Stanford [00:00:13]: When sediment makes the news, it isn't usually because of the decadal scale erosion or deposition processes that we've been talking about on this podcast so far. When my non work friends hear about sediment transport, it is usually because of some dramatic event, a sort of flood that carries so much sediment, we're not even sure we should call it a flood anymore. And as wildfire frequency and intensity increases in the western United States, I've actually been doing a lot more work on these mud and debris flow events, which means I've been in that literature over the last five years or so, and you do not have to spend much time in that literature to recognize that. A lot of it revolves around the contributions from today's guest. Doctor Richard Iverson led a research program at the Cascade Volcano Observatory of the US Geological Survey, where he and his team did some of the most important work on muddy and debris flows that I've encountered. His team built a remarkable experimental research facility, a massive 95 meters flume built on a 32 degree slope, to investigate debris flow mechanics closer to the event scale. And as we talk about in the episode, it was doing the hard work of those large scale experiments that unlocked several new insights about how these flows behave, including some surprising counterintuitive findings that he tells us about. But Doctor Iverson's team also translated the experimental program into theoretical numerical work that incorporated these insights into a robust, flexible numerical model called DCLaw, which they and others have used to evaluate and forecast geological hazards around the world. So, because mud and debris flows are getting a lot more attention in river mechanics and flood risk management communities, I wanted to include a geophysical flow expert in this first season, but my wish list was really only one name long. So I was thrilled to get to talk to Doctor Iverson and actually learned quite a bit in real time during the course of this conversation. Stanford [00:01:58]: And I think you will, too. Stanford [00:02:00]: I'm Stanford Gibson, the sediment transport specialist at the Coors Hydrologic Engineering center. And on this episode of the RSM River Mechanics podcast, we are venturing into high concentration flows in a conversation with USGS debris flow expert, Doctor Richard Iverson. Stanford [00:02:18]: Richard Iverson, welcome to the podcast. Richard [00:02:20]: Thank you. Stanford [00:02:21]: So my understanding is you grew up in the midwest. Richard [00:02:24]: That's correct. Stanford [00:02:24]: So in the midwest, there's not a lot of like 30% slopes. And so I assume there's not a lot of opportunities to get interested in debris flow. How did it happen? How did you become interested in debris flows? Richard [00:02:35]: I really became interested in debris flows. I think originally from learning a little bit in coursework, both in undergraduate and graduate school, but especially when I was working on my master's thesis in the Mojave desert, which was not work that was intended to be about debris flows at all. We were actually doing rainfall runoff experiments with a small rainfall simulator and what we discovered on some slopes that were steep enough, in that case, steeper than about 25 degrees, and that had been where the soil surface had been disrupted by the action of off road vehicles, that when we applied our simulated rainfall to those slopes, in some cases, we'd actually get little debris flows forming. Stanford [00:03:13]: Oh, so you made debris flows. Richard [00:03:16]: We actually made little miniature debris flows. And by miniature, I mean extremely miniature. These things were probably less than one liter in volume, but nonetheless, they would flow down the slope, and they would exhibit a lot of the same characteristics as larger debris flows. They would create little lateral levees. There would be grain size segregation where those levees would tend to be enriched in coarser particles. There'd be a front on the flow, a bulbous front that was sort of being shoved from behind by more dilute material. And so, at least qualitatively, they had many of the features that you'd see in a full scale debris flow. And I just thought they were. They were fascinating. At the time, I wasn't certain that I would ever be able to study them as a full time job. But as it turned out, I eventually did get that opportunity. Stanford [00:04:03]: Well, I think we've already used a term that needs definition, because in my experience, the term debris flow has semantic range, right? Richard [00:04:11]: Yes. Stanford [00:04:11]: If one person says debris flow, three people hear it differently. Actually, it seems like there are even several different technical definitions. Why don't we start out with what is a debris flow? Richard [00:04:22]: You're absolutely correct that different people might think of different things when they hear the term debris flow. In my view, the crucial characteristics are, number one, that it's a flow of sediment and water that is almost entirely saturated. There might be small portions of debris flows that are not completely water saturated. For example, at a coarse grain snout that's being pushed along by material from behind, but by and large, it's completely saturated with water. The term debris itself has a precise geological definition because it implies that there's a great diversity of grain sizes. So, for example, a flow of just sand and water would not constitute a debris flow. And similarly, a flow of very fine sediment, silt, and clay size sediment that forms a water slurry would not really be called a debris flow. That would more correctly be called a mud flow because it is just mud size material or silt and clay size material. So those are really the most elementary aspects, I think, of defining a debris flow simply is it's a water saturated mixture of sediment of diverse sizes. And when I say diverse, in many cases, it's very, very diverse, spanning many orders of magnitude, everything from clay sized particles to boulders that could be as large as 10 meters in diameter. In simplest terms, those are the definitional characteristics without getting into all kinds of mechanical nuances and so on, which we. Stanford [00:05:51]: Can later, I think, when I hear a layman definition of debris flow, a lot of times they think about debris from a hydraulic point of view of there are logs or houses have been knocked down and there's anthropogenic debris or something like that. And that might be incidental to debris flow. But what's actually causing the mechanics are the rocks and soil of very different sizes, and it's saturated, and the pore water pressure is causing some of these effects. Richard [00:06:18]: Right, right. Your point, though, regarding the wood debris, whether it's logs or whether it's anthropogenic debris, I think that is an important one, because in many cases, debris flows do carry large volumes of that kind of material. It's very conspicuous because it tends to float on the surface. And so you see a lot of it mechanically, though, in most cases, it really is playing a pretty secondary role. It's just kind of being rafted along with the other material, really having the major mechanical effects. Stanford [00:06:46]: How is a debris flow different from a mud flow or a landslide? Richard [00:06:50]: In terms of the distinction between a debris flow and a mud flow, it really is the grain size distribution. So, again, strictly speaking, a mud flow is something where the sediment is all silt and clay size. Now, that doesn't mean it can't have a little bit of sand sized material in it, but it's basically silt and clay size material, whereas the debris flow has this great diversity of grain sizes. And generally speaking, debris flows, at least in a sub aerial environment, meaning not underwater, generally, they really don't have that