 Okay, welcome everybody to this CSDMS webinar today. It's gonna be a unique one. This is a follow-up webinar to the ESPIN Summer Institutes. This was like Earth Surface Process Modeling Summer Institute that we had in August this year. And it's good to see like so many familiar faces of the participants that are back here together. But it's also a moment to share with the larger CSDMS community what went on in this summer institute where we spent a bunch of time on like training and tutorials and presentations out of the community on surface process modeling, but also on programming and on like best programming practices and collaborative work when you're developing software. And then that summer institute ended with projects that were done by different teams. And so what you're gonna see today is the products of like about one and a half days or so of like a team of eager grad students and early career folk who had some programming experience from the days before using the tools of CSDMS or the land lab suite and like how they work that into like something that's hopefully useful to other people too as a teaching resource or as an example or tutorial. So we had a Slack channel during the ESPIN period and I know that people had like worked quite a bit like afterwards still on perfectionizing their projects and like work collaboratively on like making this like a presentation that's worthwhile for other people in the community. So I'm quite pleased to see that there's a lot more people than just the ESPIN folk who are doing a little reunion today because this is gonna be useful for other people maybe for teaching or for learning about these tools. I wanted to quickly thank Nicole, Mark Piper and Benjamin Camphords, Nicole Gasperini for being here today too. They were like the main instructors in this ESPIN summary institute. And so it's nice to see that they're like here to listen again and find out what's were like last improvements that these teams did. The first team that we'll be presenting will be represented by Brooke Hunter. She's at the University of Oregon and I'm assuming that you can pull up your slides, Brooke. Yeah, I can do that. And her team has been working on exploring the effects of rainstorm sequences on a river hydrograph. Okay, can you hear me and see a screen? Yes. Okay, perfect. So, yeah, I'm Brooke Hunter. I'm at the University of Oregon and I'll be presenting our group's project today on exploring the effects of rainstorm sequences on a river hydrograph. In addition to myself, a little zoom thing is in the way. In addition to myself, other people who are part of this team are Celia, Lisa, Tianu and Yuval who are shown there at the bottom. So for our final project, we created an educational Jupiter notebook. It combines three land lab components, the precipitation distribution, the soil infiltration green amped and the overland flow component to explore how rainfall intensity and storm sequences and hydraulic conductivity affect hydrographs and cumulative infiltration over a 30 day period. This difficulty is aimed at like undergraduate courses, intro level, there's minimal like no coding experience needed to run this notebook. And that wasn't run for us because like for a lot of undergrads like coding and math can be kind of very daunting. And so we wanted to provide like a low risk opportunity to get exposure to coding and get some experience with Python. And then, but there are also like opportunities where you assign the questions we have and students can go back and change the code and rerun it and kind of expand upon those Python skills as well. So some learning objectives we have in this notebook. We introduced the concepts of rainfall intensity and storm sequences as well as hydraulic conductivity and then asked questions to the students to investigate how these like changing these variables affect stage water level and cumulative infiltration over time. In addition to these topical learning objectives we also ask or have some Python skills that can be learned in the notebook. They can see how three land lab components can be integrated in a single notebook. They can see how to create a topographic elevation grid from an ASCII file and see how to create Matlab plotlet or create plots with Matplotlet and use like CSV data and Panda data frames. So prior to running any of the code students will be stepped through a conceptual model of what the code does. So before seeing just like huge chunks of code they can see what it is visually doing. And so then rather than focus on the code during this presentation I'm just gonna step through that conceptual model and explain what land lab components we use and why and then show some of the final plots that students will use to ask or to answer questions, provide example questions and potential opportunities for like expansion upon this notebook. So first the students will import an ASCII file and set up some boundary conditions for this Hugo catchment and then they'll visualize that catchment in the notebook as well. And then they're going to use the precipitation distribution component in land lab to generate rainfall time series. And so we provide a storm parameter CSV file that has two different storm series types. So scenario one has high intensity short storms and scenario two has lower intensity and longer storms. The way that this notebook is set up if students modify the CSV they can change the parameters and create different storm scenarios and they can additionally like add scenarios if they want to have more than two scenarios run through this model. So the precipitation distribution component will provide a surface water depth for each time step from the rainfall and then you can feed that value into the soil infiltration green amp component in land lab. And here is where students can play with the hydraulic conductivity values. And so we have this K vector that there's a K value for each scenario. And so here's one opportunity where students can kind of change code a little bit there. I do want to note that our infiltration depth here is cumulative over time. We don't have a evapotranspiration or any like subsurface flow component here. So that is just continually going down into an infinite soil depth there. So then after the infiltration stuff has occurred that surface water then is fed into the overland flow component which will route flow over the landscape. And so that surface water depth in each cell is actually a function of two inputs. So the rainfall and flow from up slope and then two outputs the infiltration and then the flow down slope over the time series. And so the fact that this infiltration is just cumulative over the entire time series that is like one place where a student if they got really into this could try to incorporate an evapotranspiration piece to expand upon this notebook. And in addition, so correcting all these variables we can also look at how stage water level changes at the outlet of the catchment over the entire time series. So here are some of what the final plots would look like in the notebook. We have rainfall intensity in black on the top here