 So we're going to move on now to the next talk, which will be given by Allison Pfeiffer from Western Washington University on modeling bed material abrasion at the basin scale. Over to you, Allison. Thanks, Greg, and thanks to all of you for spending the morning together. So I'm speaking to you from the native lands of the Lummi and Nooksack people, talking about work that I've been doing in the Suyadil River, which is the homeland of the Sock Suyadil tribe. So this work that I'm going to talk about today is a little bit of modeling and maybe more a detour into some field data that has convinced me that some of the classic approaches to modeling bed material weren't cutting it. All right, so before I start, I want to acknowledge the various collaborators on all these different elements of this ongoing project, as well as the funding sources that have made it happen. So bed material abrasion is the process of bed load reducing in size as it plunks along the river bed. And it's often considered through the lens of downstream finding downriver networks. But today I want to think about it in terms of how it controls the partitioning of basin scale sediment fluxes between coarse and fine material. So abrasion in this sense is an important piece of the puzzle in source to sink sediment dynamics. So how this landslide here will turn into either in a gravel bedded river, a morphologically benign plume of silt, or a problematic wave of aggregation that could increase flood hazards or change the channel morphology in ways that would be challenging for the communities that live alongside the river. So the standard approach to modeling abrasion goes back a ways, and it is exponential decay. And so we tend to say that the grain mass at some distance downstream x is a function of the original grain mass, the distance that grain has transported, and an abrasion parameter that relates to the rock strength. So a common way to implement this would be to take sediment of a given lithology out of some gravel bar and toss it into a rock tumbler or annular flume and calculate alpha. And then to assume that that alpha represents the abrasion rate for that lithology anywhere in the upstream basin. So this common approach fails me in two distinct ways. First off, the model is inadequate. And secondly, the way that we tend to gather the field data is too simplistic to characterize or capture the patterns that we end up seeing in the field. So we've come up with a method for rapidly assessing the abrasion potential of sediment. And we'll start here. So we can use Schmidt-Hemmer rock strength to estimate the tumbler-based abrasion rate, so this sort of classic way of characterizing the abrasion potential of sediment. And using the Schmidt-Hemmer is this huge benefit because it allows us to assess the abrasion potential of a particular grain in less than a minute, which is in contrast to taking sediment from the field, bringing it back to the lab, and then spending hours or days tumbling per sample. So we'll use this approach to think about the source-to-sink sediment dynamics and how abrasion plays a role in the Suyado River, which is in the North Cascades of Washington. So here we are just south of Canada. And the Suyado River drains the eastern flank of Glacier Peak, which is a stratovolcano, and heads towards the Puget Sound. So the Suyado River is notable in this region because it has exceptionally high sediment yields, the majority of which come from just from this little area on the eastern flank of the volcano. So let's zoom in there. Okay, so Chocolate Creek is the dominant source of sediment in this basin with Dusty Creek. It's downstream neighbor as a secondary source. So these two channels are rapidly incising into a volcanic apron of mid-holicine lahar deposits. And together, these two tributaries contribute upwards of 60% of the basin sediment load, just in, you know, there, despite their very small drainage areas. So we're gonna look downstream from Chocolate Creek, down the 65 kilometer length of the channel, at the rock strength of grains on the gravel bars. And what we see is that the rock strength of volcanic clasts increases rapidly over about 20 kilometers downstream from Chocolate Creek. And then the strength remains roughly constant for the subsequent distance. And this is mostly associated with a rapid loss of the weakest grains. We see, if we look in the trends in lithology, complementary trends. So the source material in Chocolate and Dusty Creek, that lahar, is made of a messy combination of volcanic rocks. Everything from Tefra airfoil deposits to lots of non-visicular, so no bubbles in the rock clasts. And so we'll sort of, for simplicity, we'll separate these volcanic rocks into these two categories. And looking at the distribution of source material, it is dominantly about 80% vesicular grains, and with a smaller proportion of non-visicular. And the vesicular clasts are lower strength, which you can imagine from those air bubbles trapped. So looking at the downstream trends in the sediment exposed on gravel bars, we see that above the chocolate fan, so above the main contribution of that volcanic sediment, maybe 40% of the bed is made up of a volcanic material. Downstream, we see a rapid increase in volcanic proportion. And most of that is vesicular volcanics, which makes sense based on what we know about