 Institute was held in the summer of 2021. We did everything online, but still I think there was quite a bit of collaboration that went on in small teams after a bunch of trainings, et cetera, that we did. And the last few days of the Institute, people work on a team project. And so the presentations that we're gonna see today through hosted by CSDMS are the resulting sort of outcomes and findings and like notebooks, so like little codes that people put together and played with and that hopefully are nuggets for others to like either teach with or like do a little bit of like exploratory research with. And so the first project team who is gonna present worked on shoreline changes and they did this very novel thing of coupling coast-sat satellite products and maps with the coastline evolution model. And so this team consistent of Armando Dundur, who is in Delft, Benton Franklin at UNC, Connor Lester, Megan Gillen, Meredith Link, and Sam Zep. So quite large team and I'll hand it over to keep us a bit on time to the first presenter. Maybe give me a quick heads up because I might have to make you co-host who's gonna present. Yep. Can you share screen, Ameth? Or do, I think I've made your host. Yes, I can. Okay, great. Okay. We're seeing it. Great. Hello, everyone. This is Ahmed on behalf of the coastal team, Benton, Connor, Megan, Meredith, and Sam was. So are presenting the simulation of the shoreline change using the color coast-sat coastline and the coastal evolution model. So a brief introduction about what the problem behind our idea that there is around 20% of the earth that is a coastal area. And there is around 45 of the human population live in these coastal areas. And the studies showed that around quarter of the sandy coastal areas are retreating. So that brings us that we need some future prediction for this coastal area. And in order to do so, there are several ways that can be by the analysis of the shoreline, the change over the year, or could be using 2D or 3D models which need a high computational cost, or could be using physical models which also need cost. And could be using the sample models or the online models, which has a reasonable cost and reasonable computational cost. One of these famous sample models that is the coastal evolution models that was made by Andrew Ashton and Murai and others, they made this model in Xeplora purpose to explore how the coastal area change especially on the long term. And as we can see from the simulation, how the space is growing over the years. And this coastal emission model was already there in the repository of the CSDMS. And another tool that was made recently that cost-set tool that was made to extract the shoreline position at any place around the outlet from the satellite image, like from Landsat 578 Sentinel-2 and it has all image from 1980 till now. So you can just easy and go and just take a shoreline. So at that point, and that also is written in the Python. So we think, okay, why not we cover these two models that we can go at any place around the outlet, just choose the shoreline and then run it into the other model, the CEM. And we work on this idea like this. So we use the cost-set tool to get the shorelines. And from the shorelines, we construct a bathymetry that we use it for the CEM model. And also we use some data that's from some wave boys that also we can access using the Jocter node to get the wave input for the CEM model. And by that we can run the CEM model. To outline this idea, we made it into two parts. The first part that when we run the cost-set part. So you just go and select the coordinates of the area wherever in the outlet. You just set the dates of the shorelines that you want to select the satellite image and you just give it a name. Then the model will go to this area. This is our example that we choose the area space lake. And the model will access the satellite image of this area as you can, we can see here. So you look at the satellite image. If you see it clear that you can detect the shoreline you just click keep. And then you start to detect the shoreline yourself like this, for example. And then after you finish, you just click enter. If you don't wanna add anything, you click in. Then the shoreline will be saved to the geogson format. And then after that it will turn into another file that we want to run the CEM part. And also for the historical wave data from that part, we can access the MDVC wave point by just put the number of the station of the wave point. Then it will end the range of the years. Then it will collect the data and for the CEM model we just take the mean of this data to run the CEM model. For the other part of the running the CEM model that we put it in another job book. So we turn the shoreline that we just draw to epithymetry using the beam profile rule. And then we adjust the orientation for the model and make some preparation for the domain. Now we have our domain for the model ready to run. And there's some setup configurations, some input parameters. Now we almost ready to run our model. Yeah, now the model is running. So now we can see the space is growing over the 20 years. Here it doesn't look similarly what is having to reality because we still cannot yet represent the structures that protect the space in reality to move. But this is just show a case if there is no structure what would be happen after 20 years. So by that we can go to any area around the outlet just detect the shoreline and we can see on the long term what could be happen. For sure it's still need some work to if when I reckon real predictions but for now it could be used as a learning tool. And so people can learn from it and can maybe look at what is might go at their local place. And that's it, thanks everyone. And I'm happy to open for any question. Thank you so much Amit. I think we have time for one quick question for the coastline team. There's a question in the chat by Jara Martinez Sanchez and it asks what's the data source for the initial bathymetry? Yeah, so what we do here, we only take the shoreline and then we assume a dean profile if you're aware of the dean there is a question that assume that profile for any coastal area. We just use this theoretical profile and then we just put it everywhere along the shoreline and from that we create the bathymetry for them. Thank you. We hope in the future that will be an easy tool to extract the bathymetry that is not there yet. I think it would be more rare. One more year of team project work is what I hear. Great, thank you so much for an excellent presentation and thank you for the whole team for putting this together and I certainly would know that this will be used in classes because my class has just been running a theoretical CVM model and I'm gonna point them to this resource on GitHub. I'm gonna introduce the next project team. This is Kevin Pierce of UBC, Lauren Robersh, who is at Tulane and Nishani Moroba, who is at the University of Alabama and they worked as a team to include wildfires in landscape evolution modeling. And I think all three of them are gonna present some if I understood that right. Yeah, so I'm gonna screen share and then Nishani's gonna start, so... Sorry, cool. So we are ready whenever you'd like to start, Nishani. Yes, I can start. So our project is focused on modeling the impacts of wildfires on sediment fluxes and landscape evolution. As you all know, wildfires are a major form of disturbance occurring in landscapes and especially