MODELING ACOUSTIC PROPAGATION NEAR THE ATLANTIS II SEAMOUNTS

Acoustic modeling was used to predict ray paths for a future experiment at the Atlantis II seamount, part of the New England Seamount Chain. Historical sound speed data from the area around the Atlantis II Seamounts was analyzed to build the acoustic prediction model. The results of the prediction model produced propagation paths for various sound speed profiles (SSP)s. The model was used to assist in the placement of acoustic sensors prior to the experiment. The acoustic environment varies significantly due to the seamount’s location and the presence of the Gulf Stream. The model has been designed to accept the most recent SSP obtained by the team onboard the research vessel to determine the final placement of the sound sources based on the ray paths calculated by the model.


List of Figures
One goal of the NESMA research is to understand the effect of seafloor properties on and around the seamount on low frequency acoustic propagation. This thesis used the ray tracing model Bellhop (Porter M. , 2022) to predict sound ray paths at various ranges and azimuths around the seamount. To model these ray paths, sound speed profiles (SSPs) were obtained from around the seamount location. These SSPs and bathymetry data have been used to predict the paths and the expected variability in the sound speed profile for the area. This resultant ray paths and transmission loss plots determined the range of the sound source for the pilot experiment.
A few source locations were assumed at distances varying from 10 km to 100 km on several bearings from where WHOI plans to deploy a volumetric hydrophone array called the Acoustic Telescope (WHOI, 2022) The AT will be placed at near summit of Atlantis II Seamount and a depth of 250 meters from the surface. At these distances, the sound speed profile will have a significant effect on the sound propagation. Therefore, changes in the environment were reviewed using historical data from the location. This history review produced 2 SSPs for use in the prediction model.

Chapter 2. Oceanographic Environment Ocean Bottom
The ocean floor of the New England Seamounts is primarily basalt and fine-grained sediment. Between the basalt and the fine-grained sediments is a thin layer of limestone. (Il'in, 2000) The seamounts are also home to multiple biological environments.
Previously, the Atlantic II Seamount has been discovered to have a mix of deep-sea coral and sponges (Abby Lapointe, 2020). Since the monitoring equipment will be placed on the ocean bottom, the seamount has been surveyed to allow for placing the sensors without disturbing the natural biological community.

Sound propagation
The sound speed profiles along with the seafloor bathymetry allows us to model the sound propagation. In analyzing acoustic propagation, sound waves will expand outward from a source. The wavefront is the surface expanding out from the sound, typically this starts as a sphere for a point source. The sound rays are perpendicular to the wavefront.
For a large body of water with the same sound speed, this would continue indefinitely until the energy is dissipated through geometric expansion or absorption into the water.
The sound rays can also be refracted (or bent) within the water due to changing sound speed. In general, this refraction can be in three dimensions. However, in this study, we are primarily focused on the bending in the vertical direction as sound speed varies with depth. This refraction from the higher towards the lower sound speed is cause by the change in sound speed and is commonly calculated using Snell's Law. (DOSITS, Science of Sound, 2023) Like a rolling shopping cart with a slow wheel, the sound ray bends toward the lower sound speed.
However, for most frequencies, the sound energy will remain strong enough to travel to the surface and bottom of the body of water. This will cause the sound waves to reflect off these boundaries. The percentage of sound that reflects will be related to the material properties of both mediums and the angle of incidence. The reflection coefficient depends on the density and the sound speed of the two materials that create the surface of the reflection. Generally, the greater the change, the higher percentage of acoustic energy will be reflected. Next, as the incident angle becomes closer to perpendicular, less energy will be reflected. The energy that is not reflected is considered loss, as it is transferred out of the water and into the impacted material such as the seafloor.

Sound Speed Analysis and Ocean Currents
Sound speed data was calculated using temperature, salinity, and depth data collected from the World Ocean Database (NOAA, 2021). The data was converted to sound speed with the Medwin Sound Speed Calculation (Medwin, 1975). The ocean data was collected from all available CTD and XBT datasets within 0.5-degree latitude and longitude from the seamount. This created a search area of about 110 km by 90 km centered on the seamount. The data was then screened for large gaps and the depth points aligned. The 163 datasets remained for analysis and use in the model. Bathymetry was collected from the NOAA charts (NOAA, 2022).
By reviewing the collected sound speed, we can predict the variation of the oceanographic and therefore the acoustic environment around the seamount. The variations in the sound speed profiles around the seamount can be attributed to the variations in the salinity and the temperature of different water masses. The area where the temperature is changing with depth is called the thermocline and ends around 1200 to 1400 meters in depth. Below this depth, the sound speed profile remains relatively constant in time which only leaves depth affecting the sound speed. An analysis of the data shows that the sound speed cycled between two distinct patterns. The pattern can also be seen in the temperature and salinity data. The temperature, salinity, and sound speed data was plotted monthly and seasonally to rule out month and season as causes   Analyzing the NCOM data shows that a pattern can be attributed to changes in the location of the Gulf Stream with the difference between the cold and warm sides of the front changing the sound speed. These two water masses are the Slope Water and the Sargasso Sea Water respectively.
The data shown in Figure 6 allow us to also compare the salinity change with the temperature change. We would expect that the warm water from the Gulf Stream will have a higher salinity due to the evaporation of the seawater in the tropical region.
Meanwhile, the colder water (Slope Sea) to the north of the Gulf Stream should have more freshwater runoff and ice melt from the polar region leading to a lower salinity. This pattern is present in the data and matched expectations.
Lower salinity and lower temperatures reduce sound speed for a given depth. Therefore, the calculated sound speed repeats this pattern. Between 1200 and 1400 meters, below the thermocline, the temperature and salinity effects become constant and only the increasing water pressure dominates the sound speed profiles shown in Figure 7. This data is then averaged (Figure 8) to calculate a cold and warm SSP for use in the model to determine the effects on the experiment's arrival paths.