much silt and clay size. If it's as much as 20% by weight, that's a lot. That's a lot more than average for debris flows. Many debris flows have far less than that, but they usually do have some. Having at least a little bit of silt and clay helps maintain the high pore pressures that enable debris flows to have their high mobility. So that's one distinction. A couple of other distinctions that are important are in terms of sediment concentration and degree of saturation and how things we call debris flows kind of are transitional from other sorts of flows. So for flows that are a little bit more dilute, typically people differ on how they define this, but typically around 40% sediment concentration by volume is where something we call a debris flow starts transitioning to something we call hyper concentrated stream flow. So if it's less than 40% solids by volume, it might be more hyper concentrated streamflow, whereas more than 40% by volume, it's more what we would call a debris flow. The other gradation is the difference between a debris flow and a debris avalanche, because debris avalanches have those same grain size characteristics as a debris flow, the great diversity of grain sizes. But generally speaking, they're not 100% water saturated. They can be partly water saturated, and they can be partly liquefied at their bases, but they don't have the character of being a flowing slurry that's fully saturated with water the way a typical debris flow would be. Stanford [00:09:03]: It sounds like you described a continuum where there is kind of these hyper concentrated flows and then debris flows and then debris avalanches based on the ratio of water and solid particles. I guess a question I've always had is, is this a smooth continuum where the distinctions are kind of arbitrary, or are these distinct processes where the thresholds are pretty abrupt? Richard [00:09:28]: I think it's a smooth continuum, and I think that's what can often lead to communication problems, because what I call a debris flow. If we both stood, say, in the field somewhere and watched a flow come by, you might say, well, that's a hyperconnaissance stream flow. And I'd say, no, I think that's a debris flow. And the fact is it might be somewhere kind of in between, because it is a continuum and a gradation of processes. Like most things in geosciences, it's hard to draw hard and fast lines pigeonholing different kinds of behavior. Stanford [00:09:58]: So, two places where I've been working that we kind of encounter these is post wildfire rainfalls and mine tailing dam failures. And I think the post wildfire, you get the great diversity. You get concentrations above 40%, the mine tailing dam failures there, you get concentrations over 40%. But maybe not the great diversity of grain sizes. Are those more like a mud flow, then? Richard [00:10:23]: Yeah, I think in many cases those are mudflows, because so often with those tailings, dams, as in any situation where you have a slurry like that. The intention, of course, is that the sediment will settle out over time when those tailings damage. But of course, a lot of the finer stuff remains in suspension for a long time, and it's easily remobilized. So that then when the dam, if and when the dam fails, and that's a far too frequent occurrence. Failure of tailings, dams. Yeah, you'll get this great rush of material coming out that often does have the character of a mud flow. It's highly concentrated, but it has very little coarse sediment in it. Stanford [00:10:59]: And we were just chatting a little bit before we started here about the incidence of wildfire in the west here, and with wildfire in the west increasing. Is this kind of a big secondary hazard associated with wildfires? Richard [00:11:11]: Absolutely. Debris flows or related phenomena like hyperconcenary flows are a big secondary hazard associated with wildfires. And that's been particularly the case in a setting like southern California, where after a wildfire, what's often left on the ground surface there is a loose, in many cases what we would call a granite groose type material, just sort of granulated soil. There's little in the way of remaining roots holding it in place, and so it's very susceptible to being entrained by running water. Interestingly, that seems to be less true here in the Pacific Northwest when we have post wildfire situations. And I think it's largely a function of the fact that, first of all, our soils are different to begin with. But also in mountainous areas here in the northwest, there's such a tremendous amount of root structure within the soil that even if the trees are killed in large part by the fire, there's still a lot of root mass left in there, which will, over time, eventually rot away. But what I've noticed around here with our quite wet winters is that within several years after those wildfires, there's already a lot of lush regrowth of smaller plants or seedling trees, too, so that by the time six or seven years passed, and maybe the roots of those larger trees that were killed by the fire are largely rotted away. You've already got good ground protection there that's shielding the soil from the most intense rainfall effects. So it really does seem that semi arid areas like southern California or parts of the Rockies are really quite different from the Pacific Northwest in that respect. Stanford [00:12:52]: I have heard that about the northwest. In the southwest, the most dangerous time for laughter fire is immediately after Orlando. In Santa Barbara, the rainfall actually put the fire out. But in the northwest, it's this potential window between when the trees rot and fall over or the undergrowth is racing that, and there could be a window where it's but it's not necessarily immediately after the event. Richard [00:13:22]: Yeah, no, I think that's absolutely right. And in some places, that window is probably wider than in other places. I was just. I've been spending a lot of time lately out on the north side of Mount Hood, where there have been some biggest wildfires in the last decade or so. And I've been impressed at just how much revegetation has grown there in the last decade. There are lots of sapling trees that are already 10ft tall and a great deal of dense underbrush beneath those trees. So they're pretty much invulnerable from the direct effects of rainfall impact at this stage. Stanford [00:13:56]: Okay. So I think part of the thing that I think is remarkable about your career is that, you know, they're kind of experimental people and they're kind of numerical people, and those people who dabble in both. I'm someone who dabbles in both. Usually they do one well and the other one, you know, but your experimental work and numerical work are just both kind of standard setting. But you're probably best known for your experimental work. So let's start there. Your program built the biggest flume I've ever seen. Can you describe it for us? Richard [00:14:24]: Sure. So back in 1991, we began construction of a facility we call the USGS debris flow flume. And essentially what it is is a big concrete chute on a steep hillside. It's located at a place called the HG Andrews Experimental forest, which is about 50 miles east of Eugene, Oregon, in the Cascade range foothills. And so this big concrete chute is 95 meters long, 2 meters wide, and 1.2 meters or 4ft deep. And what's most unusual about it is that it's on a steep slope. It's on a 31 degree slope. The upper, roughly 80 meters of it is on this steep slope at the top of that flume. We can do a couple different things. We can load material behind a big steel gate at the top of the flume and then saturate it and let it go. By opening the gate doors and create a debris flow. That way, we can also trigger more natural kinds of debris flow onset by simply piling sediment behind a retaining wall, adding water until it fails, whenever it's ready to fail. And then it too