 All right, so good morning. It is my absolute pleasure to be here today and to be meeting off this excellent workshop. I'm gonna start with a brief but hurry thank you to the organizing committee for bringing us all together in this way, this is really exciting. And of course, the National Science Foundation for funding. So I'm going to, oh let's see, I need to be dimmed. I don't know, can I see this? So I'm going to eventually today talk about an example of a coupled tectonics and surface processes project related to M9 earthquakes in the Cascadia subduction zone and the landscapes response to those earthquakes. But before I get to that, I wanna do a little bit of setting the stage. So, oh, now you can hear me, or the mic seems to be completely set. So I wanna do a little bit of setting the stage before I go off on that example. And I wanna do this so that I can really put in perspective, as Greg mentioned, my vision or the community's vision about sort of big questions and progress and challenges and coupled tectonics and surface processes studies and future. I've really been looking forward to this meeting actually since years ago, when it was first brought up by Mark at a systems meeting one spring as a possibility. And my enthusiasm for this workshop waned only just a little, a few weeks back when I got an email from the convener saying, by the way, you're first. You're gonna set out the meeting for us. And wouldn't it be great if you touched on what you think are the big, important questions and the progress and challenges which felt like a really tall order. But the good news for me and the good news for us is that our communities have actually been engaging the deep dive on these very questions over the last year plus. And so I'm gonna highlight some things from two initiatives that many of you in this room have been a part of and many people watching online would have been a part of. And those initiatives are the challenges and opportunities for research and tectonics and the SC4D initiative to try and understand the processes that underlies the question. Both of these initiatives have recently presented NSF vision documents which you can find online. And I'm going, promise I'm not gonna read I'm not gonna read through this. I'm not gonna spend my time going through all these documents but I certainly encourage you to look at them because they're really well written. But in both of these documents, they outline grand challenges with big questions that the community is trying to make big progress or is poised to make big progress in over the next generation. And I'm very pleased to see that in both documents there's a highlight of the themes of understanding the dynamic interactions among other surface processes. So we know this is important and the community recognizes this as well. And so I think the timing of this workshop couldn't be any better because I think there's a lot of momentum in the community and we really wanna capitalize on that. And so it's very exciting at this time. So I'm going to just highlight a few things that really resonated with me from these documents to sort of trip off our conversation today. I mean, the first thing that everyone in this room will be pleased to see is that the communities agree that it's about the importance and in fact the necessity of models in our efforts to try and understand service processes and electronics. And we wanna link deep models with surface processes models. And we wanna think about how these models will operate both over short-term sort of earthquake hazards timescales and link those into longer-term mountain building timescales. And that's gonna be one of the things that I talk about in the example that I give you today. Couple other things, I'm very pleased to see that our community seem to really value interdisciplinary centers and open-source community resources and organizations. Things like the CIG and the CSDMS. And we hope that we'll see continued funding for those in the future and in fact, even more of them. Along that same vein, there was a lot of emphasis in these documents that progress going forward will happen only when we have collaborative sharing data equipment, technology, labs, even sharing of things like training our next generation users of frontier techniques. So I really like things like CIDR or the NSAID summer schools, so that graduates aren't just trained in a single lab with a single PI, but instead trained by the community. And there's just one last point here that I wanna emphasize. This wasn't really written in any of these documents explicitly, but I think it's very important. But I think it's really useful to identify, tackle and strategize these research problems that we have collective interest in together from the very beginning. So I think it's really common for people to collaborate, but they often do this in sort of a siloed fashion. So you might be on a big project where you're working with a geophysicist and you're a geomorphologist. You kind of, you get it funded together, but maybe you work on it separately and then later on you sort of come back together and say, this is what I found, what did you find? And you try and kind of glom those things together. And that's useful, but I think it's even better to start from scratch, sort of from the beginning with a blank canvas and say, okay, what are our collective interests? What are, how