Assessing Sediment Heterogeneity on Continental Shelves and Slopes

As anyone in ocean acoustics will tell you, sound is essential for much of the work done underwater, whether that be navigation, sensing objects, or passively monitoring wildlife. While much research has been done about sandy ocean floors, scientists still have a lot to learn about muddier regions with mixtures of different types of sediment. In this episode, we talk with two editors and a researcher involved with the recent JASA Special Issue on Assessing Sediment Heterogeneity on Continental Shelves and Slopes: Preston Wilson (University of Texas at Austin), David Knobles (Platt Institute of Nuclear Physics and Cosmology), and Kyle Becker (University of Washington).

Read all the articles from the special issue here!

Read more from The Journal of the Acoustical Society of America (JASA).
Learn more about Acoustical Society of America Publications.

Music Credit: Min 2019 by minwbu from Pixabay. 

Transcript

ASA Publications (00:26)

Today we're discussing another JASA and JASA Express Letters joint special issue, this time on Assessing Sediment Heterogeneity on Continental Shelves and Slopes. With me are two of the guest editors for the special issue, David Knobles and Preston Wilson, as well as another person who was very involved in the research discussed in the issue, Kyle Becker. Thank you all for taking the time to speak with me today. How are you?

David Knobles (00:46)

Fine, thank you.

Kyle Becker (00:47)

Yeah, doing really great. Thanks for inviting us.

Preston (00:50)

Yes, indeed. Doing well. Thank you.

ASA Publications (00:53)

First, tell us a bit about your research backgrounds.

Kyle Becker (00:55)

I'm Kyle Becker. Boy, how did get into this? I started as an undergrad and I got interested in acoustics while studying mechanical engineering. And from there I decided, well, let's learn a little bit more. So I applied to grad school in acoustics. I found a project on rough surface scattering. It turned out that was my introduction to underwater acoustics, which I had known nothing about prior to that. Got so interested, learned about the, you know, learning about the literature, what this all involved and talking to some of my colleagues that I went and got a PhD in applied ocean sciences. And. I’ve been doing seabed acoustics ever since, so that's like 30 years or something like that.

David Knobles (01:32)

Hi, this is David Knobles. Roughly, I got into underwater acoustics in 1980. And from roughly from 1980-86, I did things like bottom-loss analysis, modeling the seabed interaction with ray theory, normal modes, stuff like that. And then I some additional time to do my PhD in postdoc in theoretical nuclear physics. And then I came back to underwater acoustics in ‘92. And from 1992 to up until about two years ago, I've been focusing on underwater acoustic analysis, seabed physics, under the direction of ONR. And then recently I decided to go back into nuclear physics and cosmology. Yeah, so those are the… roughly my background is in this.

Preston (02:22)

Well, this is Preston and I'm presently a faculty member in the Department of Mechanical Engineering at University of Texas at Austin. And I originally got into acoustics, much like Kyle said, as an undergrad. I took an undergrad elective in acoustics, and I really enjoyed it. And then when it came time to go to grad school, I didn't really know that acoustics could be done in grad school, but I found out about a graduate assistantship, which was able to help me pay for school. I said, “Oh, this is great. I'll do that.” I had no idea what it was. And it was underwater acoustics. And I ended up being able to travel and go to sea. And kind of the romance of the ocean got ahold of me. And I've also been doing it ever since. So I started doing things with ocean acoustics on the seabed for my masters, and now 30-something years later, I'm still doing it.

ASA Publications (03:15)

So what motivated you to do the special issue?

Kyle Becker (03:17)

You know it should be pointed out that this is actually the third special issue on this on this topic related to seabed acoustics. And the actual genesis of the project actually goes back to, like, 2009, where we were invited to Washington DC to talk with some of the sponsors about the future of underwater acoustics look like?

And they sought  our advice as scientists working the field of what hasn't been studied. And the thing that we came up with then, among other things, was that we've done shallow water acoustics. And shallow water in acoustics is defined not by a particular depth, but by areas of the ocean where sound is interacting with the seabed. And historically they've studied areas that are characterized mostly by sands, which has a particular acoustic characteristic. They've not studied what are considered soft materials, like muds or clays or fine silts, so that was a gap. And so we wanted to find areas in the ocean that were characterized by that. And that's kind of where this started back in 2009 with us and some other colleagues. 

And then the sponsors actually started working in this area about six years later. There's a whole bunch of things that happened to the government. There was sequestration. Things were shut down. It slowed things down. And it took a tremendous amount of resources to get data collected at sea. And so around 2014, 2015, I had actually switched and now I was actually working for the sponsor and instead of being a scientist participating I was actually working with David and Preston as chief scientists to make these experiments now happen after like a five-year delay. And, yeah, so between 2015 and 2022 I think we ended up going out to sea four or five times, with, you know, each time we took either from one vessel, one research vessel, up to four at a time, just to give you a sense of the scope of this experiment And we did it in different seasons. We did it in winter time, where the ocean is very different than in summertime. And we wanted to basically go back and revisit the same area, which was mostly off of New England, south of Martha's Vineyard in Massachusetts, to see if we get consistent results over the different seasons. And also we kept coming up with new questions to ask, and we needed new measurements.

Preston (05:43)

I can add a few things. It's kind of like any complex topic. Once you start studying it, you realize there are more things you don't know than you originally thought. So like what Kyle said in the first round of experiments, we kind of had a very simplified geometry, a flat bottom, as Kyle referred to the winter season, which leaves the water column very well mixed, and so it's relatively homogenous. So that was kind of the easiest case, and we wanted to go back and try things a little more complicated, where we would come back in a different season, the water column would be more complicated, and then eventually even further down the slope to more deep water and different kinds of sediment. So we were kind of adding complexity as we went to keep testing ourselves and our ability to do the work.

David Knobles (06:29)

One of the things that came up in this experiment, the preceding experiment, there was some hints that a better experiment in the future would be to try to make what physicists call coincidence experiments. And that is where you make several measurements simultaneously to try to understand what's going on, say, inside of a nucleus or something like that. Similarly, the idea was to make simultaneous measurements of the geology, say, with core experiments, and then correlate them to what we're seeing with the acoustics in our models. And in that way, that will put a very strong constraint on all these inversion models and techniques that you'll hear about or read about. And so that’s one of the things that we did in the design of the experiment. And I attribute a lot of that success to Kyle bringing in geological survey groups to do all this heavy coring. That way that started to provide us with a base model. And so that was one of the objectives going into this experiment, was to really try to put in this link between the geology and the acoustics in a way that's never been attempted before in any of these experiments.

Kyle Becker (07:49)

Yeah, if I may add to that, you know, you said a couple of really important things, is that as we kept going back, we had different and better questions. And we also realized that we needed different expertise, and so we went and got them. And it wasn't just geology and geophysics. We also found as we were taking direct samples of the seabed that every sample came up with some impact or indication that there was biology. Sometimes there were actual critters in the samples. And so we actually started engaging with biologists as well. It was a truly interdisciplinary team that we assembled and you'll see that in all the papers over the course of the experiment. And we also got perspectives from people not only nationally but internationally. It was a truly international experiment in the end.