would flow away, at least in some cases, would flow away as a debris flow. The question often comes up, why build such a big flume? Obviously, it's more labor intensive to work at that scale. It's more expensive to build something that big. But the reason behind that all along was that it became pretty clear to me back in the 1980s when I first started thinking about debris flow mechanics, that there were some significant scale dependent effects going on, that we're sort of biasing people's scientific views about debris flows. The most important of those scale dependent effects just has to do with the role of the liquid phase at different scales, the liquid phase and debris flows. It's useful to think of it as being not just the water, but the water, plus the very fine sediment, the silt and clay size sediment that's truly held in hydrodynamic suspension. So this is material that does not need to be suspended by interacting with particles. It's material that's just held in suspension by viscous forces within the fluid phase. When you do a miniature debris flow experiment, say on a tabletop scale, if you use, say, a mixture of this silt and clay suspended in water with sand or maybe some pebbles, and do your debris flow experiment that way at those miniature scales, two properties of that fluid phase have quite important effects. One is simply the viscosity, which, you know, which may be somewhat greater than the viscosity of pure water, you know, maybe ten times greater. That's still a pretty small number. Even if it's 100 times greater than that of water, you're still talking about something that's a fairly small number. Similarly, that silk clay slurry can have a bit of yield strength, meaning that it has an intrinsic strength analogous to the cohesive strength that a fine grained soil would have. But again, those numbers, if you actually measure them, are quite small. They're typically a few tens of pascals and virtually always less than 1 kpa. Well, the thing is, that kind of yield strength and that kind of viscosity can really have a significant effect on retarding the motion of a flow at a miniature tabletop scale. But you put that same material into a full scale debris flow, and the effects of those forces, those viscous forces and the effect of yield strength or cohesive strengthen become really small compared to the frictional interactions of the larger solid particles in the flow, the stuff that sand size and coarser all the way up to boulders. And so, you know, ideally for doing debris flow experiments, you'd make your flume as large as you possibly could. Now, of course, practical constraints eventually come in. And so we found that really the largest scale that was practical and affordable for us was to build something on the scale, this debris flow flume, that allowed us to make debris flows up to about 20 m³ in maximum size. Oh, wow. Stanford [00:18:27]: So let me see if I can reproduce this. It's really important that the fluid phase is actually not just water, but also the finest sediment. Richard [00:18:35]: Right. Stanford [00:18:36]: Understanding the interactions and the internal pore water pressure and the viscosity, those things are all very dependent on that. Richard [00:18:42]: Right. Stanford [00:18:43]: But if you go to a small scale, a table stop scale, or even a laboratory scale, the finest particles are no longer part of the fluid. When you shrink down the scale, they behave more like the solid component, and you can't get to those mechanics. Richard [00:18:57]: Right? Yeah, no, that's absolutely right. And something I didn't mention before that's pertinent in that respect, is that in terms of generating and preserving high pore fluid pressures, in a debris flow, what's really crucial is sort of the time scale over which the poor fluid pressure can dissipate. Because if the time scale over which you're generating the pressure is comparable to the timescale over which it can dissipate, then you're not going to have a strong propensity for building up high fluid pressure. But it turns out that that time scale for dissipation goes as the square of the thickness of the flow. If you change your flow from being, say, a ten centimeter thick tabletop flow to a 1 meter thick, larger experimental flow, now, you've changed that time scale by a factor of 100, and that makes a huge difference. It means that maintaining that fluidized liquefied state is much, much easier in a large scale flow than it is in a small lab scale flow. Stanford [00:19:59]: Okay, so you've mentioned this internal pore pressure several times. And I actually want to pause on it a little bit, because I think this is one of the most important things I've learned from your literature, is a lot of fluids have what we call a hydrostatic pressure distribution, which just means that wherever you are in the water or fluid column, the pressure is the weight, essentially, of the fluid above it, and so it decreases linearly with depth. And one of the things that you really press on in your literature is that debris flows are non hydrostatic. And that's really important for how these kind of counterintuitive processes work. Can you help me understand that a little bit? Richard [00:20:37]: Sure. Sure. So that is a really, really central concept. I guess the way to think about that at the outset is to think about how do you generate those non hydrostatic pressures, those greater than hydrostatic pressures? Because left to itself in an undisturbed state, water, whether it's just a standing body of water or whether it's water in sediment, it will always return to a hydrostatic pressure. That's its equilibrium pressure statically. However, if you start with a mixture of sediment and water and it starts in a relatively loose configuration, meaning that when it begins to deform, it actually densifies. So you're reducing the amount of pore volume in the material. As you reduce that pore volume, that's putting pressure on the water in those spaces between the sediment particles. The water is trying to. It's trying to escape. It's trying to, it's always trying to get back to that equilibrium hydrostatic pressure, but it just can't do it fast enough. And so you can liquefy material quickly. And of course, one way to do this is during earthquakes. You know, liquefaction during earthquakes is well known, and that's doing to due to putting shaking energy into the material. But fundamentally, once you're shaking the material, what's going on after the shaking begins is the same. It's the particles in the soil basically trying to settle, trying to densify, and therefore driving up the fluid pressure. Well, the same thing can happen in a landslide and generate a debris flow if the landslide starts out in a relatively loose state. It can even happen if a landslide doesn't start in a loose state, if the landslide becomes looser through the action of dilation, meaning that maybe the landslide goes down a very steep slope and launches off a cliff or something and becomes very dilated. Well, now, once it's very dilated, assuming there's enough water in the material beneath it, the sediment's going to try to settle once again from that dilated state, which again will tend to in between the pores. Right? Right. Which again will tend to drive up the liquid pressure in the pores. The thing about a large debris flow, and particularly a large volcanic lahar, is that once that high pore pressure is established, it can take a long, long time to dissipate. And so what our calculations indicate is that for some of these great lahars that are 1020, 30 meters thick, it can take many weeks, in some cases even years, for that pressure to completely dissipate. And so once it has developed this state of being, this soupy mess of liquefied material, it may take a long, long time for it to consolidate back to something that's firm enough to walk on, for example. And of course, that can greatly complicate