do we wanna strategize and tackle these problems together from the beginning? So I think we have an opportunity to do that over the next coming days and I'm really looking forward to that and I hope that this is one of many of those opportunities. I hope this workshop that we're doing today doesn't happen in isolation and I hope we continue. So now I'm gonna give an example of a project that I think kind of hits on a lot of those points that I just mentioned. And this is the UW so-called M9 project. M9 here stands for magnitude nine. And this project is a really large project. It was generously funded by NSF, the Seas Hazards Initiative. This is a $3 million project led by University of Washington researchers in collaboration with USGS scientists. You can see there's lots of names here. This list could be even longer. And the idea, the main thrust of this project is that we're simulating in 3D magnitude nine earthquakes on the Cascadia mega-thrust. We're considering all of the consequences that come from those events. This is a truly interdisciplinary project. Many of the names listed here are people outside of the geosciences. So in this case, we're working across departments, across the university. Social scientists are working with engineers, are working with geologists, are working with urban planners, applied mathematicians and statisticians to tackle this problem. Before I started working on this, my view of interdisciplinary was when I walked to the fourth floor to talk to geochemists for the second floor to talk to space physicists. And this has really sort of brought in my horizons on what an interdisciplinary project really looks like. So before I tell you the details of this project, I wanna make sure that we're all on the same page with a little background about Cascadia. Not everybody lives there and thinks about it all the time. So the Cascadia subduction zone, which I'm showing an image of here. You can see the Juan de Fuca plate is subducting beneath the North American plate. And we know that the Cascadia subduction zone has a history of M9 earthquakes. And we know this from records of coastal subsidence, tsunami records and offshore turbidites. We also know that the last Cascadia earthquake happened in 1780, January of 1780, and it was estimated to be between magnitude 8.7 and 9. And we also know that we can estimate the probability of a 10 to 14% chance of another M9 earthquake happening in the next few years. So we need to think about this, both from a scientific perspective and from a hazard perspective. All right, so this is a nice little graphic that really showcases the project and its many different pieces. So the overall goal of our project is to reduce the catastrophic potential of Cascadia earthquakes through advancements in hazard assessment and adaptive planning. So I'm showing this little nice cartoon by a graduate student named Nasa Murafi in civil engineering that the real sort of, oh, actually before I show you this. This, I wanna just show these are the names of the departments of people that are involved in this project. And also I wanna say that we're not just working on this within the university, that this project has many stakeholders outside of the university that have been vested and involved in the project, including the state of Washington, state of Oregon, national agencies like FEMA, and then local consulting firms in the city of Seattle. So it's been very, very broad in terms of those who are interested in the outcome of this project. All right, so as I mentioned that the sort of heart of this project are the 3D seismic simulations that are being carried out by Art Frankel and Aaron Wirth for USGS Sciences in Berkeley, Washington. And what they're doing is they're simulating M9 earthquakes and from those simulations, get broadband synthetic seismograms. And these simulations are really sophisticated so that they accurately capture really important things like rupture, direct to the base and amplification, edge converted waves and long duration, which are all things that are gonna happen. And then from these simulations, there's a whole bunch of us who are working downstream that wanna take that output and think about what it is that that kind of ground shaking or strong ground motion will do to things like building an infrastructure and in my case, to the hillslopes. How do the hillslopes react to that? And from the scientific output, we can do things like work on early warning and help the community plan for enhanced earthquakes and work on integrated risk maps. So I don't have time today to speak about each of these things. So I'm really gonna focus my attention on what we can finding with respect to the landslide but I'm certainly happy to entertain any questions that you have during the pushing time or over chat during a break about any of the other aspects of this project. So as you can imagine, when you shake the ground as hard as you do in a magnitude nine event, especially one that reaches, say, from Northern California all the way to Vancouver, by terrifying to think about, you're going to have a really strong landscape response. So there are a group of us, myself, my colleague, Joe Wortman, and our graduate students from the College Grant who've been working on the landscape response that we might expect to see. So before I show you the results from that work, I'm gonna spend just a