David Knobles (08:32)

I want to give more of a physics perspective as to what motivated me and that was the issue of causality. It's a principle that transcends all of physics, and basically it says that the frequency response, of say, some material, such as the sediment, that the real part and the imaginary part of the speed of sound has to obey what's called a Cramer's-Crony dispersion relation. So going into this experiment, and what that translates into in the previous experiment, Shallow Water ‘06, they had some very nice dispersion relationships for sound and attenuations for a sandy environment. But for a mud-like environment, we had no type of dispersion-type of relationships. And so that was one of the things we wanted to go in and try to get this, so that we could then compare models that take into account causality, such as Nick Chotiros’s Biot/Squirt model and Mike Buckingham’s Viscous Grain Shearing model, and we could compare the outputs of these models and see if we could fit the group dispersion of the sound speed and attenuation over a large bandwidth. And if you could do that, if you could actually pull that off, then that would say that you actually learned something about that sediment. And so that’s at least my physics perspective as one of the things that we were thinking about from the very beginning. Whether we achieved that or not, you know, I think that's going to be an ongoing question in the future.

ASA Publications (10:11)

So why is it useful to understand sound propagation in sediment?

Preston (10:15)

Well, I can start out maybe with a little bit of a bigger picture, which is why would we want to go out into the ocean and use sound to do anything? Many folks who study acoustics know this, but sound is the only type of radiated energy that propagates to great distance in the ocean. And so we can't use light or radar as we can in terrestrial applications. And so any time you want to do something in the ocean and remote sense anything, navigation, where objects are, where other entities are, you need to use sound. And so when you want to do work in the ocean near the bottom, then the sound that you use is going to interact with the ocean bottom. And so to kind of optimize all the systems we use to do human work near the ocean bottom, we need to understand the acoustics of the sediment. And that ultimately plays into building better tools to do all that work. Better sonar systems, better communication systems, better object detection systems, etc. So that's the kind of the big picture about why we wanted to do this work.

Kyle Becker (11:17)

Yeah, that's excellent, Preston, and that's exactly right. And as humans, we're actually playing catch up to the marine mammals and the fishes who've been out there working in these waters forever. And they certainly know how to work in these shallow waters much better than we do.

David Knobles (11:32)

Here's one of my perspectives on that, is that in the 1960s and ‘70s, a simple Rayleigh reflection coefficient at the water-sediment interface did a really good job for sand sediments. You also have to keep in mind this is a very strongly frequency-dependent problem. And at the higher frequencies, the shorter wavelengths, they have information about the upper portions of the seabed, whereas the lower frequencies, the longer wavelengths, they have information about the deeper portions of the seabed. Now for softer sediments, which we're talking about here in the mud, we needed models that take into account the depth dependence of the geoacoustic structure of the seabed, because all the energy, most of the energy, penetrates into the water sediment interface and down into the deeper part of the sediment. This is an extremely hard problem. And so that's why researchers started to include propagation in the sediment as opposed to simple reflection coefficients to account for reflection and refraction within the seabed. Again, Ross Chapman from University of Victoria played a very important role in this work in the early part of the work. In Shallow Water ‘06, the sediments, they did have this high content. And our community was really pretty successful in fitting the observed dispersion of sound speed and attenuation with several type of model there. And again, Nick Chotiros and Xiwen Zhou and Mike Buckingham led the charge on this. But again, what about these fine grain sediments? What do these dispersion curves look like? And I would say after 10 years, we've made some progress, especially with the sound speed, but less so with the attenuation. 

And the difficulty here goes back to the fact, again, for these soft sediments, most of the energy goes into the sediment. And so your inversion model has to somehow keep track of these different parts of the energy that goes into the sediment and the nature of the physics like reflection, refraction from the different layers. And so this heterogeneity that is the focus of our special issue here, when you have heterogeneity, how does it affect your ability to go in and find out these different components in the seabed? This is a very difficult problem because of heterogeneity.

Kyle Becker (13:59)

Yeah, hey, Dave, that was great. And one thing I'll add to that, because I think it's important, that I kind of mentioned that, oh, well, these experiments came up when we started talking with the sponsors and telling them what's next in 2009. But the truth is that a really challenging topic, and, you know, we’re not the first people to study it. These types of studies have been going on since before the Naval Research Lab was started in 1923. You can find some of these original formulations about sound speed in the sediment going back to 1930 with A.B. Wood. It's just a very, very difficult and complicated topic. And, you know, what's allowed us to do and make progress now is that we have better observational and experimental techniques, and we also have fleet ships available to us through UNOLS, the University National Ocean-going Laboratory, where we can get out and get access to the sea to make these measurements, which is really important.

ASA Publications (14:50)

So David got into this a little bit, but why is it important to consider sediment heterogeneity when studying the seabed?

Preston (14:56)

Yeah, I can just kind of add on to what I said before. So if you want work in the ocean over area, you want to have understanding of those acoustic properties all over the areas you're interested in. And in many parts of the ocean, there's a lot of homogeneity, and so properties are kind of constant with range or in different directions. And it might be considered an easier place to work, but unfortunately, places that we're interested in also have a lot of variability. And so, when, you know, you’re on the land, you can look out over the land and see how things change very, very easily. You know, a human can assess how different the ground might be, you know very rapidly and you know what's going on, but it's much more difficult to do that in the ocean bottom, and so we make assumptions that things are similar. Sometimes we want to know how good those assumptions are, and then sometimes we have to physically actually measure everything and take into account all of that heterogeneity. So it's an effort to be able to do a better job of using sonar. We have to be able to characterize and understand that heterogeneity as well.

David Knobles (16:02)

The underwater acousticians, we have models that take into account heterogeneity. For example, the parabolic equation model and what we call range-dependent normal modes. Namely, if you have enough information about the ocean environment from, say, your source to receiver, the parabolic equation model affords you to put in all that information into there. So namely, the speed of sound, the density and the attenuation, those are all dependent, essentially, X, Y, and Z. 

The problem is, like Preston noted, we don't know what those values are to go into these very high-powered propagation models. And we can't easily drill down an acoustic probe or put down a core, in general, and that's why we do this remote sensing to try to estimate what these parameters might actually be, so that we can perhaps go in and have enough information to make a reasonable prediction with some simplified model, with parabolic equation model. 

And most of time what people are trying to do with these models is that we're trying to come up with an effective, a global effective geocoustic profile from the source to receiver. It may not have all the true range-dependence effects and all the depth-dependent effects in there, but we're trying to come up with a global effective model. But as Preston noted, if you drive between some places where the limestone has been cut out on both sides and you see all this layering stuff. Can all that stuff go into a model, the details? Yes, but how do you understand what that layering is? You can't go out and touch it. You can't necessarily go down and drill down to get information. So you get into this business about this idea, remote sensing with some sort of an inversion model. And to me that's what makes heterogeneity, when it's not range independent, that's what makes this an extremely difficult problem. And well, that's why they got us good scientists trying to work on it to try to solve that.