if you are from around the bold area, wildfires are certainly not a new phenomenon. You may have probably seen the blackish and turbid water that's flowing in streams after a fire event. This picture here shows a fire that happened in Boulder in 2020. So during ESPIN, we wanted to use LandLab to explore how wildfires change the erodibility of the soil and how that impacts sediment fluxes and landscape evolution. For this, we use two standard LandLab tools, Fastscape Eroder and the Space Model. Both of them are based on solving the stream power equation. So the Fastscape Eroder is a fluvial incision and landscape evolution model, which is relatively fast. And the space is also a model that simulates landscape evolution by river erosion, but specifically it's capable of tracking bedrock erosion and sediment transport on the bedrock simultaneously. So building on the capabilities of both these models, we built a new tool to simulate fire impacts using LandLab. In this tool, a fire burner is used to generate stochastic wildfires that burn parts of the landscape and then the erodibility stepper will change the soil erodibility due to the fire and then a fire plotter is used to visualize the burnt areas on the landscape. Using that, we show how sediment fluxes increase and how the landscape evolves. So next round, we'll talk about how we implemented that. Yeah, so I will walk through these code blocks we have here. We start with an import statement. And then we move on to setting the parameters, things like the time step length, the erodibility, and then you can also change the fire frequency, et cetera. So the first step to build the landscape evolution model is to instantiate the model grid. And here we have a sloped grid that is rough and it's sloped towards the Southwest where we have a single outlet for the watershed. To make the watershed the landscape, we run the fastscape eroder through a time loop for two million years. And this is to reach a steady state landscape. So there you can see this is a five by five kilometer grid and we've got some hill slopes and some channels moving down to the outlet point. Then to check if we're at steady state, we plot a sediment flux time series at the outlet. So we have sediment yield here on the y-axis and time on the x-axis. And as you can see at the beginning, the flux increases a lot as the landscape is evolving. And then eventually after one and a half million years or so, we have a steady sediment flux. So we know we're at steady state and we can add the fires in to see what the effect of the fires is. So we switch over to the space model and we switch our time steps to a single year and we'll run it for a thousand years. So this is much more detailed. And then we run the landscape evolution model. This takes around 10 minutes. So we can move on to the output. And here is again, a sediment yield through time plot on the top. And you can see that there are these spikes in sediment yield and then they slowly decay back to the steady state flux. And then on the bottom here, we've got fire magnitude. So this is the area of the landscape that's affected by fires. And we can see that the larger fires, these two big spikes create much larger sediment flux signals at the outlet of the watershed. Yeah, and to look at how fires play out across the landscape in more detail, we made a couple of videos. So the first one shows fires occurring alongside landscape evolution. And so you can see that fires have random locations. They have random aerial extent. And when these fires occur, their impact on the erodibility of the landscape decays through time. So you can imagine that as the erodibility gets increased in these locations, it causes waves of sediment delivered from rills through channel networks associated with a given fire that occurs. So to look at how that sediment flux plays out, we've formed another movie of this is essentially the sediment flux anomaly. So there's a steady state sediment flux that's occurring in the absence of fire. And when fires occur, the sediment flux goes beyond that. So you're seeing the amount of sediment flux here beyond the steady state flux. And you can see as the fires occur, they light up that portion of the drainage network, which is connected to them. And we get waves of sediment traveling downstream associated with erosion from fires. So looking at this, we'll watch it one more time. You can imagine that these waves of sediment, they travel downstream, they affect channel morphology, they affect all of the aquatic habitat associated with and adapted to that channel morphology. And so what we produced here is kind of a first order computational model to understand how sediment delivery from wildfires impacts landscapes and ecosystems. Yeah, so in summary, we simulate using landlub, the effects of stochastic wildfires on soil erodibility and how that leads to changes in sediment fluxes and landscape evolution. This tool is suitable to simulate these processes over a range of time scales, probably sent neal to millions of years. And we hope that this open access educational tool will be beneficial to students who are interested in learning about fire impacts. And finally, we thank the SPIN organizers for giving us this opportunity to explore these wonderful research questions. Thank you. Thank you, wildfire team. This looks amazing. I've been so intrigued by the connections that you've made sort of between the landscape evolution modeling and then like now putting the fires into that framework. And yes, it's gonna be like a great visualization tool for like students learning about fires. But I think there's also quite a bit of research that could still be done with a tool like this. We have time for a question. And I see some compliments in the chat, but like hopefully everybody's seeing them now. Can we do it on video? Yes, go for it, Sagi, welcome. Well, this is great work team. So well done. I'm just curious of how did you control for the effect of the wildfire on the actual sediment change? And whether or not it's, you think it's fairly realistic, meaning if someone wants to use this for more shorter and modern sediment modeling? Yeah. Okay, we're outside. We actually, for now, we just, we used fairly arbitrary values. So we have a change in erodeability that you can set. So we set that to a specific value and then it decays over the course of 10 years. But, you know, those can be changed to realistic values. And at short timescales of kind of large oversight in the models that we don't include kind of rainfall intermittency. So obviously for the first months after a fire, what really happens to the landscape is dictated by whether it rains or not. And so it's not really applicable over very short timescales, but decadal probably. Thank you. I mean, I think there could also be a little bit of play with that, even if you do it very synthetically, it's like whether the rainstorms like happen like really quickly after like exposure of the fires, like California right now. And I've been waiting that in contrast with Colorado that has like almost a full snow season. And like a little bit of like vegetation recovery. And then the like monsoon rainstorms hit it. So sort of a difference that I think you could peak that with like sort of synthetic peak potentially. But like this is such a good tool to play around with those ideas, I guess. Thank you. And so I should say the team was showing this that they are like final, like little