few more minutes talking about what these simulations are and give you a little, show you a little example movie, hopefully it'll run of what a simulation is. All right, so Art and Erin have got 50 plus separate scenarios of M9 events. So this is really exciting because prior to this, you usually focused on one single sort of worst-case scenario. But in doing 50 different scenarios, we can do things like actually have a probabilistic view of what the ground shaking is. And what we're after is really understanding what the range of possible ground shaking might be from an M9 event. Also, what are the critical ruptures for it? And so in doing this, we varied a lot of different parameters and what we found is that the strength of ground shaking can roast on things like a hypocenter. So that's the starting position of the earthquakes and we varied that from central, Northern, or Southern portions of the megathrust. We also know that it's very important where the actual rupture initiates so the inland or eastward extent of the rupture is quite important. And finally, one of the most important things, especially at the high frequencies that affect the shaking of the hill slopes, are the location of strong, ground motion generating areas or sticky patches. You might know these as sub-events or high stress drop as far as along the seduction zone. And we know from examples like the Hoku in Japan that these things end up being the most important elements for stuttering strong ground. So when we vary where those sticky patches are, we get a range of different shaking. So the main takeaway is that there are a wide range of possible outcomes that we consider when thinking about strong ground motion for M9 events. And I also wanna point out that, so these, there's two companion papers that should hopefully be out soon in BSSA or you can read about this work. But also we have put all 50 of these simulations up on the design-safe site so that we can share this data with the public in that way. So go ahead over to design-safe in an actual output. So now I wanna show you an example movie of the output. So you can see the hypocenter is just there, got an M9 earthquake happening. And these are four seismic stations and you start to see what the seismic waves look like coming into those stations. And they just wanna point out a few things. So this is just one of those 50s simulations. And one thing I wanna point out is that as we would expect, see that the ground shaking or the seismic waves are quite intense at Crescent City, which is close to the shoreline in a sense. I also wanna point out what looks really intense right here, which is Seattle. And that's because Seattle sits in a basin. And so one of the real steps forward in this work was including things like basin effects. You can see that it has a pretty profound impact on the seismic waves. It amplifies those seismic waves. And you can see the same is happening important to a lesser extent. This is an example of a station that's outside of the basin. So the city of Seattle was quite interested in this result because this has big effects on, for example, the tall buildings in the city, which have not been coded to things like basin effects. It was actually a recent meeting between our work on N9 and we went to see the city of Seattle. And we met on the 67th floor of the tallest building in Seattle to tell them this, which seems to be funny. But anyway, so it turns out one of the biggest things we've learned is that the basin effects. Okay, so now I wanna show you what we can do with output like this in particular when we're imagining things like the land. All right, so this is a picture, not from Cascadia, but instead from New Zealand. That was just taken about a month ago by my graduate student. And this is an image of the Seaward Kikore Range, which is effected as you see by the 2016 Kikore Earthquake in the wind. See, there's a lot of still very fresh looking landscape, or sorry, landslide scars across this mountain range. So we know that when we have earthquakes, it's traumatic and there are many thousands of landslides that get triggered from these events. And so what we wanna do is try and predict what that might look like for an N9 event. So I'm going to show a little bit of work by Alex Grant, he's now at the USGS. And what he did is he took the output of simulations and he used that in a new mark analysis to try and predict co-seismic landslide displacement from the modeled strong ground. And then my graduate student, Sean LaHusen, he's working on an inventory, something to map and date and look for large spatial patterns in Cascadia co-seismic landslides. That's any landslides maybe from 1700 or possibly from earlier events. You can imagine that this is pretty difficult because unlike that Kikore situation or like Nepal, which today is actually the anniversary of the Gorka earthquake, another place where we saw massive landsliding, the hillslopes, unlike those events where you might have a map from before and then a landslide map from before, then you know the event happened, then you can go right out there and you can map the landslides and you can do that kind of before and after. We don't have that luxury in Cascadia, right? The last event happened in 1700 and we have a lot of landslides that have been happened since that have nothing to do with subduction. So it is a little bit difficult and