ASA Publications (18:24)

So another aspect of this is that this is research on continental shelves and slopes. So why are you interested in continental shelves and slopes as opposed to other regions of the ocean?

Kyle Becker (18:34)

When I started a little bit earlier. We talked about the definition of shallow water, right? And it's not any particular depth, per se. It's just areas where if you're going to put a sound source, whatever sound source that might be, that could be an animal or it could be a ship or one that we use for our remote sensing, that sound’s going to interact multiple times the seabed and that, you know, in the shallow water, it's going to repeatedly interact with it. As you go to the shelves and slopes, it's going to interact fewer times. But, you know, every time it interacts with the seabed, there's some information imparted from the seabed into that signal, is exactly the information we're exploiting to learn about what the structure is, as David said, underneath not just the surface, but the subsurface of the seabed. And so I think that kind of gets at the root of the question in our interest in that. And one other thing is that historically, there's been an interest in deeper areas of the ocean, what they call blue water, blue water acoustics. In those areas, that's defined primarily by the sound speed of the water column, the sound speed profile of the water column. And that has a particular characteristic in such that it has a sound speed minimum at around 800 to 1,000 meters in depth. And the sound in that minimum actually gets trapped, it’s a sound speed channel and it doesn't interact with the bottom. And so that kind of distinguishes deep water from shallow water. So once you get onto the shelves and the slopes, that transition, that channel where you don't have interactions goes away.

ASA Publications (20:04)

Okay, okay, got it, that makes a lot of sense.

Preston (20:07)

Yeah, I was going to say, there's one other thing worth mentioning about why the continental slopes and shelves and, you know, those are closer to the land, obviously, and that's one of reasons there's more heterogeneity in those areas because things coming off the land, out of the rivers, and the shoreline, the interaction of the ocean bottom with all the sediment types that are coming from the terrestrial surface, that's one of the things that leads heterogeneity. And it's also convenient if you can only have to drive your ship for six hours to get out to your site rather than two or three days. It certainly makes your life a little easier. So I mean, that's kind of a fun little way of thinking about it. But it's true. We happen to have this mud patch 60 miles away from Martha's Vineyard with all the things we were looking for and very close to one of the world's greatest oceanographic institutes. So a lot of convenience in that as well.

David Knobles (21:02)

There’s also an applied knowledge of this, and that has to do with the idea of the effect of large ships, the sound of large ships on marine mammals. And that’s really a hot topic these days, especially in Europe. And the idea is that in deep water, it's really easy to set up an experiment to measure the sound levels from a large, say, container ship because there you don't have to worry about the bottom; you're basically within direct path and other things ,but as soon as you start getting into shallow water, as Kyle and Preston noted, the sound isn't trapped in a channel. Most of the time it’s downward refracting, and between wherever you set up your receiver and the large ship,  the sound is already interacting with the bottom multiple times. So if you wanted to come up with an algorithm to tell how loud is that ship, say, at one meter away from, say, the source, which is some sort of a standard unit, then you're going to have to take into account the interaction of the sound with the seabed, namely to build up a system to, say, it’s kind like a radar gun. You know how police like to hide behind trees and see how fast you're going with the radar gun? The idea is similar here. You have a receiver and you have the algorithm and then potentially you could say, hey, this ship is putting out X number of dBs higher than what our sound limit is, and we know which ship that is and we'll send them a bill for that. Those are actually the type of talks that are going on at least for some of the conferences I’ve been in to Europe, so it's natural that studies from the seabed in the shallow water areas are going to be of interest to that community that is concerned about the sound levels and their effects on marine mammals

Kyle Becker (22:58)

David, as you're talking, you know, I'm reminded that when I went to graduate school, which was at Woods Hole, which Preston mentioned, and I went to sea from there and I took a ship out to the same areas that we went and did seabed or through those, it was open water. When I went out on a ship last fall from Woods Hole to the same area I did my research as a graduate student, we had to navigate through a wind farm, through Vineyard Wind. The amount of offshore development going on right now is phenomenal. And these types of studies help us and them understand what the seabed looks like, not just the sea surface, but into the seabed, whether it's going to be a stable place to be building these wind farms, what the sound characteristics in that environment are going to be now that the infrastructure is in place, owing to not only the platforms, but to the increased amount of shipping around them. So really these studies that we're doing which are basic research go hand in hand with a lot of the applications that are surrounding the development of our coastal waters for whatever reason.

ASA Publications (24:05)

So you guys have sort of alluded to this already, but most of the articles in this issue had to do with the Seabed Characterization Experiment conducted in the New England mud patch from 2015 to now. What were the goals of this study and why was it so special?

Kyle Becker (24:18)

Okay, so I think, it's kind of alluded to before, is that we’d done a lot of work previously in sediments that were characterized as being sandy. And as David alluded to, both people and papers, we had actually a lot of data on sound speed dispersion and attenuation in these regions, which are defined by, we'll call them sand. Those are granular materials where the interaction—the sound vibrates and the results of the dispersion are owing to mechanical interaction between granules and the water in between them. Fine sediments, clays and muds, in addition to having particles scratching against each other, also have electromechanical properties. And they have different behaviors. And we—this is the podcast so you can't see the plot. But it's amazing how they're separable when you look at these dispersion curves when you start to have data. And prior to the experiment, we had almost no data on what that looked like. And so really, in simplest terms, we wanted to get data on these types of sediments which hadn't been studied in this sense prior to these sets of experiments. I think that's the simplest way to put it.

Preston (25:31)

I’ve got four items here that we originally set out to try to do, our scientific goals. And so the first one is to understand the physical mechanisms that control acoustic propagation in fine-grain sediments. And this goes directly to what Kyle was talking about, how the grains interact with one another and how the acoustic waves affect that. And then the second one is to quantify uncertainties in our ability to estimate seabed parameters using remote sensing. So it's not a perfect way of understanding the seabed, but we can do it remotely. It's convenient. And so how much uncertainty we have and can we live with that? So that was the second. The third one is to correlate observed variations in the acoustic field and  in the water column with measured variability of the seabed. So that gets directly to the heterogeneity. So understanding that heterogeneity has been part of this since the very beginning. And then the last one is to assess the ability of these remote sensing techniques to create geoacoustic models and to do statistical inferences and ultimately understand the acoustic properties of the seabed are to be able to assess those techniques. So those are the four that we've carried this entire time.

ASA Publications (26:48)

So a project of this scope required a number of collaborators, both domestic and international. Can you tell us a bit more about the organizations and individuals involved?