search and rescue operations in these kinds of circumstances if basically what you're dealing with is quicksand. Stanford [00:23:26]: Yeah. My understanding from your literature is that it's this positive pore water pressure or this non hydrostatic pore water pressure that is part of the reason that these masses are so mobile or that they move so quickly. But one of the things that was count, there's a lot of things about these processes I find counterintuitive. But one of the things, the first time I read it in one of your papers, I almost didn't believe it was that you said that these can have up to 20% clay by weight. But the interesting thing was that the more clay one of these had, the faster it went. And I thought that, well, there's no way is that true, because Clay's sticky. It'll slow it down. But it's connected to this pore water pressure. Richard [00:24:06]: Yes. Unless you really have a lot of clay, the clay gets to be so concentrated that it becomes just a gooey, gooey mess. That's one thing, you know, say maybe up to 20% by weight. That's dry weight. Its primary role is basically helping the pore pressure to remain elevated by making it more difficult for the pressure to equalize. And partly that's just because it increases the fluid viscosity. Or at the same time, you can think of it as reducing the permeability of the mixture. It kind of does both things together, but it just makes it much easier for the high pressures to persist. Stanford [00:24:40]: So if you were thinking about if you had a soil and you were going to pump water through it, or you had a soil and you were going to pump, like, corn syrup through it or something like that, the corn syrup would take longer to pump through it, and it would maintain a higher pressure if you were to try to compress it? Richard [00:24:56]: Yeah, yeah, that's a good way to think about it. If you made a debris flow out of corn syrup and rocks, it would have a particularly strong potential to remain liquefied. Stanford [00:25:04]: And so the more clay you add to the liquid phase, the more it behaves like that theoretical corn syrup event. Richard [00:25:11]: Right. Right. Stanford [00:25:12]: Now, you've used a couple of terms here that we are not accustomed to using in water modeling. You talked about dilation and compression. We usually think of water as a non compressible fluid, but it doesn't sound like, a, it doesn't sound like that's true about these mixtures, and b, it sounds like the fact that they're compressible is actually really important in understanding how they behave. Richard [00:25:36]: It is really important. And it's also important to think about compressibility in a slightly different way than would be typical for a pure fluid. So if we're thinking about water, yes, it's basically an incompressible fluid. If we're thinking about air, yes, it's a compressible fluid. But what makes a debris flow behave like a compressible material is not so much that either the water is compressible or the rock components are compressible, because neither one really is. It's simply the fact that the volume fraction, the concentration of the solids, can change greatly during the course of a debris flow event. And so, well, let's say I can kind of give you a scenario where we start from the beginning and follow a debris flow throughout its sort of life history. So we might start up on a slope somewhere, a steep slope where we have relatively loosely packed sediment, and it's been loosened over the years by maybe growth of tree roots, maybe by burrowing animals, maybe by various volcanic processes like hydrothermal alteration. But any event, it starts out in this fairly loose state, and for whatever reason, it's nearly saturated with water. Maybe it's heavy rainfall, maybe it's snow melt, whatever. So now the slope fails because it starts out loose. That material will tend to densify through the fact that the solids are sort of collapsing on themselves and increasing the solid concentration. It's acting compressible in the sense that its bulk density has changed. The mixture's bulk density has changed quite a lot. Even though the solids themselves are incompressible and the liquid water itself is incompressible, the mixture acts compressively. If it has sufficient fine sediment in it, it will probably stay rather liquefied as it continues to flow down the slope and so on. And then eventually it will stop somewhere and form a deposit, in which case it will densify even further. And in that case, it may actually consolidate and water may start leaving the mixture, you know, seeping out around the margins and so forth. What keeps it from densifying quite so rapidly, as long as it's moving, is this tendency for agitation in the flow to persist and to keep it from densifying as much as it would if it were just sitting there statically. Stanford [00:27:58]: So I've anticipated this question a little bit just because we got into the processes already. A lot of these processes that we're talking about, these are learnings that came out of some of these big experiments that you did in this flume, which, incidentally, I once climbed to the top of the flume, and I'm not an athlete, but I run every day, and I coach youth soccer, and I was winded by the time I got to the top of this flume, just to kind of give you a picture, you're climbing a lot of stairs. Richard [00:28:23]: Yeah, I believe there are 271 stairs, because I used to do a lot of running, too, and did running even when we were working at the flume. In fact, I remember one year I was there, I was training for a marathon, and I would go out. I'd run 5 miles in the morning, work all day at the flume, and then run another 5 miles, which was a pretty. Made for a pretty tough week. But I would also do sprints up those steps at the flume. Yeah. So I timed myself many times. Stanford [00:28:52]: Maybe that is that. Maybe that's some career advice. If you want to get into mesoscale modeling, you need to be in shape. Richard [00:28:58]: Yeah. Yeah. That probably does make sense. Stanford [00:29:00]: What are some of the other big ideas that emerged from these iconic experiments? Richard [00:29:05]: So I would say some of the most important ideas, one just has to do with this basic kind of mechanical structure of debris flows, which we've already covered to some degree. The fact that the flows are largely liquefied, they also develop this kind of classic head and tail structure where you tend to, because of grain size segregation effects, you tend to wind up with a coarser grained aggregate of material right at the snout of the flow and also at the front of secondary surges, because these flows on these steep slopes do develop flow instabilities, where you get things like roll waves that form in steep water flows. Stanford [00:29:42]: I feel like that might be surprising. It was certainly surprising to me the first time I looked at your diagram in one of your classic papers. That seems counterintuitive. It seems to me like the fines would be out in front and then the boulders would be rolling behind it. But that's not how it works at all. Richard [00:29:58]: No. What happens? Grain size segregation is an amazingly efficient process. And the classic experiment, it's actually one of the papers that described us, was given the title, why the Brazil nuts are on top. And it talked about, if you take a can of mixed nuts and shake it, you don't have to shake it very much. And you open the can, you find out all the big ones are on top. That is the Brazil nuts. But the same was true with any other ground material that's subject to agitation. Size segregation is very efficient. In the case of our flume experiments, you can see evidence happening within the first 10 meters or so of travel downslope. And so what happens is that the larger sediment tends to work its way to the surface, which in our case was usually gravel. Sometimes we used cobbles up