I'll talk a little bit more about that. Okay, so first let's look at this model output. So this is a map showing peak ground acceleration and this is the median peak ground acceleration of peak model simulation. You can see the warmer colors are where you would expect them to be that are closer to the trench and as you get away from the trench, you have a less intensity of ground change. And so even though Seattle, which is in that little blue box there, is not right on the coast. Alex decided to start his analysis there for a couple reasons. One, you can see that even though it's not as dramatic as the shaking at the coast, there's still 0.2G that would be affected or would happen in Seattle and anywhere from 0.1 to 0.24 is the range depending on things like directivity and the location of the sub-event. But the thing about Seattle is that we're notorious for having unstable hills, right? So we know that the landscape here is quite vulnerable. There are plenty of slopes. The material is very hard to find bedrock in Seattle. It's mostly very weak glacial deposits, including very porous sandstones that sit on non-porous clay. Makes for a really dangerous combination. And we also know that this is a population center. Obviously, it's also where the University of Washington is. So it made sense for us to start our analysis there. What I'm showing here is a map where there's color-coded dots of landslides, historical landslides, but there are quite a bit of them. There are many of them. They tend to be focused along the blocks of the ground or along the drumlands. You can see here basically anywhere where there's steep slopes or the potential interaction with water. All right, so here's the recipe of what Alex is doing. So you find yourself a place. So in this case, we've got Seattle. You have some sort of gridded slope data that you also need information about material strength and ground saturation. And then you need to couple that with some equations that you use for different landslides. And then from that, what you can do is a homework analysis or a hazard model where we consider the probability of co-size like block displacements, even the shaking intensities that are coming out of our seismic simulations. We're just a little bit more about the modes of failure that we chose. So in this case, we decided to have two modes of failure, shallow translational slides and rotational slides. And from those, you can just have an equation for a simple factor of safety, sort of is it failing or are we expected to failure to be expected to be stable? But here are some equations that we use for that. And what we're really interested in this case though is how close are we to failure, right? So you might have a whole bunch of cells that are stable right now, but the real question that we're asking is, what's gonna happen when they are exposed to strong ground motion? So what we can do is we can consider what the yield acceleration is. What's the acceleration above which down slow motion will occur, right? So if factor of safety one is sort of right on that edge of safe, what kind of shaking do we need to trip this into instability? And so here I'm showing that yield acceleration versus the ground acceleration and what you can see is that if the slope can be considered strong relative to ground shaking, if your yield acceleration is much greater than the PGA, but in cases where the PGA is much greater than AY, then your slope is weak and it's going to fail. So Alex has been busy generating a bunch of maps from that output. And he's done this for both dry summer conditions and wet winter conditions. And if you're looking at this, trying to figure out where the colors are, if you're not blind, it's actually really hard to see because they're here but they're really focused in the, well, they're cell by cell and they're focused in just a few clustered areas. So I'm going to try and make this a little bit better by zooming in on West Seattle, which is a place that has a lot of landslides. Hopefully you can still see, I'll use the pointer to try and connect your eye a little bit here. So here the shallow translational slides are the pinks and the deep rotational slides are the blues. What you can see is that during the summer, this area here has quite a bit of shallow translational slides, but you're not seeing a lot of deep rotational slides. And then though once we introduce water or winter conditions, you can see that there are the addition of many more deep rotational landslides. What he's found is that he sees a 515% increase in areas of greater than 5% predicted probability of deep rotational landslides from dry to wet. So I think one of the main take-homes here is that we hope that the next magnitude nine event happens in the summer, because it'll make a really big difference to Seattle. And I'm not going to talk about it here, but Alex has also done a lot of really neat work where he compares an M9 event versus a Seattle Falls event, which is a fault local to our city, versus a Nisqually type event, which is an earthquake on the down going plate. And what he finds is that even though the M9 event is big and scary and dramatic, it's actually much worse when we have a Cascadia Falls event because it's so much more vocal. And in all cases, we wanted to happen in the summer, every time. Okay, so that's what, those are some preliminaries to do as well as out of