Preston (26:58)

Going back to the beginning when Kyle was talking about the beginning of the project, we were working with different sponsors in the Navy and elsewhere. And then we started doing this through basically the University of Texas, where David and I have been working. And then we began getting collaborators in other universities all over the country and then even internationally. And so I don't have an exact count of the total number of people, but it's in the hundreds. And in addition to those people who are responsible for the science, we have all the people that help make the science possible. So the UNOLS team that operates the ships, the ships’ captains and crews, and then each of these researchers back at their home institutions, they have people supporting them, technicians or engineers building equipment. And we have people who help us get the gear into the ocean. And we would work with oceanographers to help us understand the oceanography. And then we would work with local communities and their fishing fleets to try to minimize our impact on those. And so it's kind of an amazingly large amount of people that have gone into this, hundreds of people over a decade.

Kyle Becker (28:14)

Yeah, I think I think something that you we might point to is that David mentioned some of the models, and there's a laboratory over in Italy. It's now called the Center for Maritime Research and Experimentation. It's hosted by NATO. That goes back 75 years, that lab, and a lot of the models that were developed for underwater acoustics were actually developed in that laboratory and are still used today with some of the people, and so some of you know… There's a history of international collaboration in this space, and we're just kind of continuing that. We had NATO partner as part of this experiment.

David Knobles (28:50)

Another thing that comes into play is what Preston was talking about,  you know, support from these different universities and labs. It's not only the principal investigators, but it's also all the graduate students and post-docs that they have working for them. And it's basically, it's a training ground for our next generation of ocean scientists that are gonna be involved in this area here. The field is very rapidly developing; for example, in the area of machine learning, we weren’t even talking about machine learning when we were coming up with these grand plans 20 years ago that Kyle was talking about. It wasn't even on the radar, but now it is very much on the radar about these next generation mathematicians trying to figure out how machine learning can be used to do this type of remote sensing. 

And I just want to point out a couple of places in seabed characterization that's doing that. One is the University of Delaware. Specifically, there are two graduate-- or I guess they're postdocs by now-- working under Mohsen Badiey, John Castro and Christian Escobar Amado have come up with some pretty revolutionary papers in machine learning, and also there's a program at Brigham Young University headed by Tracy Nielsen, with all of her students that is also looking at new ways to put in this machine learning. So, again, my point is when you do a large scientific experiment like this, and you have all your grand goals that we’ve talked about, it's the unexpected things and the new paths that start taking place. And very often those are connected with this new generation of scientists that are being trained in these types of experiments.

ASA Publications (30:40)

Yay, graduate students!

Preston (30:41)

Here, here. Yes, thank you, graduate students. They do the heavy lifting, quite honestly.

Kyle Becker (30:46)

They do 

ASA Publications (30:47)

Yeah, right.

Kyle Becker (30:48)

And that’s where we all started, right? So continuing that is so important for the field.

ASA Publications (30:53)

The circle of the research life. So as we've been discussing, the fundamental thing we want to know is how to remotely sense and figure out what the properties of the ocean bottom are. And acoustic sensing is a convenient way to do that. How do direct measurements, simulations, and inference play into achieving this role?

Preston (31:10)

Well, I can start out maybe with a big picture way to think of it. So, you know, we have this physical system, the seabed, and we want to eventually do remote sensing. We want to be able to assess whether our remote sensing is accurate. And so then that would require you to do what we call direct sensing, or in other words, maybe use the words ground truth. So need to go down and actually measure the sediment sound speed directly as best we can. Once we're there, we want to remove some of that sediment, bring it back to the lab, so we can understand all the things that are in it. And ultimately, as we've also mentioned, we want models that kind of relate the physical characteristic of the sediment, whether it's sand or mud, and whether it's got infauna, which are the critters that live in it. We want to relate that back to the acoustic properties. So what I'm kind of imagining here is kind of a circle. And the circle if you think of like 12 noon, there may be an acoustic model of the sediment, and then that leads to of a direct measurement of that sediment sound speed, and then that might relate to doing an inference using acoustics. And then we might have an inference procedure, right? And so all these things are kind of dependent on one another, and they really all need to be self-consistent. And if we can come up with a self-consistent system with forward models and inference procedures and direct measurements, and if they all kind of tend to agree with each other and are self-consistent, then we can convince ourselves we understand this system, and then we can go out and use it to do this remote sensing and be confident that we're getting the right answer. So it's kind of a big picture way of thinking of all those things.

David Knobles (32:49)

I wanted to add a mathematical perspective at this time. And that is, the idea is we go out and do remote sensing and we try to get more and more information about the seabed. In principle, we take that information, all this additional complexity, and  we try to put it into our PE models. But here's the mathematical issue that we're really up against in that. And that is in machine learning, it turns out that as we increase the complexity in the inversion models, guess what happens to our ability to make a prediction? It goes down. So, namely, we can fit the data with more and more complexity, but then our predictions go down the toilet there. 

And it turns out this is not just an idea, it's actually a fundamental law. And our group really has not come to grip with this. And namely, how much remote sensing, how much information do we need to get for our models? So this thing going again back to mud, when we go to the mud, to the shelf break, and then we go down to slopes, the seabed actually undergoes large-scale changes. And we've made a lot of progress, but we are far from characterizing these changes that, say, could go into a parabolic equation model for long range prediction with a, what I would call a tractable number of parameters. That’s something that we are going to be faced with and trying to make progress on for a long time in the future.

Kyle Becker (34:28)

Yeah, and I think if we go back and revisit some of the things we said earlier. So, you know,  high-frequency sound doesn't penetrate as deeply into the sediments as low-frequency sound, alright? And so depending on what you're modeling and what you're trying to predict, you need more or less information. And depending on what tools you bring to your remote sensing or inference problem, you know, every observation system has a type of lens. And it'll allow you to see or get different information. And, you know, we want to try to understand how we observe the system impacts what our inference is. And then so if we're looking at making measurements at a high frequency versus a low frequency, we want to kind of connect them together so that, in a sense, they all represent the same seabed. Hopefully I'm making sense. But I can tell you now that there are some databases out there that represent the seabed for different models, but they’re completely disconnected from one another, even though they may claim to represent the same geographic location in space. If you're going to work with a high-frequency prediction model, you might pull up completely different sediment properties for a low-frequency model. And we're trying to bring these things closer together.

ASA Publications (35:39)

Yeah, that does sound a little problematic, to say the least.

Kyle Becker (35:42)

Yeah, but that's the nature of it and you know these inference problems are typically non-unique, right, so you can get an answer that matches the data or the prediction, but not across all frequencies in many cases. Andwe're trying to, you know, in this self consistency thing, make that a little bit better.

David Knobles (35:58)

Back in the early days of geocoustic inversion, there were a tremendous number of papers that were written in JASA. And IEEE JOE, basically saying, hey, I have this piece of data. I used the geocoustic inversion model. I did inversion. And I look at the fit that I got between the model and data. It's fabulous. The problem is, like Kyle was saying, you really have no idea whether or not the optimal solution has anything to do with reality. But we've made a lot of progress in trying to answer that question, but that is one of the fundamental issues that we have about remote sensing.