to sort of fist size. That was the biggest material we ever used as a flume. But that material would tend to wind up at the top. And because there is some velocity profile in the debris flow, not well defined, but we do know that at the surface, it's moving faster than it is right at the bed. You know, over time, that coarse material that's come to the surface of the flow will wind up being transported to the front. And once it gets to the front, it kind of gets stranded there because it just keeps getting pushed along by the material behind it. Stanford [00:31:14]: I never understood this. You have the brazilian effect that pushes these rocks to the big rocks to the top. Richard [00:31:19]: Right. Stanford [00:31:20]: And then you have a differential velocity. That's essentially a conveyor belt on the top. Richard [00:31:24]: Exactly. Stanford [00:31:25]: The conveyor belt to the front. And then there's just this big, coarse snout in the front. Richard [00:31:30]: Exactly. I mean, you can almost think about it like the tread on a caterpillar tractor or something. And you can imagine if it were carrying along on its surface this coarse sediment that tumbled off the front of the tread as the cat moved forward. If that were fine sediment, it would just get rolled under. It would just drive over the top of it. But if that sediment were coarse enough, it would just kind of get pushed along in front of the cat track. And that's kind of what happens with the coarse sediment in these debris flows. And so the fact that that structure, that head and tail structure was so, so readily developed and so persistent, that was a key observation at the flume. It also helped us understand the relationship between the liquefied core of the flow and the more resistive front that was being pushed by that more liquid core. So that basic observation of the structure was really important. A couple of other things that were really important, kind of fundamental properties that we learned about pertain to the effects of the basal boundary condition. And I'll talk about just two. One is the difference between a smooth bed and a rough bed, and the other is the difference between a rigid bed and an erodible bed. For the first, I guess it was the first eight years that we operated the flume. The bed of the flume just had a relatively smooth, finished concrete surface. It was a broom finish, like they'd put on standard sidewalks. And the reason we did that was it was sort of the simplest thing to do, and it was sort of the obvious place to start. But we always knew that it wasn't terribly realistic in terms of most real debris flows. So, in the year 2000, we took the plunge of surfacing almost the entire bed of the flume with a bumpy bed with bumpy concrete tiles that we specially fabricated for this purpose. We did a lot of trial experiments to kind of come up with the bump configuration that would give us the amount of friction we wanted at the base, and we settled on something that looked, oh, if you imagine something that. Where you took ping pong balls and sliced them into hemispheres and just had the whole bed of the flume studded with hemispheres. When we switched from the smooth bed to the bumpy bed, certainly the flows became more realistic, and they became both deeper and slower than they had been on the smoothbed. But quite a surprise was that they actually ran out further at the foot of the flume, even though they were moving more slowly up within the flume. When they reached the relatively horizontal surface at the foot of the flume, which is just a smooth concrete pad, they would actually run out considerably further than they had with the flows running over the smooth bed. And the reason for that all comes back to this importance of grain size segregation, again, because it turned out that by having the bumpy bed, that that increased the amount of agitation in the flow. As you might well imagine, the stuff is just being rattled to death as it comes down across that bumpy bed. And that enhances the grain size segregation process. And when the flows issue from the mouth of the flume, the grain size segregation causes them to form coarse grained lateral levees that then channelize the ensuing flow. It doesn't readily escape from over the top of those levees. It just keeps its momentum going straight down. Stanford [00:34:48]: So it's like a rifle. Richard [00:34:49]: It's just shooting like a rifle, shooting the flow out. And so with the smooth badge that we had originally used, what would happen is that we would get levees to some degree, but they were not well developed. And so the fort. The deposits that would form at the foot of the flume would spread out quite a bit laterally and kind of form a pancake deposit. And then once the bed was roughened, we'd get these long, linear deposits that would often be twice as long. And it just showed how important that grain size segregation effect is, not just as a phenomenon of its own importance and interest, but the fact that it has feedback effects that influence the overall dynamics, the debris flow. And so that's sort of a fascinating thing from the standpoint of basic physics, because it's showing that there's a feedback mechanism whereby what's happening in a small scale process, that is, the grain size segregation can end up then influencing the large scale process, the debris flow dynamics that are actually the source of the phenomenon to begin with. And so there's some very interesting two way coupling going on there. Stanford [00:35:59]: And that just sounds like a classic moment in just science. You fabricated these bowling ball plates essentially, and things behaved the way they should, except they also behaved completely differently in a way that only made sense in retrospect. Richard [00:36:13]: Exactly. Exactly. Something we would not have anticipated. You know, it's the sort of thing where you can say, well, you should have anticipated. Right, right. Stanford [00:36:20]: But that's why we. It's like they say in football, that's why we play the game. Richard [00:36:24]: Exactly. Stanford [00:36:24]: In science. That's why we run the game. Richard [00:36:25]: That's why you do experiments. Exactly. And so then the other thing regarding basal boundary conditions, that was just as important and really even more exciting had to do with the effect of erodible boundaries. Now, it turns out that doing erodible boundary experiments to flume, it's a difficult thing to do. Stanford [00:36:42]: That sounds really challenging at this kind of slope. Richard [00:36:44]: Really challenging to do at that slope. Stanford [00:36:46]: It's really challenging to do at. .001 and then what made it even. Richard [00:36:50]: More challenging is we realized fairly quickly the water content of that bed sediment was going to be really important. And when things get really dicey is when you have this sediment sitting. And typically the sediment would be ten to 12 cm thick, sometimes thicker, sometimes less. But that's kind of typical, ten or 12 cm on the bed of the flume, same sediment that made up the debris flow itself. But we decided that we needed to add water to it, because if that sediment was just completely dry, what would happen is the debris flows would, they would hit that dry sediment and they would just sort of bog down, because as they incorporated that dry sediment, they'd scour some of it, incorporate it, and then just sort of dried out the debris flow. Stanford [00:37:30]: Yeah, you'd lose the water into the substrate. Richard [00:37:31]: You'd lose the water into the substrate. So we needed to wet the bed sediment. And then it became a really dicey procedure of how wet can you get. Stanford [00:37:39]: It without triggering a debris flow? Richard [00:37:40]: Without triggering a debris flow? Because sometimes we did trigger a debris flow. Stanford [00:37:44]: That's a bad day. Richard [00:37:45]: So we