Alex's work. He recognized that this is far from the places where people with table there are many of them and we're really interested in doing next, taking that same kind of analysis and applying it to places like the Olympic mountains and the coast ranges, places where we have high topography and are much closer to each other. So that's where we're headed with this next. But we don't wanna just model it and make predictions. We still, I mentioned that it's tough because the last earthquake happened so long ago, but we're not giving up on trying to find evidence for previous Co-Size McLanslades in the landscape itself. And I think that's really important to compare to our models and to see an iterative process. And so I'm gonna just tell you a little bit about some work that we've been doing to try and get a better sense of where the M9 Co-Size McLanslade is. And this question is kind of rhetorical in some ways because you might be surprised to know that there is not one single definitively dated 1,700 landslades. We know they must be out there. It's difficult, we could talk later, I don't wanna get off on too much of a tangent, but why it is that those may not have been identified yet. But the point is that we don't yet know where are the M9 Co-Size McLanslades. And I think it's important to note that we, we're not just looking for one, right? We really want to have some information about the spatial patterns, the spatial and temporal patterns of what the Co-Size McLanslades is generated. So we wanna have information about not just one or two, but many hundreds or thousands of landslades. So I'm gonna just take a moment to tell you a little bit about how we might get an inventory of dated landslides at that time. And so to tell you about that, I need to just do a little sidebar. I'll tell you about a project near to the Oso landslide. Any of you may have heard of the March 22nd 2014 Oso landslide because it was one of the deadliest landslides in the United States history. Three people were killed on that day by this long run out landslide here. It was really dramatic. So I'm going to share with you some work done in collaboration with Adam Booth at Portland State and my colleague Dave Montemarie at Houston, Washington and of course, Sean and Dennis. And here I'm showing you a Google Earth image of the North Forge still at Guamish River Valley, which is a river valley that's an Oso landslide on the soil of Oso. So if I gave you a few seconds, the best geomorphologist in the room might have a way to sort of kick out the rest of what the rest of the hillslides look like in the image, but it's pretty hard to do, right? I mean, this is really obvious. The river's pretty obvious. But looking at the rest of the landscape, it's pretty hard to see what else might be going on. And this is the lidar of the same exact place. And I can't really think of much better examples of utility of lidar data than in trying to map landslides into the Pacific Ocean. So when you see it like this, it's startlingly clear that the Oso landslide, I'll just point out here. So here's a map of the many other landslides that are in this community. The Oso landslide happened right here. This is the precursor landslide to the Oso landslide. And the community of Steelhead Haven, where most of those people were, so everybody except for Jesus, I wait here at the very moment that this landslide came across the road. They were all in this community. They knew about this landslide, but they thought they were safe over here because they didn't ever. But when you look at an image like this, what you see is that there are plenty of other examples of huge, big landslides that haven't backed on it. So a really important question when thinking about things like risk, how old are these other landslides? Do they happen a really long time ago and are in an issue or are these things happening? And so in order to address that, we have to have some introduction. So how do we get timing on landslides? I really love this picture because it's such a classic advisor scene, right? An advisor hovering over a student while doing nothing, but while a student does everything. So what Sean is doing here is he's digging in a landslide deposit that has a small exposure from a little river gully. He's looking for wood that's in trees and deposits. Wood that would have been killed, trees that would have been killed in the act of the landslide that could provide some sort of timing on. And so you might think, okay, great, well, you can just go around and you can use carbon 14 to pay all the landslides. And the truth of that is you can do that in some lucky situations, right? So even if you had all the money in the world, you're never gonna have access to all. Pretty reticent to let you know and it's not always easy to get there. But even more than cost or say one of the biggest issues that we're talking about trying thousands of times, this is just not reasonable. And it's really not reasonable because you can't always find wood. I mean, if I showed you a picture of the Oso landslide right after the event up close, you would see that parts of the slide exposure are just chock-a-buck full of huge trees everywhere. But then there are other parts of the landslide that have no trees at all. So now you fast forward hundreds or thousands of years and try finding that needle in a haystack. Try finding that one. You