, T.-T.; Song, G.-L. Using Petrel II Glider to Analyze Underwater Noise Spectrogram in the South China Sea. Acoust. Aust. 2018, 46, 151–158. [CrossRef] 23. Grigoli, F.; Cesca, S.; Rinaldi, A.; Manconi, A.; López-Comino, J.; Clinton, J.; Westaway, R.; Cauzzi, C.; Dahm, T.; Wiemer, S. The November 2017 Mw 5.5 Pohang earthquake: A possible case of induced seismicity in South Korea. Science 2018, 360, 1003–1006. [CrossRef][PubMed] 24. Kim, K.-H.; Ree, J.-H.; Kim, Y.; Kim, S.; Kang, S.Y.; Seo, W. Assessing whether the 2017 Mw 5.4 Pohang earthquake in South Korea was an induced event. Science 2018, 360, 1007–1009. [CrossRef][PubMed] 25. Merchant, N.D.; Barton, T.R.; Thompson, P.M.; Pirotta, E.; Dakin, D.T.; Dorocicz, J. Spectral probability density as a tool for ambient noise analysis. J. Acoust. Soc. Am. 2013, 133, EL262–EL267. [CrossRef][PubMed] 26. Merchant, N.D.; Blondel, P.; Dakin, D.T.; Dorocicz, J. Averaging underwater noise levels for environmental assessment of shipping. J. Acoust. Soc. Am. 2012, 132, EL343–EL349. [CrossRef][PubMed] 27. Alvarez, A.; Stoner, R.; Maguer, A. Performance of pumped and un-pumped CTDs in an underwater glider. In Proceedings of the 2013 OCEANS-San Diego, San Diego, CA, USA, 23–26 September 2013; pp. 1–5. 28. Robinson, S.P.; Lepper, P.A.; Hazelwood, R.A. Good Practice Guide for Underwater Noise Measurement; No.133; National Measurement Oﬃce: Teddington, UK, 2014; pp. 1–95.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).