Preston (36:40)

Yeah, I've got to bring up this analogy, which I learned from my former PhD advisor Bill Carey, which is very useful here. It's an astronomy or a cosmology analogy. And you may recall that Ptolemy had a geocentric universe as a model. And then, after that, people like Copernicus and others had a heliocentric model. Each of the times when they were working, they had data from astronomers that matched their model. And so, from that point of view, the measurements and the model were in agreement. So everything was great, except for, you know, of course, one of them was totally wrong. So as Kyle mentioned, if you want to now look at it in a slightly different way, that measurement, that model will no longer agree, and it's not going to be a helpful remote sensing tool. It's very important to have all of these different ways of looking at this sediment, from direct measurements to models to inferences, and make them all self-consistent and useful.

ASA Publications (37:36)

So let's get into some of the actual research within the issue. There were a lot of different measurement and analysis techniques covered in this issue. Let’s start with some of the direct measurement techniques. What studies use these sort of measurements and what did they find?

Preston (37:52)

So, my group, which led by Megan Ballard and included colleagues of mine, Kevin Lee and Andrew McNeese and a number of our students, led some of the direct measurements where we were inserting acoustic probes into the sediment bottom. So all the way down on the bottom, sitting on the bottom with instrumentation that's connected back to the ship with a cable. And it's pretty simple measurement. You insert a sound source and then you also insert a couple of receivers. You send out sound, and then you can see the time of flight difference between the two receivers. And then we would take the distance between the receivers and calculate the sound speed. So really, know, kind of freshman, high school physics kind of stuff. So directly measuring the sound speed is one of the things that we did. 

And then we would also do that at the end of a coring device, which would allow us to do that from the surface down much deeper into the sediment, say on the order of a few meters up to maybe three or four meters at most. We would also try to measure the attenuation, which is a little bit more of a difficult problem because there are more systematic errors that you have to unravel and account for. But sound speed attenuation, for the compressional wave, which is kind of the standard way we think of acoustics. You would hear between you and me talking in a room. That's a compressional wave. But we also try to do all the exact same things I just mentioned for what's called the shear wave, which is what you might imagine with a slinky, kind of doing a transverse wave with a slinky. The ocean bottom can support shear waves, and they also have a speed, and they also have an attenuation. So we would measure that as well.

And then the last thing I'll mention is when we remove those cores, we measure all kinds of things about the physical nature of the sediment, the kinds of grains, the grain size, and as Kyle and David both mentioned, the infauna that live there. So we're characterizing all of those things.

David Knobles (39:47)

There’s something that did come up in the direct measurements that I think is a little bit unique to the mud patch. That is that even historically we knew that the mud isn't simply a uniform glob of a sediment. Specifically down towards where the mud is over a sand, but that interface is not a pure mud-sand interface. It's a gradual interface, and what happens is that the mud gradually transforms to a sand. So there's this region of about two or three meters above the sand interface where you have this mixture of mud and sand, so folks like, as Preston was saying, that his research group that included Megan Ballard, they were able to get their acoustic probes down far enough and saw that you can actually map this transition. And it's not a uniform type of, namely, it's not always three meters, it's not always two meters. There's heterogeneity associated with that, what we call this gradient layer between, oh, somewhere around six or seven meters and then on down to the sand around ten meters.  And John Goff has now published a very interesting paper about the geological process that led to this deep gradient inside of mud. And one of the very interesting questions that I think are going to come up is that there are other mud patches in the world like in the central yellow sea and other places. And the question is, do they also have this same gradient or a similar type of gradient? And those types of questions at this point are unknown. But I think that that's one of the most interesting things that came out of the direct measurements that Preston talked about.

Kyle Becker (41:38)

Yeah, so that's right. So the direct measurement, you mostly talked about the acoustic measurements, which you had these acoustic instruments on a core barrel, actually. The core barrel itself, when you extracted the core, that was full of the actual physical sediment material, which could also be analyzed as Preston talked about. And so you had this consistency check, right? Does the physical material look like what we saw acoustically? And then the other thing I think we should really kind of highlight with these soft sediments is they have a unique property in that at specific angles, called the angle of intromission, they're perfectly transparent to sound, and so you know when we talk about remote sensing, we're making all the measurements in the water column; they're not actually going into the sediment, right, but yet we want to see into the sediment through them to this transition layer, and these direct measurements give us evidence that what we're doing when we're remote sensing is actually, you know, we're seeing through that layer down to the gradients and even deeper. And so you bring all these things together to build up your evidence to start to say something meaningful about the volume of the seabed  that you’re studying.

David Knobles (42:39)

And just to follow on a little bit about what Kyle said, this angle of intermission, those main measurements were made by Charles Holland when he was at Penn State, when we were out there, I guess, in the first experiment, 2017. And it's not quite a direct measurement, but it’s the next step away from, say, the acoustic core that was in there. And he's right. One of the things that we definitely found out, and we kinda suspected this going in, is that the speed of sound in the sediment actually decreases relative to the water, and that makes this angle of intermission. Whereas for a sandy environment where the speed of sound is greater than that of the water, there's a critical angle effect. And those two different types of sediment causes a big, big difference in the nature of the propagation. And so I think we've learned a whole lot about the heterogeneity of that critical angle effect. And I think we're set to learn more about as analyses comes in. But that’s part of what I'm going to consider almost like ground truth type of measurements, are these semi-direct measurements that were made by folks like Charles Holland, who's now at, let’s see, Portland State University.

Kyle Becker (43:54)

Yeah, and if there are any students out there listening and you’re thinking about the classroom stuff, I remember taking ocean and seabed acoustics, and you learn the analytical formulas for these things, critical angle and angle of intermission. To go out on a ship and come back and do the data analysis and see these things for real, it's really powerful and meaningful.

David Knobles (44:14)

Yeah, students don't miss class that day where your professor is talking about reflection coefficients.

Preston (44:19)

Yeah, let me add one piece about that, which is somewhat related to the heterogeneity part.

With the fine-grained sediments, a relatively small percent change in sound speed can make you go from having an angle of intermission to not having one. So that's one of the reasons fine-grained sediments are more difficult and why our work is more important. We need to have higher kind of precision in our ability to infer these properties because very small changes in the properties can have big changes in the acoustic effects. So that makes it even more difficult and important.

David Knobles (44:55)

And these measurements are very difficult to make, I would add. In the early days when they were doing these direct measurements, basically they were coming back with an iso-velocity sediment, all the way past the mud. They were not seeing these gradients. But they kept refining their technique with the signal processing and sampling rate. And that's when they, in not the first special issue, starting with the second special issue maybe, and this last one, that's when they started getting enough resolution to where we started seeing these finer effects that the mud is not a purely uniform slab of mud down there. It's complicated.

ASA Publications (45:36)

It's a variety of muds and such.

Preston (45:39)

Yeah, one of the things you can think about mud is mud is everything. You can have sand and silt and clay in various different percentages of each component. And then it can have all the things that live there and then all the things that the life does, digging holes, pumping fluid, going to the bathroom, eating, all of that in that upper layer of the sediment. So it's one reason why it's so complicated and dynamic. It's a living entity. And give Kyle credit for being one of the people to realize that we need the biologists, and we need the geologists here to help us understand that the acousticians alone need those other areas of expertise to study this as one system. And that's really one of the things Kyle brought to this is, it's not just a biological system by itself. It's not just a geological system by itself. It's just not an acoustic material by itself. They're all together, they’re all impacting one another, and we need the expertise from all those areas to understand it.