slide. Like all of these tools or most of these tools are all like available through GitHub, through the Espin team repository, but some of them are also like actual labs in the educational repository already. So we can point people to them. The third project team worked on landscape scale modeling across a variable slip fault. So like more of a technical focus. The team consistent of Grasshopper who's at the University of Massachusetts, Tvera Oren-Gwis, Oren-Gwis, Judeb, Katrina Gellwick who's at the ETH and Francesco Pavano at Lehigh University and Joss Wilbert who's at University of Toronto. And Tvera will be the presenter for this team. And I made Tvera close. So she should be able to share a screen. Are you seeing my screen? We do. Okay. Perfect. So yeah, good morning everyone. Thanks Irina for the introduction. Today I'm going to be presenting on behalf of my team that this was our project. And as Irina said, our title is Landscape Scale Modeling Across a Variable Slip Fault. And before I start introducing what we did, I would like to tell you what make us to be together as a group. And it was all of us, we were interested in the role of tectonics in landscape evolution response. So we started talking about how we can include the tectonics in modeling and because models can help us to figure out what is happening in the landscape. And we decided to focus on a strike a slip fold. First, because, well, as you can see here, this is one of the most famous strike a slip fold instructors in that is the San Andreas fold. That is a plate boundary. And strike a slip fold are really important and not just because of themselves because also other plate boundaries also include the strike a slip fold as a component. So that represent almost 50% of the structures are oblique faults in plate boundaries. So considering that a strike a slip fold are important and they are complex, analog experiments, landscape model can help us to visualize these complexities. And the community have been working on developing models. And with this, we decided to focus on one specific work that was published in 2019 developed by Madine Raymond. This is an open access model that you can find here in Sonata. This is how the website looks like. And in this website, you can find all the contents you run the model by yourself. So this is a basic model framework that we decided to use because include the lateral motion of the strike a slip also add a complexity that is the discreet earthquake that is really important in active strike a slip fold and is based in land lab that was one of the main takeaways from our summer institutes. So with this framework, we were asking ourselves like what we can include what is important in the topography related with the tectonic role in landscape evolution models. So because we know that strike a slip fold are complex and strike a slip fold are understood as mainly lateral motion, but in the reality, they are much more complex than that. And the geometry can be generating oblique folding that also impact the topography. So our main question to answer with this work was how the oblique folding impacted topography in a strike a slip setting. So these figures here are illustrating how the geometry of a strike a slip fold can generate tension in different direction that are accommodating deformation to the strike a slip folding. So we set three learning objectives. First, we wanted to use an open source model in Python that was one of the objectives of our institute. So with this, we can access to open data from other public decision studies and the one that we're doing in this case and you can modify it by your own changing the parameters and Jupiter notebook is a great tool to visualize the cells and the code that we are using. Our second objective was using landscape evolution in active technique setting, adding in a complexity in this case that was the oblique geometry and compare these outputs with what we can find in the natural work conditions. And as a broader impact objective we wanted to use this notebook as an educational resource where students that are interested in tectonic geomorphology can identify geomarkers and processes that are happening around a strike a slip fold settings. So the first part of our notebooks looks like this where you have to set up your Python environment, import the packages that we're going to use, define the parameters that the model is going to call. The parameters are the sleep of the fold, the time that you want to run the code, how the regional uplift that was included in the main loop created by Redmond et al. So the first part of our notebook is going to, you can go through the main loop that she created and understand what is happening in the code. But now, because of time, I'm just going to show you what we did, what was our contribution. And with this figure in mind, this is how we tried to simplify the uplift that was going to happen in the local fold along the fold. So because we knew the size of the grid that we were using in this model, we created a new grid with the same size and we divided the grid in the upper part of the fold and down part of the fold. And we divided this along fold in five parts and because we were trying to simplify how this figure looks like, the higher uplift was going to be in the fifth part along the fold, in the north part or upper fold and in the south part that is down off the fold. So if we see a profile of the uplift that we look like this. So the north part, the peak is around 200 meters and in the south part, the peak is around 800 meters along fold. So if we visualize this on plan view, you wanna see the uplift, this is how it looks like. So the higher uplift is in this, I don't know if you are seeing my cursor here, but the higher uplift is in the yellow part down in this figure and south to the right. So we again, run the main loop but not in created and this is the original uplift to the left and to the right is our output. So you can see that obviously something is happening, the topography is very different. This is a set for random parameters that we choose, you can change them, it's really easy, nothing complicated, you just make it as fast and you can see that obviously something is happening and this is the differential topography after the simulation. So our uplift along the fold is changing the topography, looks something less pretty or more similar than what we can find in the nature and the same happened with the channel response. So this is how her original output looks like. So we see offset channels, she in this one is illustrating the channels that are connected and this is how our channel response looks like. So it's much more irregular and you start seeing other geomarkers that are characteristic of a straight to sleep fold. For the part that we are plotting the channels, we also use landscape landlamp components but later in channel profile that are really useful if you are interested in looking for the response of the drainage network in a straight to sleep fold. And then so we moved to the Jupiter notebook as a educational resource. As I show you, this is the topography that we got, this is the channel response that we are seeing in plan view. And if we go through one of the most classic resources just to know about tectonic geomorphology workbook and Anderson 2011 is showing in this figure, the geomorphology of the straight to sleep fold zone and we can recognize these features in our output. So we