have to have an exposure to it and you have to. So we need something else. We can't just leave it. So the something else is actually locked in the landscape itself. So if we're hundreds of years, geomorphologists and geologists have known that landslide deposits, so this is just a little cartoon showing you that when you have a landslide that really roughens up the landscape. You have all kinds of things like hammocks back to the scarps and tension cracks and all kinds of things that roughen the landslide. But as time goes on, things go over. Eventually, rivers take back hold. And then really, toward the end, you actually roughen it again just a little bit by having the red smooth up. And so the good news is that we can use this as some information about timing. People have done this for a very long time. It's a relative. But we'd like to do more than just let this be a relative. And we can do that because we know something now about how it is that the landslides move in time, right? It doesn't happen randomly. It happens by equations like this. So this is nonlinear coastal climate plus. But we can make some predictions as to how that's going to go in time. This is a model that I'm showing here and this is a model of landslides that plays this equation here. And what you see is that they're at the predicted curve. And this is really valuable because if we have just a few pieces of information about absolute age from those two spaces, we can look at the landslide. Then we can calibrate a surface roughness age curve as long as we have lidar where we can get a surface roughness measurement from our other landslide. And we can estimate their absolute age to allow us to get at least some sense of timing for hundreds or thousands. So when we did that in this area, see these landslides are colored by age bands. What we see is that there's quite a range. All of these former landslides are not all ancient. We remember quite young. You can see that we can do things like get a recurrence measurement. It's just on average this one big valley long runout landslide about everything. And then in a companion piece led by Adam Boos, he expanded and mapped over 200 landslides up and down this valley. From this data set, we can really study things like look for changes in space and time related to things like climate change or the onset of an earth. So it's really valuable to put timing on data sets this large. Okay, so the idea is we wanna take what we've learned from the North Fork Siliglamas Valley. We wanna try and apply it to test ADS projections done. And we wanna look for spatial and temporal clusters of timing of landslides. And so I'm focusing in here on Oregon, a particular part of Oregon that has a tie-up region. I'm doing that because we can't just speak what we've learned from the North Fork Siliglamas from here. We have to recalibrate our curve because the rocks in that valley, they weren't rocks at all. They were actually glacial sediment. So if we wanna try and apply this to the bedrock of the coastal ranges, then we need to recalibrate. And we're focusing in here on the Oriental Oriental Coast Range because not only is it not either, but we know from great work by Josh Warring and others that there are plenty of deep-seated landslides in this place. So this is a good place to focus on trying to look for it. Some that may have been generated by this place. So in order to do this, we need to have mapped landslides. I can't just see any old mapped landslides, so we'll really just need to just map the deposit itself as well we're interested in. So Sean and two very industrious undergraduate students, Kyle Lowry and Valerie Bright have been mapping these landslides. This should probably be 4,000 by now and this box has probably been generated. But the idea is that we wanna map out all the different landslides in this area. We really only want the landslide body itself for our surface roughness. And so this is just one half of it though, right? So now that we've got this, we can start working on the surface roughness, but we need to calibrate with the community. So the good news is we're not the only people interested, obviously, in looking for those landslides timing. Josh Waring and his graduate students, who will have a really nice excellent set of different places where they've been working to get very high-resonation timing on some landslides. And in particular, they and their colleagues at Dagami and with Brian Flack have been doing a combination of dendrochronology along with carbon-14 dating. And in places like this, where you have a large deep-sea immensely that does things like block a river, you can get a lot of really good information out of a location like this. So you can get carbon-14 out of the Stumpster landslide deposit like I choose before, and even in the dam lake. But then you can add to that things like coring of the trees, well trees, and you can do dendrochronology. If you want to really pinpoint something like 1700s, there's gonna have to be something like this. Carbon-14 by itself is too large to be pointing at though. But so anyway, the idea is that in the next months or a year we'll hopefully have more points that we can use to help calibrate our model. This is just a few that we have going, but we'll need a lot more than this to get that rough and smooth. So I just have a few last slides here to sort of bring this home from