spent a lot of time monkeying around with lots of little intricate sprinklers and so forth, and we finally got something that worked reasonably well. It wasn't 100% reliable, but it worked reasonably well. And it turned out that by varying the water content of that bed sediment through the range of about 10% by weight versus 30% by weight, which is pretty darn wet. Not completely saturated, but pretty close to it got enormously divergent results when we ran debris flows across them. When the bed sediment was sufficiently wet, we'd get this explosive positive feedback. As the debris flows ran over the top of this sediment and began to entrain it, they could compress and shear the underlying sediment sufficiently that would actually liquefy the bed sediment. And so now you've got a debris flow that not only is the flow itself largely liquefied, but it's running on top of a low friction bed. And those things would really move and grow explosively in momentum as well as speed because they were picking up mass at the same time they were picking up speed. Those were really exciting and kind of scary, actually. I mean, we destroyed some equipment, inadvertently destroyed some equipment. Stanford [00:39:07]: So you have collected the videos of these, and they're dramatic, they're so interesting. You've collected them all online. We'll put a link to them in the description below. But for those who are just having the audio experience here, your flume is roughly 100 meters long, and the 100 meters dash record, like a human being, can do this in about 9.5 seconds. Richard [00:39:28]: Now. Stanford [00:39:29]: So how would your debris flows compare to that? Richard [00:39:32]: Because we did various different kinds of debris flows. I'll kind of categorize these a little bit for the debris flows on the smooth bed, the smooth concrete bed that had not been roughened with the bumps. Those reached speeds that were somewhat faster than Usain bolt ever reached. For those of you who don't follow track and field, he's for some time now been the world record holder and 200 meters, but not a lot. They would reach speeds of maybe 11 meters/second the flows on the bumpy bed were somewhat slower than that. A little bit slower than Usain bolt. Stanford [00:40:07]: But still on that order. Richard [00:40:08]: Yeah, but on that order. However, those ones I was describing with the erodible beds that were subject to liquefaction and those reach speeds as high as 16 meters. Stanford [00:40:17]: Oh, my goodness. Richard [00:40:18]: So that's really moving. Stanford [00:40:19]: Yeah. About 30 or 40% faster than a hundred yard dash. Richard [00:40:23]: Yeah. Yeah. Stanford [00:40:24]: This is one of the interesting things about the program here, is that it's enough that you learned all those things about these processes, but isn't kind of where your team stopped, you took those processes and you quantified them and you built a really robust, really flexible geophysical flow model called DCLaw. And so what are some of the distinctive capabilities of DCLaw. Richard [00:40:48]: So the thing that distinguishes DCLaw from the models that came before it, and I make that state explicit, the models that came before it, because since we developed DCLaw and published out, there are other people who continue to build on that. But ours was really the first debris flow model that tried to account and did account explicitly for these effects of density change or dilation that we felt was a really important step forward because it allowed us to explicitly simulate these feedback processes that generate pore pressure or allow it to dissipate, which in turn influences the effective friction so much because the higher the pore pressure, the lower the friction. That's sort of been established for 100. Stanford [00:41:34]: Years now or so, just to kind of untie that. The higher the pore pressure, the lower the friction, the faster the mass moves. Richard [00:41:42]: Right, right. So that was a major feature of the model. Another feature of it that was something that I had in mind from the very beginning, and this did work out in the long run, is that in the limit of vanishing solid volume fraction or vanishing solid content, the equations reduce back to the standard shallow water equations. Stanford [00:42:06]: Okay, so I could just run a flood model. Richard [00:42:08]: You could run it as a flood model. If you just specified that the. The solids fraction is zero, basically, it would become a flood model. At the other end of the spectrum, if you say that the fluid has no density and viscosity, it now becomes a dry granular flow model. Stanford [00:42:25]: So it's like kind of a geotechnical landslide model. Richard [00:42:27]: Yeah. And then, of course, it can span that whole intermediate space, which includes debris flows and hyperconcenated flows and so forth. So the model is, it's somewhat complicated mathematically and numerically, but in our view, the complications are worth it in terms of what you get in terms of functionality out of it, a couple of. Stanford [00:42:47]: Things that are important about it. It's a multi phase model. What is a multiphase model, and why is that important? Richard [00:42:52]: So, it's multiphase in the sense that it tracks both solid phase and fluid phase somewhat independently of one another. But it's also a mixture model in the sense that ultimately, in the momentum equations, they're not separate momentum equations for the solid and fluid phases. Those are merged through coupling terms that are basically the result of solid fluid drag, relating the fluid phase motion to the solid phase motion. But there are entirely separate equations for mass conservation of the solid and fluid phases. And then there's kind of an equation for the evolution of the pore fluid pressure. And then it's the interaction effects between the solid and the fluid that produce this extra equation, this pore pressure evolution equation. Stanford [00:43:35]: And that's really important because we've talked about this compressibility, or kind of quasi compressibility, and the non hydrostatic pressure distribution is really important to how these masses move and how fast they move. Right, okay, so you've already alluded to this, but we're sitting here in the Cascade volcanic Observatory, which I think is one of the coolest federal offices in the portfolio. The Cascade Volcanic Observatory was stood up to monitor the activity of the volcano, the Cascade range in the northwest. So what is our premier debris flow facility and team doing in a volcano lab? Richard [00:44:15]: The reason debris flows became such a focal point of investigations here at CVO, because CVO was established, really, in response to the 1980 eruption in Mount St. Helens, is that it had been appreciated, really, since the late 1960s, but even more so after the Mount St. Helens eruption. In many ways, the single greatest hazard associated with cascade volcanoes are the lahars that run down the valleys that drain the slopes of the volcanoes. Stanford [00:44:46]: So I heard the term lahar many times before I understood what it was. Could you just maybe define it? Richard [00:44:51]: Well, in simplest terms, lahar is just an indonesian term for a debris flow, or potentially a hyper concentrated stream flow that originates on a volcano. And importantly, that does not have to be an origination that includes an ongoing volcano eruption. It could be a debris flow that originates from a big landslide on the side of the volcano that might take place where there hasn't been an eruption for thousands of years. But the point is that it begins on a volcano. And we tend to use the term lahar because it's a very broad term, and it sort of frees us up from the burden of saying it's explicitly a debris flow or explicitly a hyperconcence. It's a bigger bucket. Stanford [00:45:32]: That's great. Richard [00:45:33]: It's a bigger bucket. And