can see a chatter rate in this area. We probably will see a scarf here due to the higher uplift in this region, sac pound that is accumulating water. About channels, we can see behave channel, offset channels and in this bottom to the right, we can see one of the processes that is really interesting to study that is the river capture. So this model can allow you to identify geomarkers and also if we, because we are compressing the time we can see how the river capture process works in the straight to sleep fold. To summarize, we have three main takeaways. We are using geoproteinalbook as a tool to run existing open source model with your own parameters and use to visualize while you are running and understand what is happening. We are including a simplified local uplift variability that is generating a different landscape response that is not purely lateral and we can compare with natural work conditions and our educational impact is related with how you can use this to visualize geological processes as river capture and identify geomarkers that are classic of the straight to sleep fold. That's all. I'm open for a question. Thank you. That was a great talk Tamara. We can take one. Yeah. I see two participants raise hand. I don't know how to see people's voice. Yeah. Do you pick one there? Yes. Great talk. I'm not a geomorphologist, but I was particularly interested in this talk. In the presentation, my question is, do you think the code could be adapted to take the inputs from, let's say, a larger tectonic model with multiple faults and use the velocity field from those multiple faults as an input to the code? Would there be any technical obstacle to that or would that be feasible? Assuming we took the velocity field and imported it into a Python framework I could use in the array or something like that. Yeah. So, as I understand, if we work with a bigger grid, I'm sure that we can be able to generate multiple straight to sleep folds. In that case, we will obviously add more complexities in terms of the steps that we are connecting the straight to sleep folds because the straight to sleep fold in our grid is like a complete input in the grid. So, I'm not sure how we can put a straight to sleep fold that just ends in the half of the grid. So we will have to do something to connect them. And yeah, if we have the input for the velocity field, that's not a hard step that we... We are already doing that with the sleep break that we are defining for this is a straight to sleep fold. So, yeah, I see it as a possible next step for our model. Excellent. Very cool. Thank you for the great presentation. Thank you. Hi, this is Shusong. I thought maybe I just speak here because it takes a while to type my question. But it's nice work. And I might have missed about your tectonic input, but I'm wondering for the tectonic uplift and the subsidence, is it only the vertical displacement or it also has the horizontal displacement? Yeah, so how we did our contribution to this model, the model already has the lateral motion. It's purely lateral and we are not changing that. We are in... On that, we are adding a vertical motion in these two regions that I showed you. So the sleep to the side is still happening. We are just adding the vertical motion, but we are keeping the same lateral motion that the striket sleep fold has. And we're just changing the vertical motion in these points and extrapolating in between. I see. I see. Yeah, cool. Thank you for the questions. It's great to see that people are here who are just checking out what the teams have been doing and that the community is here to see some of what the work that had been done over the summer and maybe make some connections to ongoing projects too. So I'm pleased that you are here. We're moving to the fourth team. It was like team climates, consisting of Risa Madoff, who's at North Dakota, Jakob Kirsberg, who's in Switzerland, and Ali Baalter at Le Monde d'Oratory. And they will speak about paleoclimate and elevation data used to implement the frost cracking window model concept. And Risa will be the presenter for this team. And Risa, you can share screen because I made your co-host, hopefully. So yeah, we have it. Risa, you might be on mute. All right, ready to go. Okay, I realized we were all introduced already. I just wanted to go over, I guess, just our backgrounds and how we converged on this topic. I think Ali has worked on subglacial erosion and Jakob is working on the climate impacts, climate change impacts on mass movements on alpine environments. And I've been working on hill sub-diffusion and climate connections. And so we sort of converged on this topic where there was already a lab set up. And so I guess just to give some background just about frost cracking, I think everybody is familiar with the concept, but it's very fundamental and for just basic landscape evolution, it's what produces the basic material that degrades landscapes. And early work of Anderson provided the sort of frost cracking window, the temperatures at which the process of free saw acts. And then the intensity is the amount of time that the rock would spend in that window. And then since then, there's a lot of research that's been going on. Hales and Roaring looked at climate variations at the interglacial-glacial time scale and saw that again, climate is playing a significant role in denudation in alpine landscapes. And Jill Marshall continues on the research with them, but they're being strong temperature control on surface processes. And so it's pretty clear that climate is impacting at least weathering in cold, alpine, glacial interglacial time scale landscapes, changing climates, and all of that is important for weathering erosion and natural hazards. And so we sort of our diverse backgrounds have had to have some kind of common threads. And our basic question is about how we can use past climate records to model and teach about frost cracking, these frost cracking windows, because it's reasonable that in different locations with different kinds of climates, maybe different climate pasts, there are gonna be different ranges of those windows. And then since that's impacting degradation and hazards, that would be something really good to know. And also I guess connecting it to, I guess the framework of using these Jupiter notebooks, how do we initiate those really new to this approach altogether? And so I guess this presentation is a little different than the previous three in that, I guess there's a lot that goes, there's a lot of prep I think that's needed before even you start thinking about coding, like what is going on in the computer before you can start thinking about how to put together a model. And so I guess I brought in Greg Tucker's lab, just to I guess show that we incorporated the basic sort of core parameters for how soil and temperature interact and how temperature changes that frost cracking depth. And so this is something in the labs that anyone could go to and we use the code for that as sort of the basis of our model, except for this lab, the changes are going to be just going on daily and seasonally. And we wanted to incorporate climate data set for paleo climate and then for specific locations. And so that's sort of where things got a bit more complex. And so I sort of put together a conceptual workflow of sort of what's needed. If anybody wants to put together a model to do this, because well, I'll say what we did and like hopefully inspire you to go further