something that I've been talking about so far, which has been sort of coupling surface processes and tectonics over just this single earthquake event to imagining what this does over the long term. So I think that I'm really excited to see what we find about how the coastal subspond to a magnitude nine event. But I'm really tantalized by the idea of imagining how that affects landscape evolution. When you have thousands of N9 events over a month. So what I'd really like to see is a model or models that we couple that help us to stretch across those timespales. And one of the reasons why I think that it could be a profound impact on the landscape is because we know from studies like this, this is Brian Enrightus's work, he said one, that rivers react to larger earthquakes. And they do this because we dump a lot of sediment into the river, it has a strong effect on rivers. So I'm really excited about the possibility of imagining how different amount range like the Olympics would look, for example, if you had an earthquake that dumped material from hillslopes into rivers every 500 years on a regular basis versus a place that this A-sized magnet doesn't do that. And so if we wanna do that, I think what we need to do is we need to have a surface processes model that has the seismic cycle in it. And we need to make sure that we don't just have rivers used and most of our landscape evolution models have rivers but they don't all have a treatment, a specific treatment of landslides. But for this, what I'm proposing, we would need that. And so the good news is that there's been a lot of recent progress on this front. So a nice recent work with LandLab of shallow landslide probability. It's a great work by Adam Booth and others with a general deep-seated landslide model. So I think we have something we can work with here but what we wanna do is we wanna add in the co-seismic cycle to this. And of course I haven't forgotten the geodynamics tectonics piece, it's huge in this. The landscape is not operating in aesthetically. So what we need to do is we need a couple, that kind of model to something which is generating the topography in the first place. And from an outsider's perspective, it seems like there's been a lot of really great progress on modeling subduction zones and particular things like even paying attention to how topography develops even at the initiation of subduction and complicated oblique subduction systems. And I'm really delighted to see and I hope we hear more about this at this meeting that there's even progress in connecting the seismic cycle to the long-term topographic elephant evolution at Cambridge Marriott. So I think we have all the pieces in place to really start talking about how to link those. And I'm really excited about the possibility of having models that connect us from short earthquake timescales to long-time scales. And so with that, I'm going to say thank you and hopefully to see some great discussion. Terrific, thanks so much Allison. So we have some time for questions and discussion, just a reminder to ask your questions into a mic. I've got one I can hand you or there's one back here so that the folks who are dialing in can listen. Thanks for a great talk. I wonder as to the first part when you use the scenario rupture computations to infer a likelihood of landslide. When you compare the numerical assembler computation with what you get from a simple attenuation relationship given just the distance from the fault, I wonder how much additional information really get because it seems like the scales of the landslides are sort of mismatched to the presumed resolution of the rupture propagation computations which depend on the BS30 and the regional sites. And specifically, I mean, you were far out in terms of where you were in Seattle where this might be even a bigger concern but even closer to the rupture because then you're really controlled by the details of what happened in terms of slip on the rupture that I'm not, I just wonder how much additional deterministic information can you gain beyond an attenuation relationship? So that's a really good question and a really good point. I would say that, I think the biggest issue that we have is what you're pointing out which is that in the case of the high frequency motion which is what we care about. So for landslides what matters is above three curts. So that's the, that part of these simulations was generated entirely stochastically from the high stress drop disparities for the locations of the sub-events. So where those sub-events are is absolutely controlling the TDA. The problem with that, we have superimposed, we've changed around four possibilities of where they might be. We don't actually know where they were in the past or where they'll be in the future. And so it's a little tricky with the models because most of our ground x shaking is really tied to that. So the seismologists always look to us and say, well, if you map out the landslides in great enough detail, maybe that is how we can figure out the Helio sub-events. So it's a little bit circular that we're trying to use those sub-events to predict the ground shaking because the reality where they are is really good. So I think your point is a good one. Not really sure the way around it other than try and find the landslides in real life to help us. So that was really interesting to try to put the landscape