that makes life a little easier when you're communicating with the public. Stanford [00:45:37]: Yeah. Can you just tell us a little bit about what makes Lahars special, what makes them distinct? Although you did just tell us it's a bigger bucket. Richard [00:45:45]: So it's a bigger bucket. And of course, it does have to begin on a volcano. That's another distinct attribute of lahars. But I think the other thing that's distinct about them is just that they can be so immense compared to any debris flow that would occur in a non volcanic setting. And so the world's largest lahars, the ones that we know about not so much from historical times, but from the geologic record, you know, were several cubic kilometers in volume, you know, so several billion cubic meters. To put that in context, the largest debris flows I'm aware of, at least in historic times, of non volcanic origins, are more on the order of perhaps one or 200 million. So we're talking about orders of magnitude difference and maximum size. And also just the abundance of very large events from volcanoes is much more substantial than in non volcanic settings. To me, one of the most interesting events in recent decades that's been instructive in many ways was an event that did not draw that much attention here in the United States, but it occurred just 100 miles north of the border, up in British Columbia, but in a rather remote part of British Columbia. There's a volcano there called Mount Meager, which is at the very northern end of the Cascade range of volcanoes. It's about 100 miles straight north of Vancouver, BC, without any triggering event, no volcanic activity, no earthquake. The south face of that volcano collapsed. About 50 million m³ collapsed, and nearly all of it transformed into a debris flow, or lahar. Since it was coming from a volcano, we know an awful lot about that event simply because it was big enough and there were enough seismic instruments in the vicinity that we have quite a detailed understanding of how fast it moved, how quickly it mobilized. And, of course, there was a lot of mapping and so forth done after the event. And there was a little bit of somewhat sketchy eyewitness account as well, although it happened in the middle of the night. So these eyewitnesses, it was really more than what they heard. Ear witnesses. But nonetheless, because it happened recently and because it was possible to document it using modern technology, and because we have all the seismic evidence, we know a lot about the event. And the reason it's important is that it provides an excellent modern analog for what we know could happen at any of the cascade volcanoes. And the fact that it happened without any obvious precursor is really important from a hazard standpoint. Stanford [00:48:20]: Lahars is not something that. That I've done any work on. Everything that I know, I learned from reading to prepare for this. It seems like you mentioned three triggers. There can be volcanic activity, earthquake, or kind of none at all, just a gravitational trigger and maybe reaching back a little bit farther. So that was a spatial analog. It seems like there's a temporal analog in some of our us population centers. The electron debris flow, can you tell us a little bit about that? Richard [00:48:47]: Sure. The electron mudflow. We call it a mudflow simply because that was the name that was originally given to it by the first archaeologists who study it nowadays, we would call it a lahar. It. As far as we know, it occurred roughly 500 years ago. It originated high on the western side of Mount Rainier. And as far as we know, based on the geologic evidence, it was not the result of volcanic eruption. There's simply no eruptive materials that have been found in association with that deposit. And so we believe that what happened is that it was just a big landslide, similar to what happened at Mount Meager in 2010, but much larger. And it traveled down some very steep slopes and canyons. As it left, the immediate vicinity of Mount Rainier reached the lowland area, where there are now several tens of thousands of people living. Particularly the closest large town is a town called ording, but inundated, a broad swath of lowland area that's now heavily populated. And so if something similar to that were to happen today, it really could be potentially quite a devastating event. And that's one of the reasons that the USGS has put a lot of effort into installing monitoring equipment to try to detect an event like that. We've also put in a lot of effort recently in modeling similar kinds of events to try to understand their dynamics better and refine hazard forecasts. Stanford [00:50:11]: And so you have a kind of a geophysical flow model that spans all of the entire continuum that we've talked about, from hyper concentrated flows to debris flows to landslides. And you've done some modeling on this. How fast is one of these lahars? Richard [00:50:28]: So the speeds that lahars can reach in the steep upper canyons where they're leaving the immediate faces of the volcano at high elevations, the speeds can be astounding in those, those kinds of localities. For example, with this event that occurred at Mount Meager in 2010, what the seismic data interpretations indicate is speeds up to about 200 miles an hour. So that's pretty darn fast. Stanford [00:50:54]: 200 miles an hour? Richard [00:50:55]: 200 miles an hour. Stanford [00:50:57]: I don't know any other geophysical hazard that wins the Olympics of the geophysical hazard. Richard [00:51:04]: That may well be true. And I think that's the reason for those high speeds, is the combination of very steep slopes and very large volumes of material. So there's just a whole lot of momentum involved. Pyroclastic flows can move very fast, but they seldom have the great volumes that some of these larger laharas have. Also, our model results, our simulations for the west side of Mount Rainier indicate similar kinds of peak speeds. So that seismological inference for Mount meager is not at all out of line with what we infer from the modeling, other kinds of evidence are the fact, for example, that that Mount meager event at one point reached a t intersection, a confluence of two streams that basically formed a t intersection. And there was a mountainside on the downstream side of that t. And that flow, when it hit the mountainside, ran up more than 800ft vertically up the side of that mountain. Stanford [00:51:58]: Just turning kinetic energy into potential energy, basically. Richard [00:52:01]: Yeah. Stanford [00:52:01]: How many feet? Richard [00:52:02]: 800. Stanford [00:52:02]: Oh, my goodness. Richard [00:52:03]: 800 vertical feet. Yeah. Stanford [00:52:05]: I've never heard of anything like that. Richard [00:52:07]: So, yeah, it's somewhat staggering. And of course, in simplest terms, you're right, it is absolutely just a conversion of kinetic energy to potential energy. But of course, there's also the fact you've got momentum continuing to come in and push in from behind. It's continuing to feed more and more momentum into that snout as it climbs up the slope. Stanford [00:52:29]: The electron mudslide lahar is estimated at about 260 million m³. Richard [00:52:35]: About 260 million m³, which is what you did. Stanford [00:52:37]: You used that kind of as a model for your modeling. Richard [00:52:39]: Right, right. Stanford [00:52:40]: So how fast does an event like that in your model reach population centers? Richard [00:52:46]: So that town I mentioned of ording, Washington, which is not necessarily the closest inhabited area, but it's the largest inhabited area that's relatively close. So it reaches ordinance in about 1 hour. And the model results there are not too different from what had been inferred previously, just based on kind of empirical correlations between lahar speeds and travel distances and so forth. Stanford [00:53:10]: Right. But that does