with this, but there's the idea of a computational grid and others have mentioned it and that's sort of central. It's how the results will vary across space and time. And so it's what the computer uses to think. So it's doing the work for us so that we don't have to have a spreadsheet and do it by hand. And then and so with this grid, I mean, each computational process is going through each cell and then there are sort of rules relating to how each cell is going to relate to the next cell. And I realized for those of you experts in programming and in Python and such, this is very basic, but for those who haven't done it, this is sort of what's going on behind the scenes. And so we made use of a land lab component sort of provided the skeleton for this and then Python tools, but it's really just applying a finite difference approach. And it's just relating to different cells to each other. And so we used, what we added to sort of the basics was elevation from a specific location and then temperature, temperatures through time. And we used PMEP6, which is a paleo climate model. And so I guess in general, someone could either use make up their own model grid or as what we did, we imported a DEM and then we had to code for change of slope for each cell. And also in each cell, each cell is getting to read a temperature through a time series. So each cell is computing one-dimensional heat diffusion equation and for a certain elevation, because changes in elevation will affect the change in temperature locally. And so I guess the complicating factor is adding for time series on the order of thousands, tens of thousands, hundreds of thousands of years, whatever your model has it. And so I'm not really showing code from our notebook because we still have to make adjustments to generalize some of the code so that someone could input their own climate model data, temperature data. And so this is just showing sort of the initiation states for each cell in elevation and then in temperature. And so each cell is reading like a package of data as getting input and then computing it through time. So and then the end result was we, we had a location, but we only computed it for a single cell and then add a certain time for a single time because we didn't go far enough to be able to compute it for the entire time period and for a whole like watershed that we had. But this is showing you just how much more complex even just for a single cell is compared to the original model that was shown here before. And so what this is showing is we computed a window, a place-based window, frost cracking window. So this is the temperature range that frost cracking would occur. And then the other concept is a frost cracking index. And so that's the time spent at each depth through time. Okay, so again, this is just one point in time for a single cell and this is showing the index for that. Okay, so this is where we got and there's still more work to do to at least generalize the code more. And so just this is kind of in summary is to like, what are the learning objectives you could do with any of this? And so I just came up with some general and in specific. So in general, I think what the big idea, what the big ideas are, are using real-world grid of data and applying it to a physics-based model, quantifying change through time using a real-world time series data. And I guess I don't know if this is more advanced or just like, it depends on what level course you're teaching, finding relevance resources and importing it to answer a question. And then depending on the objectives of a course, how much programming you want students to have or do themselves as opposed to importing all the data for them and having them just sort of answer questions based on the outputs. Yeah, I think I'm going to... Am I over? Yeah. I'm sorry. Wow. Okay. That's all right. I didn't get you started like quite a long time, but like for the sake of time. Yeah, I didn't realize it. Yeah, it just seems to go okay. Yeah, so I'm done. It's okay. Thank you, Risa. I encourage people to like ask you questions also in the chat or ask the team questions. And so like maybe keep an eye on the chat and then we will be moving to the team that worked on storms and erosion. This is on Angel Monzovay, Sam Anderson, Sophia Alfieus and Muriel Buchner, Muriel Neston and Grace Gurion. So quite a large team and I was the presenter if I have my notes right. Yes. We see your screen on show. Yes. Okay, perfect. Thanks. You're asked in short. Yeah, sure. Of course. Good job. Thanks for the invitation. I'm presenting on behalf of my team. My name is Angel Monzovay. And we played a little bit with storms and rainfall. So our presentation I have those storm intensity, duration and frequency influence, river channel and incision. And this is of course the results that from the summer school, which was pretty, pretty, pretty good. So I'll try to summarize most of the results because of, in terms of time. I don't know what's, oh yeah, there you go. So our particular project was trying to combine the erosion, but as a result of rainfall. So we know that the landscape evolution in particular channel, channel morphology depends on hydrologic processes, but all the variability that, that high. So, hydrologic processes, but all the variability that, that hydrologic processes has to including geology, so conditions and weather. Well, almost everything varies, especially so it's very, very difficult to include all these variables in a model. So at a small scale or at a local scale, this is very, very challenging. So we tried to include some of the most, or what we think is the, are the most important components. So first we, the most important part that we thought it would be important is to have different types of different scenarios of rainfall. And that's what we applied into, in a couple of simulations that I'm going to show here. So the objective or, sorry, I have, so the objective or, actually it was not developing a code, actually coupling different models was to build this model that illustrates how different precipitation patterns produce different erosion patterns. Okay, so the most of landscape evolution models assume constant steady flow state of water internal, which is really nicely described in this figure, adapted from Adams in 2017, which basically is most of the models, what they do are, or is to consider a certain rainfall event and calculate based on that magnitude. So what that means is that when the rain stops, all the processes stops. And we know that that is not true in nature. So after the rainfall stops, it continues flowing through the catchment and erosion and other processes still continue happening. So this is what we were trying to explore in our, when we were coupling these different models. So we use the overland flow component in land life. We were in all the team members were somehow related with river. So this particular component was really nice to, first to learn how to use it and then actually use it in different situations. So the model is nice for long-term simulation because it's really fast. It's really stable. And after a little studying, it's relatively easy to use. So we were really happy when we were able to use this component. So the model is a 2D model. It allows us for long-term modeling, as I told you before. And it's based basically on a DM or digital elevation model, a certain roughness. So some of the most important characteristics of a given