the source material. Downstream of Dusty Creek, we see a rapid loss of volcanics, which is almost entirely a loss in the vesicular fraction. And over that distance, there's an increase, a slight increase in the non-visicular fraction. And then this remains roughly consistent downstream, maybe until some interesting things going on in the lower kilometers. So what we're going to do with these field observations of rapid loss of volcanics over maybe a 10 to 15 kilometer length scale is to try to model them. And so we'll start out with a naive approach. So we'll say, well, what if we had driven to the one road bridge that crosses the Suyadil River and measured the mean volcanic Schmitt-Hemmer rock strength at that site? Well, if we had done that, we would calculate that that abrasion is essentially negligible in this system, which would be very nice because then we could not have to deal with that problem in our models of source-dissic sediment dynamics. We could use remote sensing to estimate the size of mass wasting events or sediment contributions in the uplands and assume that that sediment works its way downstream and is sorted by size selective transport. Well, maybe a more thorough approach. Let's say we might have taken a couple day hike up the river and measured the mean Schmitt-Hemmer rock strength of that source material and used that in our simple Sternberg equation. Well, calculating this, we'd find a slightly greater degree of abrasion, but abrasion still appears to be a pretty minor effect. We might note, though, that experiments have suggested that tumblers systematically underestimate abrasion by a factor of maybe two to four. And so if we accounted for this, we greatly increase the total downstream abrasion. But abrasion is happening quite gradually over the length of that channel. So going yet a step further, we might use the distribution rather than just the means of Schmitt-Hemmer rock strength. And of course, because there's this wide distribution of rock strength in these volcanic clasts, we get a rapid loss of the weakest clasts. So we do have some more rapid abrasion when we separate it out. And now we're distinguishing between these two populations of grains. But problematically, the rate of volumetric loss of this bed material is still quite gradual relative to what we see in the field data. So this brings me to transport-dependent abrasion. So Macauletal and Jerome Levé published in 2009 some flume experiments that showed that abrasion measured per kilometer is enhanced at higher sediment transport rates. And so they observed that, well, at the lower transport rates, abrasion is maybe constant for a given lithology, but then certainly increases after about three times the threshold for motion. And so what I've done is I've said, well, let's take sort of a simple version of their observations and model it. And so let's say that any grain has a baseline abrasion rate. That's a we can estimate using those Schmitt-Hemmer measurements in the curve I showed earlier. And, you know, we'll put a tumbler correction factor on there. And then above three times or 3.3 times that the critical shield stress will increase the abrasion rate. So what I've done here is I've said that we're going to start with a population of grains that represent the grain size distribution, the Schmitt-Hemmer rock strength distribution, the density distribution, which is also a function of Schmitt-Hemmer rock strength of that source material. And then we'll numerically abrade them going downstream. And we'll, for simplicity, assume that we're in sort of an equilibrium state where we continually have material added from upstream. And we're just thinking about the mass lost to abrasion or volume lost to abrasion heading downstream. And I'm happy to talk about how I dealt with the shear stress at each link going downstream if people are curious. Because I'm not doing a sort of dynamical model where the bed material grain size distribution is changing through time or changing with the evolution of these grains. So what we get sort of playing with this transport dependent abrasion really depends on how high we tune that transport dependence. And the flume experiments don't give us particularly strong constraints on how intense this transport dependent effect is. But either a weekly or a highly, I'll flip back through those, a weekly or a highly transport dependent abrasion effect does give us that rapid loss of the weakest grains in those first 10 to 15 kilometers downstream from the source. It also results in downstream strengthening of the grains in the bed surface. And so this is to say that transport dependent abrasion seems to be one way to replicate some of the patterns that we're seeing in the field. So the takeaway. While a simplified approach to characterizing abrasion is tempting, sediment heterogeneity and transport dependent abrasion can be important controls on the downstream fate of course sediment in fluvial systems. And there are a couple of caveats here. And one of them is that, I've shown you a site where sediment heterogeneity is substantial on the flank of a strata volcano where you have both andesite flow material and airfall deposits that are mixing and are mapped as single units on the geologic maps. But I think that the sediment heterogeneity problem is probably much more widespread than just volcanoes