*3.2. Spectra Comparison of ISW Noise and Ambient Noise*

This section served to illustrate the effects of ISW noise on the ambient noise. Figure 4a shows the power spectra of the noise received by the hydrophone located at 700 m during the passage of internal waves and the ambient noise received three hours later (from 9:00 to 9:02 UTC+8). The data duration was 2 min. The differences in PSD between the two PSD curves are shown in Figure 4b. The spectral level of ISW noise was about 20–40 dB higher than that of ambient noise at frequencies between 10 Hz and 50 Hz. Due to the effects of low frequency harmonics, the maximum difference was 47.7 dB. As the frequency increased, the spectral level gap narrowed. Up to 2 kHz, the difference remained around 10 dB. Flow noise was the dominant source below 100 Hz. Distant ships may be responsible for the peaks in the ambient noise between 200 Hz and 500 Hz. The noise comparison revealed that the ambient noise was completely drowned out by the ISW noise below 2 kHz.

**Figure 4.** Power spectrum comparison of ISW noise and ambient noise. (**a**) PSD comparison. (**b**) The differences of PSD between ISW noise and ambient noise. ISW noise is from the 2 min data received by hydrophone C058 at depth of 700 m between 6:00 to 6:02 UTC+8 on 21 September 2016, and the ambient noise is from data three hours later (from 9:00 to 9:02 UTC+8). **Figure 4.** Power spectrum comparison of ISW noise and ambient noise. (**a**) PSD comparison. (**b**) The differences of PSD between ISW noise and ambient noise. ISW noise is from the 2 min data received by hydrophone C058 at depth of 700 m between 6:00 to 6:02 UTC+8 on 21 September 2016, and the ambient noise is from data three hours later (from 9:00 to 9:02 UTC+8).

during the passage of internal waves and the ambient noise received three hours later (from 9:00 to 9:02 UTC+8). The data duration was 2 min. The differences in PSD between the two PSD curves are shown in Figure 4b. The spectral level of ISW noise was about 20– 40 dB higher than that of ambient noise at frequencies between 10 Hz and 50 Hz. Due to the effects of low frequency harmonics, the maximum difference was 47.7 dB. As the frequency increased, the spectral level gap narrowed. Up to 2 kHz, the difference remained around 10 dB. Flow noise was the dominant source below 100 Hz. Distant ships may be responsible for the peaks in the ambient noise between 200 Hz and 500 Hz. The noise comparison revealed that the ambient noise was completely drowned out by the ISW noise

#### *3.3. Low-Frequency Noise Induced by ISWs 3.3. Low-Frequency Noise Induced by ISWs*

below 2 kHz.

#### 3.3.1. Relationship between Low-Frequency Noise and ISWs 3.3.1. Relationship between Low-Frequency Noise and ISWs

We interpreted the observed low-frequency noise as the flow noise induced by ISWs [23,27]. For this purpose, we analyzed data collected from other ISWs. Figure 5 shows the time–frequency spectra of low frequency noise and their corresponding flow velocities (The data time is consistent with Table 1). The increase in flow velocity was almost synchronous with the burst of low frequency noise. The noise on 21 September 2016 was incomplete, as the maximum velocity occurred at 5:54 UTC+8. As discussed, these results were in accordance with the properties of flow noise induced by ISWs. We interpreted the observed low-frequency noise as the flow noise induced by ISWs [23,27]. For this purpose, we analyzed data collected from other ISWs. Figure 5 shows the time–frequency spectra of low frequency noise and their corresponding flowvelocities (The data time is consistent with Table 1). The increase in flow velocity was almost synchronous with the burst of low frequency noise. The noise on 21 September 2016 was incomplete, as the maximum velocity occurred at 5:54 UTC+8. As discussed, these results were in accordance with the properties of flow noise induced by ISWs. *J. Mar. Sci. Eng.* **2022**, *10*, x FOR PEER REVIEW 7 of 14

**Figure 5.** Flow velocities and time–frequency spectra (dB) of the C058 hydrophone (700 m) on (**a**,**b**) 14 March 2017. Flow velocities and time–frequency spectra of the C010 hydrophone (950 m) on (**c**,**d**) 1 October 2016 and (**e**,**f**) 21 September 2016. The depth on the figures represents the depth at which the flow velocity was measured. **Figure 5.** Flow velocities and time–frequency spectra (dB) of the C058 hydrophone (700 m) on (**a**,**b**) 14 March 2017. Flow velocities and time–frequency spectra of the C010 hydrophone (950 m) on (**c**,**d**) 1 October 2016 and (**e**,**f**) 21 September 2016. The depth on the figures represents the depth at which the flow velocity was measured.

It was interesting to note that each segment of noise induced by the ISWs had fundamental and harmonic waves. To elucidate the characteristics of the harmonic waves, we used a 16.384 s data block with 50% overlap (65536-points FFT) from the center of each It was interesting to note that each segment of noise induced by the ISWs had fundamental and harmonic waves. To elucidate the characteristics of the harmonic waves, we used a 16.384 s data block with 50% overlap (65536-points FFT) from the center of each

signal for spectrum analysis. The Welch method and Blackman window were used to calculate power spectra. Figure 6 shows the power spectra corresponding to the three signals

signal had a different fundamental frequency. This phenomenon may be associated with the velocity of the ISW. It was worth discussing these facts, in terms of which part of the mooring system the internal wave interacted with, to produce low-frequency flow noise

and harmonics.

PSD (dB)

PSD (dB)

PSD (dB)

signal for spectrum analysis. The Welch method and Blackman window were used to calculate power spectra. Figure 6 shows the power spectra corresponding to the three signals in Figure 5. Table 2 shows their fundamental and harmonic frequencies. Each segment of signal had a different fundamental frequency. This phenomenon may be associated with the velocity of the ISW. It was worth discussing these facts, in terms of which part of the mooring system the internal wave interacted with, to produce low-frequency flow noise and harmonics. signal for spectrum analysis. The Welch method and Blackman window were used to calculate power spectra. Figure 6 shows the power spectra corresponding to the three signals in Figure 5. Table 2 shows their fundamental and harmonic frequencies. Each segment of signal had a different fundamental frequency. This phenomenon may be associated with the velocity of the ISW. It was worth discussing these facts, in terms of which part of the mooring system the internal wave interacted with, to produce low-frequency flow noise and harmonics.

**Figure 5.** Flow velocities and time–frequency spectra (dB) of the C058 hydrophone (700 m) on (**a**,**b**) 14 March 2017. Flow velocities and time–frequency spectra of the C010 hydrophone (950 m) on (**c**,**d**) 1 October 2016 and (**e**,**f**) 21 September 2016. The depth on the figures represents the depth at which

It was interesting to note that each segment of noise induced by the ISWs had fundamental and harmonic waves. To elucidate the characteristics of the harmonic waves, we used a 16.384 s data block with 50% overlap (65536-points FFT) from the center of each

*J. Mar. Sci. Eng.* **2022**, *10*, x FOR PEER REVIEW 7 of 14

the flow velocity was measured.

**Figure 6.** Power spectra of the noise recorded by the same hydrophones as in Figure 5. (**a**) The data after 18:09 UTC+8. (**b**) The data after 18:06 UTC+8. (**c**) The data after 6:00 UTC+8. The data length is 16.384 s.

**Table 2.** The fundamental and harmonic frequencies of the three signals (Hz).