watershed are included in this component. One big difference is one big advantage of this component is that we don't have to define the channels. The channels are basically a result of water flowing through the model. So that is really, really nice. The other component is the detachment limited erosion. So this is the part that takes the input from the, from the water source, from the previous one, from the overland flow and erodes the basin. So it's the nice thing about this particular component is that it actually uses this chart as an input to erode instead of the drainage area, which is basically, which is typically done and for long-term simulations. We are still working on the spatial precipitation distribution because we need to, the challenge there is to modify the source code in the overland flow to make it, to make it especially in our case, all our storms are uniform across the whole watershed, but we are making some good progress in that part. I hope that it will soon will be available for everybody. So the simulations what we did are ideal models or some idealizations of a square basin, which is really, it has steep gradients and are really short in time. So because it takes a long, long time for this model to, I mean, computational time to model when we wanted to simulate this type of rainfall events. So I'm going to show three different rainfall events, but the things, so basically they have the same amount of water between them, but they differ in the intensity and the duration and it should be, so it's really subtle the difference in the animation, but I'll try to explain why it's not happening. So first of all, okay, now that you saw the first movie, all the events have the exactly same amount of water, so we are taking that variable out of the results. So, but if you see here in the upper panels, you see how the flow is routing through the watershed. So I'll try to make a case here where this is, in time 120, excuse me, 1200 seconds, so it will be around here. I don't think you can see my pointer. I don't see it. Okay, whatever. So the thing is, you can see that the response in the different watersheds, it's different and that also affects, so these are the three cases. Okay, and this is the hydrograph at the watershed outlet. So all of them are different. Basically, they follow up similar shape of the rainfall event, but the nice thing is that they vary in time, which is one thing that we really wanted to include and the incision rates are almost also the same shape as the rainfall event and the hydrograph. So, but something that is really important here is not really to see the incision rate, but to see the areas of the incision. So that will tell us how much incision happened. So, in this figure, you can see that all of them, although the different rainfall events have different responses, which is something that we wanted to show and this is the incision at the watershed. So I apologize for the, they looked very different, but it's because of the scale of the watershed. So it's really, I tried several combination of color maps and this was the one that it was better. So what is surprising that it was the rainfall intensity that it was the smaller one, but with larger duration had larger incision. So that was really, really cool to observe in this case. Also, we tried with different frequencies, so the rainfall will start and stop and start and stop and several times to see what's the effect on that and of course it affects. And also we have some, this really nice patterns. I'm sorry that I can't show you, I can show the mouse, but this really wavy responses, which is really cool. The sources and all the codes will be available. I didn't have the time to upload them, but they are of course running because I was able to get those figures. So if, I think I was pretty quick. So if anyone has any question, please let me know. Jo, you were pretty quick, but like I was like late in the previous talk. So like we're going to still like refer people a bit to the chat and hopefully a few people will hang at the towards the end and like ask questions then still too. And we'll move on to the team that works on simulating of sediment pulses in the land lab network sediment transporter. This was Shijun Chou, Muneer Mahath, Marius Hooper, and Nelguiro, and Shijun will be the presenter today. Great. So our team will talk about network sediment transporter, a land lab component that tracks sediment from its source, transport and fate across the entire network, entire stream network at any time step we define. Okay. All right. So sediment is usually introduced from erosion of hill slope or within and near stream network, and the river generally moves downstream to the sink. Sounds simple enough, but still sediment transport is one of the most complex and challenging problems in germinology. And that is due to wide-ranging processes that define sediment transport and storage. And those processes depend on multiple environmental factors including particle size, channel and floodplain morphology and energy of alluvial system. For example, coarser size sediment like sand and gravel travels on river bed by rolling, sliding, skipping while finer particles, sediment gets picked up by turbulence of water and moves downstream in suspension. So there are different, you know, transport processes and sometimes sediment get trapped in a river system. So its residents tank and range from seconds to thousands and millennia here. So you can see how varying those processes are. NSD particularly deals with bed load transport. And we care about sediment transport because channel morphology, stability and aquatic habitat, as Kevin talked about earlier, all depends on sediment transport. In addition to the threats both human and natural infrastructure. And we care about that because we live in a world of fast transition and large scale landscape disturbance both human and natural. Human in this disturbance like agricultural development, urbanization, mining and logging, all can introduce huge amounts of sediment to stream and affect sediment transport downstream. There's also natural events like wildfire like Kevin's team and Lawrence team talked about and flooding and then slide, which have become exacerbated by climate change. So it is really important that we understand sediment transport processes and have the capability to make predictions in order to prevent and mitigate negative impacts of those landscape disturbances. So our project objectives are twofold. First is to demonstrate a potential to couple network sediment transport with other existing landline models that can generate sediment sources or simulate other sediment input conditions. And secondly run NSD with random pulses of sediment to understand, build a predictive capability to see how sediment moves through the network and so that we can begin to understand the impact of landscape disturbance and sediment yield. So I'm going to go to my entrepreneurial notebook. So I'll first talk about the first objective and then second one. First objective is nicely illustrated by this conceptual model. We want to couple other landline components like space like that could generate landscape evolution as a result of buyer occurrences and now I'm seeing this presentation. Rainfall Irresistibility long-term simulation will change the landscape topography and that, you know, it'll be really interesting to couple that with network sediment transporter to see how the in-stream