MATLAB modeling
The Bellhop program was used to calculate the acoustic paths (Porter M. , 2022). The Bellhop program uses a beam tracing model to predict the acoustic propagation. The 2D version of this modeling program uses bathymetry and sound speed environment files to generate Transmission Loss (TL) plots, ray traces, or time front plots based on the inputs provided. As discussed below, MATLAB programs were written to make the Bellhop input file setup and output file analysis easier.
Ray angle processing was used to plot rays between the source and the proposed Any computer with MATLAB and the Bellhop program installed will be able to run this program. This will allow the deployed team to use current sound speed data on site during the experiment. This provides the team an opportunity to verify placement of the SUS charges when conducting the testing.
One limitation of code is the assumption that the bottom is flat perpendicular to the 2D bathymetry. This limits the model to 2-dimensional propagation. This assumption is mostly valid for the location of the selected bathymetry trace used in the model. However, 3-dimensional effects will most likely be seen as the experiment moves around the seamount and the source changes relative to the bottom topography.

Chapter 4. Ray Tracing Results
Ray modeling ran over ranges from 10 km to 96 km calculates that the optimal arrival angle at the receiver on the summit is around 15km. The 10, 15, and 20 km ranges show the effect of the SSP on the arrivals. This is around the radius that the experiment will use to circle the seamount with SUS charges. Figures 9 to 20 compare the cold and warm transmission losses and ray paths for 10km, 15km, and 20km ranges.
We can see in Figure 9 and Figure 11 how the cold water sound speed profile pushes the sound rays down quicker than the sound rays shown in Figure 10 and Figure 12 with the warm water sound speed profiles. With the source at 15 km, this will maximize the incident angle for the reflected wave. However, since the sound rays remain higher in the water column with the warm water sound speed profiles, the Acoustic Telescope (AT) could be saturated by direct path at 10km for a warm SSP.
The next group of figures show what will happen when the source is moved to 15km.
The transmission loss plot in Figure 13 and Figure 15 shows the cold water sound propagation. This range will still provide incident angles above the critical angle which will allow for the sound wave to reflect off the bottom at the OBX location and that wave will be detected by the AT. The warm water sound speed propagation is shown in Figure   14 and Figure 16. This shows that the direct path arrival will be refracted deep enough to be below the AT allowing for a cleaner reception of the wave front reflecting off the seamount.
Therefore, the sources should be at least 15km away to reduce the sound level at the AT. However, if the cold SSP exists, the 15km range may cause the OBX below the AT at the summit to not receive the sound from the source.           For longer ranges, the rays will reflect off the slope before bouncing off the surface and returning to the top of the seamount. The reflection is only modeled in a 2D plane. The reflected ray will also incorporate a 3 rd dimension due to the grazing angle of the slope.
For the paths modeled, the sources to the north will travel mostly parallel to the slope of the seamount. This reduces the error for the selected models.
As we continue out to ranges past the first convergence, we find that the sound rays are refracted downward into the seamount again. The range that this occurs depends significantly on the sound speed profile. Results of the model are next.
In Figure 21 and Figure 23 the cold water sound speed profile results are plotted. The range of 75km provides good acoustic arrival angles. Due to the ray path dependence on the sound speed profile, Figure 22 and Figure 24 show that the warm water profiles will pass over the seamount and will not provide reflected waves from the seamount.
The modeling shows that the warm water sound speed profile will produce long range arrival paths around 96km. Figure 25 and Figure 27 show the cold water paths for comparison with the warm water paths in Figure 26 and Figure 28. The arrival path for the incident ray will be similar to the short-range arrival angle as seen with the blue ray in Figure 28. This is due to the convergence peaking at a similar depth as the original source.          Returning to the 15km plots, we can place a receiver at our desired depth to find the expected incident angle. Both SSPs provide a similar angle of 30 degrees above the level plane as shown in Figure 29 and Figure 30. The final incident angle will also depend on the slope of the seamount where the OBX is place.

Chapter 5. Conclusions and Recommendations
The model implemented utilizing bellhop allows for the prediction of the sound ray paths.
Historical sound speed data from the seamount demonstrates the value of measuring the sound speed on site to predict the ray paths.
The SSP at the time of the experiment will have a significant effect on the sound travel paths from the source. The NOAA NCOM data should be used prior to departure for the site to provide up to date prediction. Then the SSP data measured onsite can be used in the model to verify source plan. Recorded data will then be used for further research.
The current models support an OBX at the summit with the AT moored above. The other OBXs should be placed on the slope 9.5km from the summit at a depth around 2850m.
This placement allows for some uncertainty of the bathymetry without placing the OBXs below their operating depth. The arrival angle will be above the critical angle which will allow for transmission to the seafloor and the reflected wave will be measured by the AT over the summit of the seamount.