make it one of the shortest travel time hazards, like flooding, except for a dam breach. In a very close community, we usually have more warning time. Richard [00:53:21]: Right. And there are other communities around Mount Rainier that are closer to the volcano in different drainages. They would have even less warning time. Stanford [00:53:30]: Okay, so I have a question about lahars that might be really naive, but it seems to me like we've talked about a solid phase and a liquid phase and debris flows. But ice would be a pretty interesting component because it starts out as rock and ends up as water. Is that a minor effect or is that something that really comes into play? Richard [00:53:49]: Ice, particularly in some of these large volcanic events, we really think ice can play a significant role. And in part that's because when the face of a volcano fails as a big landslide, as could happen on Mount Hood, and certainly did happen at Mount St. Helens in 1980, there might be an awful lot of glacier ice that gets entrained into that landslide and ground up, pulverized and so forth, which makes it easier to melt. We know for a fact at Mount St. Helens in 1980, that one of the reasons that a large lahar issued from the distal part of the great debris avalanche that occurred there down the toutal river valley, was the fact that there were chunks of glacier ice incorporated within that landslide, which melted and added to the water supply that was available to transform some of this debris into a lahar. Stanford [00:54:44]: Oh, wow. That's amazing. Maybe taking a step back to wrap up here, what are some of the really memorable events like this that you've seen in the field? You mentioned the one up in British Columbia, but are there others? Richard [00:54:55]: Well, the most memorable event I've seen in the field was one that I did not. Where I did not actually witness the event in real time, but got there shortly thereafter. And that was the Oso landslide up in northern Washington state in 2014. That was memorable on several accounts. One is that when my USGS colleague Jonathan got and I got to this site, there hadn't been any geologists there who were really landslide or debris flow specialists in advance of Jonathan and myself. The first thing was just the shocking amount of devastation that we saw, because that was the deadliest landslide in us history, outside of one that occurred in Puerto Rico back in the 1970s. And not only was it deadly, but the character of the event was such that really all of the remains of human civilization had all been pushed to the distal edge of the deposit, which is where we arrived. Stanford [00:55:50]: As if it was that snout. Richard [00:55:52]: Yeah. And the thing is that the leading edge of that event did act like a debris flow. Stanford [00:55:57]: Oh, wow. Richard [00:55:58]: That's something we learned after we studied it for some time. And so there was that sort of shocking evidence of just destruction of habitation. And some of it was heartbreaking in the sense that you'd walk out there and you had to walk along sort of gangplanks because there was quicksand everywhere. And if you stepped off, you'd drop right into the castle. Stanford [00:56:20]: Evidence of liquefaction. Richard [00:56:21]: Yeah, liquefaction process, absolutely. But, you know, you'd see things like teddy bears and kitchen utensils and mangled bicycle wheels and just all the things you associate with people just living their lives. And it was all just this tangled mess. And Jonathan, I thought, characterized it well. He said, it looks like the debris from a very violent tornado, and then you pour quicksand over the top of everything. That's what it looked like. So it was shocking in that respect. But then shocking in a scientific respect was when we made our way out to the distal end of the landslide. And I had my primitive little analog device, my Abni level with me, and I shot an angle from the distal end of the landslide up to the top of the headscarf, and it was six degrees. Oh, wow. At that time, we already had a kind of a rough idea of what the volume of the landslide was. We knew that it was somewhere around eight to 10 million, knew that for a landslide of that size, that this was sort of unprecedented. This run out angle was unprecedented. So the mobility as measured by that angle was extraordinarily high. Stanford [00:57:33]: The mobility was high because it had run out so much, which left a low angle. Richard [00:57:38]: Exactly. Our job there initially was not really to do science. It was to inform and assist with the recovery effort. I. But of course, it eventually did transition into more of a scientific effort. And to me, the outstanding problem there was explaining the mobility, which also was. Stanford [00:57:59]: Kind of what made it so deadly. Richard [00:58:01]: Yeah, absolutely. And it was also clear that a big part of the story was basal liquefaction. Now, exactly how much of the mass was liquefied is hard to say, but we do know for certain that parts of it were liquidity, because there were pools of quicksand standing around. There were sand boils, which are features that often form during earthquakes when liquefaction takes place. So anyway, we ended up using the d claw model to simulate this event, which helped us understand it a lot better. And then there was a lot more detailed field work done as well and so forth. But it was a really poignant event for me to experience that. And partly because it was in my home state, it wasn't like I was halfway around the world doing this. Stanford [00:58:43]: That's right, yeah. Richard Iverson, thank you so much for talking with us today. Richard [00:58:47]: I appreciate it. It's been fun. Stanford [00:58:52]: Someone once asked me, if you could do a different job, what would it be? I remember saying, I think I'd enjoy mycology. Fungus is a weird and wonderful world, or mud and debris flow seem almost magical. I'd really like to understand them better. I remember my friend looking at me weird and saying, isn't that second one. Stanford [00:59:09]: Pretty close to what you actually do? Richard [00:59:11]: But not really. Stanford [00:59:12]: These high concentration flows are their own weird and wonderful world. And in some ways, I'm thankful for this podcast, even if it just gave me an excuse to ask Doctor Iverson some of the things I've always wondered about these processes. If you've learned as much from this conversation as I did, my work might be done here. We are winding down on the first season and are already recording some great episodes for the second season, and that second season is taking on a different tone, more like this episode. I'm still interviewing my mentors, but I'm also reaching out to people I kind of wish were my mentors who I've always wanted to talk to about their work. So next season will include more professors, academics, and practitioners outside of the Corps of Engineers. If you have recommendations about who I should talk to or classic river mechanics papers that I should consider for a classic paper series we're working on, there is a Google forum on the podcast website where you can recommend either or both. There's a link to that podcast website in the notes, and we'll have lots of links there this week to Richard's papers and videos. I also need to recognize that this episode of the podcast was funded by the course post wildfire R and D program, which is led by Doctor Ian Floyd. Ian's the one who actually introduced me to Doctor Iverson's work many years ago. Thanks for that, Ian. These are informal conversations and the views expressed by the host and guest do nothing necessarily reflect the official positions or policies of the USAC, the USGS, or their partners. Mike Loreto edited this episode and basically the whole season and wrote the music. I'm Stanford Gibson, the sediment transport specialist at HEC. Thanks for tuning in.