channel processes are affected by long-term evolution of rainfall. So the connecting tissue of other landline model to NSD is the digital elevation model. So we started working with, you know, Shelby who really gave us a lot of help to develop a function to convert DM to network grid because NSD operates on network grid, not landscape topography. So for example, we have received this DM from space team and we decide to run a model that can convert the DM to network grid and so create network function that does exactly that. So it looks good. It creates nodes and links from the DM. So we're like, all right, it's going to work. And then we discover a bug that actually duplicates some of the links between some of the nodes. So we have more number of links than nodes, which is, you know, preventing us from running NSD. So we step on that part but we're continuing to work with Shelby and Eric to debug that model because there's still, I feel like there will be a lot of utility in a function that could create network grid from DM for not only NSD, but other member component that operates with the network grid as a foundation for model execution. So from here on, I'm going to move on to objective two, which is to run NSD with randomly generated segment parcels simulating the effects of fire and landslide using example shapefile. So we use methods of basing, which is including the land lab library when you download that package. So, but, you know, it's a shapefile. We still need land lab in order to learn NSD. We need a network grid. Rich shapefile is a function that converts the shapefiles to network grid and picks up some of the attributes that are included in the shapefile. So, I'm running the rich shapefile. We have converted the shapefile to network grid with the links and nodes. And now we have right number of links and nodes. There should be one more nodes and links as it should. And it also picked up some of the physical attributes from the shapefile like rich length, drainage area and topographic elevation, which are important factors but we still need some additional topographic and hydrographic information to run NSD. So here we define the bedrock elevation based on topographic elevation and assign some arbitrary channel width and flow depth. The meat of our objective two is the ability to introduce additional sediment pulses in our network grid. So we have done that by creating random sediment parcels and data records and data distribution. So this histogram shows how many numbers of parcels that we're creating for the link. So we have an array of randomly generated sediment parcels. Now we can assign the random sediment parcels to the links which will be tracked using element ID for all of the links at each time step moving forward. And then once we have the array we have to we also need to define some sediment that has a grain size distribution and so that we can track each parcel downstream as they move forward in time. But in order to track sediment we have to classify parcels as other active or inactive. The sediment that are on top layer are active and buried are inactive. And active parcels are the most recent parcels arriving links so it's important to track which are active and which are not active. And the active parcels are moved downstream using a sediment transfer formula. For this we use World Cup Crow which is a bed load equation. Remember NSD currently only does do bed load. It's a bed load equation designed for a well graded poorly sorted river containing both sand and gravel. So now we can begin assigning each parcel an arbitrary arrival to the sediment transporter. Now we have all the elements defined. We can now collect those arrays into dictionary variable which will be entered into NSD model formulation. And in order to run the sediment transporter we have to define some time steps and we run it. And then once we have the time steps we enter the variables that we already defined and then we can run the model through this. 10 minutes so short. Now I'm going to show you some of the model results. There are land lab plotting tools specific to NSD. In particular plot network and parcels create plan view map of network parcels and links. We can color them both. So first I thought it would be interesting to just kind of look at how much sediments generated in different links. So at time step zero another example if you look at the link 25 which is here we have generated some sediment at time step zero. And I wanted to create animation but I didn't have the computational savvy to do it. But okay this is time step zero and if we move to time step five we see that the sediment that was generated in link 25 has moved from first order stream to second order stream and in the final terminal time stream so that's really cool. But we can also plot all the parcels using the same plotting tool. So here we see at time step zero all the parcels that's created in all the links that we defined here at time step zero. And when you move to time step five we can see that a lot of the parcels have moved downstream and some of them exited the network at time step 10 further moved downstream. There are some not networking plotting we did just you know as we can see that as the time moved forward we had less number of volumes as we as one would expect. And same with this plot is showing the cumulative distance versus grain size. So it means that larger the grain size less distance it traveled as one would expect so it's generating results that are reasonable. We're not done yet. I mean I feel like this is a continuing work. First we want to complete the modeling framework to couple more than one land map component and after today's presentations I feel like there's other ways to connect with other teams to see how you know how we might model and make predictions on how reverse transport sediment downstream that results from long-term evolution of atomic events, rainfall events, fire landslide like highlands. So we're going to continue to work on debugging that function that translates DM to network grid. And secondly we want to develop a suspended solid load algorithm to append to current NST right now only handles bed boat. And reason for that is water quality concerns are largely on suspended solid loads. So it will be interesting to add that capability to NST and we'll have a huge scientific value. Okay, I'm done. Thank you for a great presentation and this network transporter is such like a novel implementation in the land lab framework that it's been great that your team kind of picked that up and it's been working on it and I agree I could see the linkages with other teams too. I'm going to give the floor to the last team that worked on simulating simulating craters on planetary surfaces and this team consisted of Emily Bamber was at duty Austin and Gaia is to be the key who is now at Harvard and Emily is the presenter. Ron, it's so nice to see everyone from Austin again. I'm excited to close with you, but it's a bit of a different kind of thing from the rest of the presentations which are based on kind of transport and stuff. So Gaia and myself are geomorphologists I would say but we I am mainly looking at surface of Mars and Gaia house in the past as well, look at the surface of Mars and other planets and we're interested in other planets as well. So it's been a summer institute we wanted to build something that could mesh well with landlabs so we could run landscape evolution models but we're