David Knobles (46:36)

And I can attest that it has lots of living organisms in there. I was on the ship with Jason Chaytor when we were bringing up the cores for the mud patch. And I can assure you, it smells horrible. I mean, if you've never smelled mud coming straight out of the seabed, it took me a month to get that smell out of my nose and face. So it definitely has lots of living organisms and all the things that come out of them there in the mud.

ASA Publications (47:04)

I don't think you've sold me on that experience. I think I'm gonna avoid smelling sea mud now.

Preston (47:09)

There's a kind of a fun little anecdote about bringing together all these scientists. So if you ask a physicist, what is density, they'll say, “Oh, it's mass per unit volume.” So we think of that, you know, in acoustics, density, the mass per unit volume is a very important parameter. But then if you talk to a biologist, they'll say density is the number of critters per unit area of seabed. And so, you know, part of the challenge of initially bringing all these people together is we all speak a different language, a different scientific language. And so we had to start learning how to communicate with each other and to know how to say things in a way that each other would understand. So that's been an enjoyable, although sometimes challenging, aspect of this work.

Kyle Becker (47:50)

Absolutely. And you know just staying on the critters, right? So, okay, we know they're there, but what are they doing? Well, it turns out some are building mounds as they excavate, some are bioturbating and just making holes, some are changing the surface by some kind of mucous-y material, and, you know, these different species are kind of studied at the individual level, but how do they contribute kind of in a bulk way to the sediments? I mean, all these questions that when we first started this, we had no appreciation for. So it's just been really amazing how this evolved over the almost decade that we've been studying it.

ASA Publications (48:30)

So some other studies used sub-bottom layering inversions from noise or low-frequency sound transmissions. What did this entail, and how is it useful in understanding the seabed properties?

Preston (48:41)

Maybe I could start with a big picture, and I bet David or Kyle could give a more detailed, but as you know, we mentioned, we had to go out, and ships with sound sources and receivers cost money and requires time and effort. But the idea of a passing ship to be a source of opportunity allows one to do the remote sensing with less gear that you have to bring. So our experiment was located in between two shipping lanes going into and out of New York City. So these big container ships were going by. And it turns out, yeah, you can learn a lot by using the sound from those ships to do these acoustic inferences, to do the remote sensing. In some cases, you may not need to bring your own source. So I can start with that, and if someone wants to give more detail.

David Knobles (49:24)

Yeah, so, when we were planning this experiment, and we found out that the mud basically was in between these two major shipping lanes, there were a number of folks that were not very happy about this because, well, gee, all that noise is going to affect my ability to do inversions to find them. But a number of us saw this as an opportunity about the idea of using ship noise to estimate what parameters are. 

I want to go back one step. Some of the founding fathers, or founding people, that use the idea of noise and seabed acoustics, I have to mention Peter Gerstoft and Martin Sidereus, who have made their careers out of the idea of using noise and different types of noise sources, to give us an idea about properties of the seabed. I do want to also go back, in my own case, I used to work at ARLUT, and that's where I learned about the use of ship noise. I had a very close colleague and friend; his name was bob cook, and he really was a pioneer of the use of ship noise, not wind noise but ship noise, to estimate parameters of the sea bed. He was that type of person that would stay buried in his office, didn't want to get out in front of people and take credit for stuff. And he would teach other people stuff, like me, to go out and say, “Hey, look at this!” and write papers on it. So this business about why the ships were very important is that if you want to get low frequencies, the lower frequencies require a larger and larger and larger source. And at some point, it doesn't make sense to put in a source in the water the size of a house in order to generate low frequencies. And so that's why these ships are very important here. And there were some major discoveries with these low frequencies. And Peter Dahl and David Dall’Osto wrote two papers about low-frequency noise from ships and about the relationships between the noise and the deep sediment layers below the mud. Okay, these I think are very interesting papers. There is a graduate student from Tracy Neilsen, her name is Alexandria McDaniels, okay, and she expanded on a previous discovery from some 2017 data there in the mud patch that at the very, very low frequencies, very interesting phenomena happens. And that's where two modes can end up having the same group velocity at certain frequencies. And it leaves behind a very distinctive signature in the spectrogram. And all of a sudden, this generated a lot of interest. And it was part of a design of the seabed characterization experiment in 2022, and we were actually quite successful in seeing this particular low-frequency signature on three different arrays that we had out there: the ARLUT array along with the two vertical line arrays that Bill Hodgkiss deployed. This is the beginning. This work is far from completed. And this is one of these unexpected things that happens in a very large experiment. And again, it's made possible by this business with the ship noise.

Kyle Becker (52:57)

I think as that gets developed, right, a lot of folks that are out there around the world that are doing passive acoustic monitoring for various reasons. For when they're putting in new infrastructure, or they're doing it monitor fish or mammal migrations, all kinds of things like that. Ships are going to be driving by. So, you know, as we develop this and mature it, we could transition it to those types of systems, and we can then use those data that they're collecting to learn about the seabeds in those areas as well. We haven't even explored that yet.

David Knobles (53:28)

Yeah, one last thing here, and I told some students and stuff, is that one of the greatest discoveries in all of underwater acoustics was Ewing's discovery of the SOFAR channel. That's the channel that Kyle talked about in deep water that sound gets trapped. And in shallow water, in this channel, due to deep sediments that trap the low-frequency sound,

it's really down there and it exists it and if they can travel to very very far ranges, this could also expand this whole concept of channels environments in ocean environments. I think it could potentially be a real big thing.

ASA Publications (54:07)

Yeah, yeah. So some other studies considered inferences from solutions to inverse problems. What kind of research questions were covered in these studies, and what did they ultimately reveal?

Kyle Becker (54:18)

Yeah, I can say one thing because David, you mentioned machine learning earlier, right? And people like Stan Dasso and Jan Dettmer and Mei Zhang have been looking at these inference problems using active sources for a long time, Bayesian inference and whatnot. And although we haven't talked about them, they didn't talk about them as machine learning approaches. They really are, right? And so we've been doing machine learning for a long time in our community, and so we should get some credit, although we haven't been. Just want to get that out there.