because sedimentary rocks can be exceptionally heterogeneous within a single mapped unit. And as our metamorphic melanges. And so these are many of the landscapes that are that are rapidly eroding and that have these mass wasting events that from sort of a societal perspective were particularly interested in being able to model using source a source to sync framework. And so I would I would argue that you know this heterogeneity problem can probably be ignored in many places. But many of the places that were interested most interested in I think it probably can't be. And the second caveat here is that I've modeled transport dependent abrasion but I think that there's a second sort of complicating effect in in the abrasion process that I haven't accounted for yet. And that's in place abrasion. So an immobile grain can get smaller without ever moving downstream just as smaller grains come and hit it as they transport along. And so this in place abrasion effect was you know described by Shemin Stevens in the 70s and I think Jeff Pransovic is doing some work to maybe get us closer to having sort of the numerical tools to tackle it. So where where do we take this? I am headed towards a model for basin scale sediment routing that's that's where I want to to put this abrasion work. So the network sediment transporter is a new land lab component that was just published last fall. That's work by Katie Barnhart, John Chuba, Eric Cutten and I. And what this new component does is we can take a a river network represented as as a land lab network model grid and model parcels of sediment and in a Lagrangian framework transport them downstream and abrade them and and sort them into a surface and subsurface distribution. So each one of these parcels is given a baseline abrasion rate density a starting volume a starting grain size and a starting location on the grid. And so then via sediment transport equations we can evolve the channel bed elevations in the bed size in the bed grain size distribution. So this is the sort of framework that it'll take for us to explore in combination abrasion and size selective transport with some of these elements of sophistication. The network sediment transporter can't yet do transport dependent or in place abrasion but coming soon. So this is a tool that we can use then to explore the morphodynamic response of the channel bed to sediment pulses and when I say we I don't mean me I mean all of us. So this component is live in land lab but along with tutorials to help you get started using it. But it is also very much a work in progress. So this group of people along with anybody else who wants to join in is working together to create the really wide variety of utilities that will make a tool like the network sediment transporter truly useful for both sort of the intellectual curiosities and hopefully some of the more practical questions related to source to sink sediment dynamics. So with that I'll leave you with this thought that modeling abrasion requires quantifying heterogeneity and input parameters and a greater complexity in the modeled processes. And if there's time I'll take a question or two. Fantastic thank you Allison. So let's see so I see a question from Risa Mattoff in the chat. She writes very interesting discussion Allison would it help considering how the fines facilitate or limit transport of larger clasts. That is a great question. So to a small extent we've done that with sand. So we're using the Wilcock and Crow bed load transport equations which do take into account the sand content of the bed. But John Chuba John Chuba's PhD work is really what the network sediment transporter is based on. And some of his explorations have suggested that the way that it ends up playing out we don't actually keep as much sand on the bed as we observe in our field measurements of bed surface grain size distributions. So I think that that's probably something that we need to explore more making sure that both our equations can handle the sand content and that our model can retain enough sand to actually match what we see in the field. Other questions for Allison you can either post in the chat or just do the raise hand. Okay here's one from Vanessa Gable very cool talk Allison I'm curious about what causes the jump in transport rate that leads to transport dependent abrasion. Does it occur after some critical amount of finding has already occurred? That's also a good question. So the transport rates are high from the get-go going downstream and that's actually the the sort of I mean we start with this really wide grain size distribution everything you can see this is the source material in my background photo here everything from you know two meter boulders down to silt though we neglect the fines in this in the model and the wide grain size distribution contributes to really high transport rates but I think maybe even more importantly the really wide distribution of density. So we have occasionally at this site you'll find a grain that's pumice and it will float. We have lots of grains that are you know 16 or 1800 kilograms per meter cube so much much lighter than our standard assumption of of quartz for granite density. So that I think is what's responsible for the the transport rates that are much higher than I think many of us would be used to seeing in our standard sorts of models.