on a created kind of background terrain as an initial condition. So what is an impact crater? It's just a depression basically in the surface and it has usually has a high standing room around it. So here's a photo of me to see giant structures like crater on the moon which is a quarter of an eighth of the moon's surface or something ridiculous. And why do we really care about impact crater? Why do we want them landscape evolution models? So they're actually maybe the main geomorphic agent on every surface except Earth and they probably were important in earlier sea pollution as well because they were affected by patectonics or everything's been printed and the craters from the past still exist but they've been affected by landscape processes. So we can necessarily use that as an initial condition but we want to kind of go back and reconstruct the initial crater to photography as a kind of input to a landscape evolution model. Craters are also really important for dating surfaces because the number of craters and size distribution of them is like dating for our model as well to kind of integrate that with time steps of like normal because it's kind of a random process. And they also record solar system evolution throughout time that can be at any scale. So meteor crater is like a kilometer across I think or something and then you're still showing a crater so these are really like lots of different scales and it can be really important whether you're looking at kind of small scale things like kilometer scale to global scale kind of things on planetary surfaces. And the reason we're interested in kind of the geomorphology hydrology of craters is really important. I think it's really important to understand in the past kind of surface environment like I'm personally studying how did rivers form into these craters so I want to run landscape evolution models kind of on a created landscape and find out how inlets are forming and what we can understand about them because it all has implications of the habitability of lakes that we're exploring in the delta there so pretty cool stuff. There's lots of reasons to study it. And the main problem is that there currently isn't an open source accessible code to generate a cratered surface. There's lots of codes that generate like an individual crater and I think as I downloaded this presentation it doesn't link to the nice video that would have been here but there's lots of videos about the crater and how the ejector around the crater forms which is just the kind of material that the crater or meteor blows out as it creates the crater. And Langlab is such an amazing powerful tool and you're all adding like even more to this. It's so exciting that I want to be able to use some of that basically. It's called Mousom from the Alan Howard's group and it's been around for 20 odd years now but it's written in Fortran and I love Python and I'm not sure I want to learn Fortran. Langlab originally also did have a cratering component but it seemed that it wasn't updated into like subsequent releases and updates of the source code and there's also we found some like cratering kind of things on GitHub but they deal with this thing called crater saturation which is kind of measuring until like what time how many craters do you accumulate on a surface until at the point you're adding one you're also obscuring another one so you're not adding any more craters to the surface it's kind of reached its kind of boundary so we wanted to add that in so we could model the evolution of the topographic surface and so the basic idea behind this is just to the it's taking work done by Alan Howard and basically writing it coding it in Python so the initial the shape of the crater is already fixed and so we can actually set the inside of the crater has a shape that can be defined by polynomial fit and that's based on data taken from a lot of planetary surfaces of fresh of whether fresh craters such as the moon and that haven't been affected by a ton of geomorphic processes and then there's also a shape for the rim and outside of the crater that can be fit in the shape of the crater and what's inside it versus what's outside and has a different equation to describe the shape which will include the eject as well which is that blanket material that gets deposited of the stuff that gets ejected from the crater and is the the actual equations that can be used and in the notebook that we wrote and the code overall is model the shape so go from just like saying there's a crater here to actually having the shape of the crater in personal landscape and have a population which reflects how craters form so they actually it's a random process what's going to impact the planetary surface differs on a bunch of things so it's coming from a population of like asteroids or comets and randomly hitting the surface at random points in time so we wanted to kind of simulate that stochastic process and have a population of craters on the surface and then eventually what we want to do with this code is run some of the land lab and then run and all the things that you've all been talking about to see how that evolves in time and that was written a bit into the lesson we produced as well and so I can actually post a link to this in the chat so this is the lesson that's now live on the TSTMS website so you can actually go and check out the notebook on there that we have in the lesson and talks you through how we choose from a uniform sorry how we choose locations from a uniform distribution on the square grid but we choose the shape it's like a cumulative distribution because you have smaller craters and you would have larger craters it's kind of displaying all information so you can like digest it and understand how we're simulating this very random process I will stop here I was going to talk through the notebook but in the interest I know that everyone didn't have questions and stuff so maybe we can have a time for questions like everybody at the end of this it's better for everyone I think thank you thank you Emily this is such good work and that it was like really nice to see that like all together and like it's been up for a little while and I think some of the repo notebooks are very cool yeah that's super cool and if you do end up using it at any point or want to use it for a lesson have any questions like please reach out I'd be happy to or any comments as well that's a very dangerous offer I mean we're coming to like sort of the conclusion of the official part and I'm hoping that some people will stick around for a few questions to each other you know like we all like are trying to like sort of like pack our days and go from zoom to zoom or from class to class or however the situation is for everyone I'm going to keep the zoom open for about like 10 more minutes or so so that's if people are keen to like ask a quick question then they can do that but otherwise I wanted to like thank everyone like I want to thank everyone for just attending and like to the aspen teams and being there for questions and I especially wanted to thank the 2021 aspen participants and presenters for today who are willing to like sort of step it up and like share their latest and greatest so thank you everyone and these team projects are always so impressive to us so like we're really like proud to see how far along they get in just a few minutes and like some work like still like going on afterwards for like many of you so thank you everyone