development framework and linking it to the seismic cycle time scale. I was also wondering about, so you talked about co-seismic. Do you think there could be any fingerprint of the co-seismic readjustment in a landscape because it's kind of a transient fast episode of deformation? Yeah, so that's a really good point. And I don't know, I think maybe Josh could speak to this, but it's a small numbers problem so far. But a lot of the places where we think we get close to finding 1700, we actually find landslides that are like 1750 years, 1730 years. And it's like, maybe that's a coincidence or maybe it's not. There's a time scale where we've shaken and disrupted the landscape and there's going to be some transients associated with that. So I think that's really interesting. And I really think the modeling could really help us with that. So we can sort of watch it go and see those feedbacks. So I just wanted to ask how you distinguish between say a landslide that's not related to the seismic cycle that can happen in most, that happen in tropical regions, the drain, et cetera. So in your analysis. Yes. So I thought that would be the first question. It usually is. There are so many landslides that happen. We don't have a loss of landslides. We know that we have them all the time and regularly. So this is an issue. In a model that we set up, we could have a rule for precipitation driven landslides and then sort of shake the system every 300 or 500 years. That's pretty cool. But when we're looking in the real landscape, we have this kind of data set where we're looking far and wide. I think our best hope aside from projects where we're actually fingerprinting and dating 1700, that's great, then you can really say. But if we're gonna look over sort of the whole seismic area and try and have broad pictures, I think the best thing we can do is we can hope for spatial clusters and we can use tricks like, we know that the topographic amplification, for example, can trigger landslides in different places. So there is information out there that I think can help us to tease between the two. But one of the reasons I think we've struggled to find whose seismic landslides from 1700 is because there's an overprint from so many landslides that have nothing to do with it. The other reason why it gets back to the first question, which is that this, the trench is offshore, right? And so the work that we've been doing with crustal earthquakes tells us that co-sized landslides tend to be most clustered and concentrated to the fault itself, right? In fact, a really nice new paper just came out from Kaikoura that shows that very same thing. Our fault is offshore, right? And so if the topography, if sea level changed and there were mountains there, I'm sure that we would see many, many more. I really like what you're doing right at the end. So it's a different kind of question, or maybe it's more of a, were you reaching to the, thinking about what do we need to really couple surface processes with deformation? And so of course I like the idea of developing models and the landscape models that have the hill slopes in them. So you can look at the landscape response, shaking. And then of course you need to have a better river models. You know how the river responds, all that sediment in there. But before we go chasing into building more and more complicated models, I think we have to step back and ask, how is the coupling gonna happen? And so imagine you compare two landscapes, one with lots of shaking, one with less, but maybe the same net uplift rate. Are you changing the erosion rates? Are you changing the relief? Are you changing the mean elevation rate? Is it the surface process you've mattered the long-term? It's gonna be because they changed the stress state in the cross, the change in the target. So like I said, it's not so much a question, it's just the thing to think about. We can see ways that those effects will change that. Maybe just directly go out and measure. Do we see that change? If we do then, whoa, to break those effects. That's exactly right. I mean, this is why I wanted to present this here at this workshop, I don't know exactly how to get it done, but I think I'd like to strategize that, but I just really, sexually I find it taken by this idea that you just mentioned, which is, we can step back, we don't necessarily need to get fancy with our models and we can just ask, do a place like the Olympic Mountains, how different is the landscape there over time because we have M9 events? And what if we compare that to a place where there weren't any, would there be a difference? That's really the sort of question I'm asking. Hi, Allison, that was a really nice talk. One of the questions that I have about the project is it seems like you guys are having some difficulty finding these landslides and the landscape. Have you guys considered other techniques like lakes or other things to try to work on that? Right, so yes, and you guys, I wanna be clear that we're not the only people, I mentioned Josh's work, but there's also other groups that are working and some people are, well, Josh, you can comment on that if you want to. Also, I just heard that there's both things. So yes, people are looking at it. I think we're gonna have to, this isn't gonna happen by one or two people trying to go data landslides. We need a really wide look to try and put this together.