David Knobles (54:46)

Let me put this question a little bit, at least in my opinion, in historical perspective. Back in the 1970s, there was this concept of what's called match field processing. In that way to localize sources. And it was a big thing. The problem with match field processing, it was susceptible to errors that you made in the assumption about what, for example, your ocean environment was. You got all these things that happened to where you would end up with multiple sources that were out there. And it just acts, it didn't work at the end of the day, although there must've been five hundred to a thousand papers in JASA and IEEE published about this stuff here, so people like Stan Dosso, Jan Dettmer, who you mentioned, we started to think about the idea that geoinversion it would be the next step above matchfield processing, and basically the idea was, hey, instead of just trying to find out the source location, what if we put in the sea bed as one of the parameters that we're looking for, and that we invert simultaneously for the sea bed and localization and where the source actually is, and that's where a lot of the models that have been developed in past 20 years, that's what we're using in some ways to analyze data for seabed characterization. Namely, it's this business that we started off with matched field processing. We realized that you couldn't do just matched field by itself. You had to include the seabed as a parameter, or parameters, in order to solve the bigger problem. And as I pointed out before, as the seabed becomes more complex, then that more complexity, you can fit the localization problem better and better, but its ability to make predictions for data that you haven't analyzed starts to decrease. And what we've learned, one of the things that we've learned, is that you have to build in a protection device in order to keep this nasty aspect of inversion from biting you. And like Stan and Jan Dettmer and others know a whole lot about this business about how to build in these protection devices into the mathematics. And again, it’s all strongly related to machine learning and all the mathematics behind that.

ASA Publications (57:23)

So there were a few other types of studies that came up that didn't fit into these categories that we just discussed. What techniques and/or findings stood out among those?

David Knobles (57:31)

This to me is kind of a continuation of what we just said. Something that came out that's new that didn't quite fit into the paradigm of some of the models that we had going in is this business with machine learning. And some of the big people behind that in our group are Potty, Nielsen, Castro, Escobar Amato, and it may be our best hope to try and predict seabed classification when the waveguide actually possesses large-scale heterogeneity. So again, I repeat, it was really great seeing the younger generation, instead of us old fogies, coming up with ideas, it was a younger generation that took the bull by the horns to try to tackle this very complex problem because they are the future of ocean acoustics. All the people I've mentioned just recently, they're all under 40. Whereas people like Preston and me, well I'm not going to tell you how old I am, but you know.

Kyle Becker (58:33)

You know, to address this, I don't think this is necessarily explicit in this special issue or any of the other special issues, but I think something that was unique about this set of experiments in general is the fact that most observational oceanography and acoustic work, which is primarily measurement-based, we get an opportunity to go to a location and make some measurements and then report on it.

And here we're studying the seabed, right? Ostensibly the seabed is a fixed, we're talking about as a dynamic system, but in, you know, relative time scales, it's not changing from day to day. It's going to be relatively consistent from year to year with slow changes. But as we already mentioned, we went out in the winter time where the ocean is very, very dynamic and mixed up. And that means an iso velocity water column. And then we went out in the summer where there's heating at the surface and you get something called the stratified sound speed profile; it's a very, very different environment. And by going out for the different seasons and repeating in the same space, we got to see if that assumption of a relatively stable seabed is actually true. And that's important, right? Because we're trying to characterize something in a somewhat lasting matter, right? We talked about building infrastructure, right? Like, so if you went out in your backyard and the soil changed from sand to muds between one day and the next, it be really hard to build a house there. And so it's kind of behind everything that we did over these several years. I don't think it comes out explicitly, but it's something that we were paying attention to. So if we're going to go measure the seabed, are we actually measuring the seabed? Or are we just getting a different result because the conditions were different?

Preston (1:00:09)

Yeah, I want to add to that with a little more finer detail. It was a great segue. So Nick Chotiros has a paper in the special issue where he investigates the effect of the salinity of the pore water on the acoustic properties. And as we've all kind of mentioned, mud is made up of these clay platelets with these electromechanical characteristics. And so if you want to kind of imagine an acoustic material is kind of having some compliance or some springiness, kind of the harder it is, the faster the sound speed, the softer it is, the slower.

With mud, the salinity, which is the electrolyte in this electrochemical system, the salinity changes the way those electrochemical, electromechanical forces are present, and hence it has the ability to change the stiffness. And so, that's something that people have been kind of poking around at for a few years, but in this recent special issue, there's more mounting evidence that this is true. And so, how that impacts through the heterogeneity is now we have to not only worry about the way the ocean is today, but potentially the way it was last week, because if saline water comes in last week and that salinity diffuses down into the sediment, those sound speeds become what they are based on that salinity.

Tomorrow, an ocean front comes in and now the water above the sediment is very different in salinity. Well, it still takes time for that diffusion to occur. So the bottom water is of some particular salinity, but the pore water is of the salinity from last week. And we now know that that plays a role. It's a small effect, but it's something there. So that's one of the new results, you might say, coming out of this special issue.

ASA Publications (1:01:55)

Very interesting. So were there any key takeaways from the special issue you would say?

Kyle Becker (1:02:00)

It's as hard as we thought it would be.

ASA Publications (1:02:04)

Ha, yeah.

David Knobles (1:02:06)

Well, let me throw a few numbers out that I think that the special issue came up with. Stan Dosso wrote a very nice kind of a, it was a short paper, but it's a nice summary paper that tried to put everything together. The shear waves, the shear numbers, the compressional numbers, attenuation. And I go back to this idea that there was a confirmation that the sound speed ratio average over the mud patch was somewhere around 0.984. Again, that's the ratio of the speed in the sediment to the speed in the water. That number seems to be holding up pretty well. And what's interesting to me was that number kind of popped out very early in the analysis with Charles Holland and there were some of his colleagues. Belcourt, I'm trying to remember her first name, but she wrote a paper right around the time that Holland came up with his work, and that number seems to be really be holding up. There was a researcher, her name is Jie Yang over University of Washington, Seattle.

Early on, they deployed this direct measurement technique, this triangle, where they put probes down into the sediment, a little bit similar in scope to Megan Ballard's technique. But the numbers that they came up with, you know, around .99, .9 whatever, at the mid frequencies, those averaged over the entire mudpatch, they had about 15 or 16 sites, those numbers have held up. And I'm really impressed. I've always been really impressed Jie Yang's measurements, or the University of Washington, because to me, they have withstood the test of time. And I think they are about as close to ground truth as you can get, as far as numbers are concerned. I think there's another thing that came out of the… the recent measurements. There’s been a reconciliation between the acoustic cores and the physical cores, because when they first did the measurements back in 2017, the physical cores were showing missing mud, or they were showing  these things that were missing. Whereas the acoustic cores and the measurements by Goff, they didn't show that. And so, like Kyle said, we went back in 2022, they made new core measurements, they made both acoustic cores and physical cores, and that all got reconciled. And they also found the small heterogeneity, east to west heterogeneity of the grain size, or if you like the surface sound speed. That was kind of known going back to Twitchell who wrote a paper about this area before that there was this east-west heterogeneity. But it was really nice seeing that all these measurements and also the acoustic convergence were also seeing a similar heterogeneity there. So I think those are some of the big takeaways. 

Another takeaway is that we still don't have the answer for attenuation dispersion. And so that's been something that's been troubling us. But in fairness, attenuation is the hardest thing to get out of a set of measurements or inferences compared to things like sound speed. Hopefully we will reconcile that as we go forward in these future analyses. But anyway, those are two things that I thought that kind of pops to mind about what were some of the takeaways.

Kyle Becker (1:05:43)

Yeah, you know, just one thing before, you know, we're kind of wrapping up that I think is just really important. Like when we met initially in 2009, it was the Ocean Acoustics Program at the Office of Naval Research and it was, I don’t know, less than half a dozen, but a few good, you know, good prominent scientists in ocean acoustics. And if we had just stayed with that team, we would have done well, right? But we didn't. We expanded the team to a whole bunch of different disciplines. And as a result, we couldn't do one special issue. We ended up doing three special issues. The work's not done. You know? The focus won't be there because the mechanism to keep it going the way it was isn't there. But it really took this team of folks with different perspectives and different expertise to come together to make the progress that we did in this area. And I think that's important. I think it's a model for how big science can be done.

David Knobles (1:06:39)

And in every large group, there's always a bean counter. Okay, so I'm gonna be that bean counter here. Tvhose three special issues that Kyle pointed, and all the JASA papers and IEEE papers and some geophysics journals, starting at the first pieces of data that we've taken and then all the way to the 2022 data, I've come up with a bibliography of peer-reviewed-- I'm not talking about the POMAs and all those things. I'm talking about peer-reviewed papers. There's been 77 so far peer-reviewed papers have come out of seabed characterization. Quite frankly, I don't think that's too bad. That’s a testament to how much work and the large group of scientists that have been involved in this type of put into it. And that’s going to be part of the legacy of seabed characterization is this large volume of peer-reviewed papers that people can go back and look at the details of what was done and accomplished.

ASA Publications (1:07:43)

Yeah, that's not shabby at all, I have to say.

Kyle Becker (1:07:46)

We're not prepared now, but we should be because we know that some of these students have gone on and they're going to stay in the field and they're going to have great careers.

ASA Publications (1:07:54)

So maybe this is a good segue. What are the next steps for this research?

Preston (1:07:57)

We're not quite done yet. So even though we've got three special issues, the most recent one has not made any really set of grand conclusions yet. So we've left these open questions and commented about what we're good at and what we're less good at, but we haven't answered all four of those questions that I spoke about a little while ago. So we’re going to have another workshop where all of the scientists will get together and we will try to compose our kind of summary statement or summary statements, which will also include kind of a way to summarize all of the results and all the data and all the things that we've learned. So we haven't done that in a concise way yet. You know, the special issue is 17 papers or however many there are. We would like to distill all of that down into a more, kind of a single one or two publications, peer-reviewed, we'll all be co-authors on that. And that will hopefully end the seabed characterization experiment, but I will say there are other things that will keep going. So there's another project sponsored by the same Office of Naval Research Program Office, now studying sediments in the much deeper ocean, and including effects of sediment stability and sediment transport, which were not really studied in the current one. So the work will continue in that way and in others.

ASA Publications (1:09:19)

Very exciting, at least. Do you guys have any closing thoughts?

David Knobles (1:09:23)

On a personal level, participating with Preston as a chief scientist was one of the most rewarding, if not the most rewarding, experience of my scientific career.

Preston (1:09:36)

Geez, David, that's amazing. Wow. Well, I can say exactly the same thing. Working with Kyle, it's been amazing. Yeah, it's been a great team. I mean, the team, scientists working together is an amazing thing. We can do things as a group that you can't do by yourself, and this is a great testament to that. So thank you for your kind words. I'm indebted to both of you all, as well as to the support we've had from JASA and the editor, Jim Lynch, and Kat, you as well, putting the issue together is a lot of work for everyone. So thank you very much for all of your help too, Kat.

ASA Publications (1:10:12)

Of course.

David Knobles (1:10:13)

I wanted to say a few words about foreign collaboration. And just to reiterate some of the things that were said about the critical importance about bringing in a team that included experts from all around the world and why that was so important. And I just wanted to mention a couple of names. There was Dag Tollefson from FFI Norway, and Dag and I go a long way back, and we've been trying to think of a way to collaborate for a long time even before seabed characterization. And all of a sudden seabed characterization popped up, and then Dag got into the experiment, and he brought in a one kilometer long horizontal line array that he deployed in 2000. Heck, I didn't even know they made arrays that long. And I felt that I knew something about arrays, acoustic arrays. So it was a wonderful success having… And Dag was well plugged in to folks like University of Victoria with Stan and Ross and doing source estimation levels. So I'm just so personally happy that Dag came in and we finally realized the dream that he and I had had for a long time of seeing a collaboration in seabed acoustics. 

Another person I want to point out specifically is Julien Bonnell of France. He's currently at Scripps, and he is fast becoming, if you just look, even a cursory examination of the literature, you're seeing that he's becoming a leading expert in ocean acoustics. And the decision by folks like Kyle to bring in Julien was in my opinion, well, quite frankly, it's one of the smartest decisions that you guys made. Okay, it was. He added so much to seabed characterization. 

Also want to point out Lin Wan from China. He used to be at UDel, but I just want to point out that Lin, that personal experience that I had where Lin published a very important paper in 2017, where he was the first person to observe what we call the low-frequency airy phase. And he observed that from data that he was analyzing on the ship while we were out there on the ARL-UT-Swami array in the NEMP. And I was on the Armstrong in the dry lab one day, and he came by and grabbed me, and he said, “David, you got to come and look at this on my computer.” And he was showing me evidence of this airy phase on the data that he was analyzing. So it's those type of things that I remember having that personal interaction with my foreign colleagues that I saw firsthand why it's so important for an experiment like this to have foreign collaborators.  You can have it, but it's not going to be as good if you don't have these folks.

Kyle Becker (1:13:24)

Yeah, I have very little to add.  I mean, it's just a remarkable community. And I think what was the wonderful thing, particularly in this day and age, is that when we were having meetings and we were all together, not just the three of us, but the whole community, everyone was just really working to learn, right? And it wasn't without debate. There were some really good questions and things being debated amongst the community. But I think as a result, the community got closer and stronger. And it's just a really wonderful time over the last 10 plus years.

David Knobles (1:14:02)

And something that will always need to be remembered about this feedback characterization for time immemorial is that Kyle went to sleep and woke up thinking about this problem, how could he help support the principal investigators and how to keep this experiment going when all the superiors around him are saying, “Gee, we got better things to do with the money.” Kyle had to fight this battle on a daily basis, and he did ,and that was one of the reasons that Preston and I were able to go out with this large, keep this large group of people ,help keep them together and I think it goes back to give credit to Kyle on this.

Kyle Becker (1:14:47)

Well, thanks for that.

Preston (1:14:47)

Yeah, absolutely. Absolutely, I agree with you, David. Thank you, Kyle.

Kyle Becker (1:14:50)

My pleasure.

ASA Publications (1:14:57)

Well, thank you all again for taking the time to speak with me today. I personally never knew how important ocean sediment could be, so this was a really enlightening discussion, especially finding out about how complicated mud is. I wish you all the best of luck in your future research, and have a great day.

Preston (1:15:13)

Thank you.

Kyle Becker (1:15:13)

Thanks, Kat.

David Knobles (1:15:14)

Thank you.