and then that stage water level in blue over the time series. And on the bottom we have the stage water level and then that cumulative infiltration value. And it also counts the number of storms that you see during it. And then the mean storm depth as well as the peak water level here and just the hydraulic conductivity variable that you use. So some assignment questions I could be easily asked in this notebook without changing any of the code would be how does the hydrograph response differ between a high versus low intensity rainfall scenario? So for scenario one, students would see that higher rainfall intensities you have higher or greater responses in the stage water level. And this may seem like a pretty simple question but for like an introductory course that the student has never had any exposure to these concepts before it's a really good leading question. And then you can start to ask more complex questions like describe the decay of the stage water level height after the storm has stopped. Does it stop like abruptly or gradually and then try to get those students to think about like why there is this gradual response in the hydrograph. So yeah, neither of those questions would require any modification to the code but then you can also ask questions about hydraulic conductivity which would require changing some stuff. So you can ask students to require a rerun the notebook and make sure your rainfall time series are the same, so provide the same parameters and then change your hydraulic conductivity values and then ask them to compare and contrast the final plots. And so asking like what effect does increasing the K value have on the hydrograph? What about the cumulative infiltration? So they can start thinking about how infiltration rates and sort of properties might affect those catchment river responses to storms. As a conclusion, we created an educational notebook for introductory courses that will be publicly available soon. We haven't totally finished that part yet but it introduces concepts of rainfall intensity, soil infiltration, hydraulic conductivity, hydrographs and because it does require a little manipulation to the code it provides like a low risk chance for students who have never looked at Python code before to kind of get some experience and see what it looks like but then there are also opportunities to expand the notebook and like make challenge questions for students who do get really into the coding part of it. And I can take any questions for some time for that. Thank you, Brook. That's a really good overview and it's quite impressive to see how much you developed it again towards like students and like giving them like opportunities and sort of like laying out like how people could use this which is a big step from where it was last time. We're opening it up for a question and people can just, I think people can unmute themselves. Maybe type your question in a chat if you can't unmute yourself. Brook, I have a question. This is Greg. So this is really impressive, super cool. I can immediately see a million different ways you could use this in a classroom. One thing I was wondering about is how easy is it in your notebook if you wanted to swap in a different DEM? So for example, if students had hydrograph data from some nearby watershed, could they explore that instead or is there a way to build that in? So this is actually something I was talking about with one of my committee members because I kind of showed them LandLab. They're an ecology and they hadn't heard of it before. And I haven't tried to do it but I imagine it wouldn't be like terribly difficult to do. They were more interested in the evapotransportation part in comparing how like soil infiltration and like soil moisture values might compare to like a real life catchment. But I feel like it could be pretty easy to swap in the DEM. I haven't tried it but... And then compare those to like real values could be really interesting. Can I ask a quick question? Yeah, please. So this is gonna call great job team. I'm super excited. And I just have a question. So you have these beautiful sequences that create beautiful hydrographs. Did you have to work really hard to like figure out what sequences you were gonna use? Like how much tuning did you do to get those really nice illustrative results? So like of the storm parameters. So I have to thank you for a lot of that work. He spent a lot of time trying to pick parameters that were representative of somewhat like realistic. I think for the scenario one thinking of kind of like intense like desert really storms where they have like these like bursts of rain. And so he did spend quite a bit of time trying to make it look realistic. And then I spent a bit of time trying to find some like soil infiltration parameters that made things look somewhat realistic. I think Albert are you have a question too? Yeah, I have a question. First of all, excellent presentation. That was very nice and great group work I think. So when you presented the rain events and the discharge that followed kind of those rain events, they were laying on top exactly almost. And is that because you got a really small basin? So it's very reactive. When you've got a precipitation event it's almost immediately shows up at the outlet of the river or is there something else going on? So that's something I was looking at as well. My guess would be that it's a small catchment. I don't know, Celia might have an idea as well she's also on this call here. But I think it's just because it's a small catchment. Maybe there's something in the code that we could like see if there's something going on. But I think it's just because it's just a very quick response in a small area. Thank you, Brooke. That was a great talk. We're gonna move to team two. They are very large team. And they worked on like fluvial geomorphology and different components of like fluvial geomorphology and Rachel Bosch of the University of Cincinnati is gonna present their teamwork. Thanks for that great introduction, Irina. Let me share this right to the beginning. Can you all see that? Yes. Okay, great, thanks. So I am presenting for team two today which is the really big team. Sheldie, Josie, Eric, Francois, Hima, Vinnie, Edwin, Mochit and myself were this team. So we took on a pretty big question. We wanted to couple different types of grids within Land Lab at different scales to explore some fluvial geomorphology questions. So because it was such a big team, such a big project, we ended up distributing tasks throughout the team. And so we did a lot of work collaborating and learning about Git and GitHub and the whole process of continuous version control and it was a very interesting introduction. So here is the conceptual model for our simulation. We have three different types of grids that we're looking at here. We have a large-scale rectilinear grid representing our upland catchment. We simulate storm events on the upland catchment, gather those in and feed them to a network type grid which is the vector channel network that you see going across the upland catchment with nodes along it. So there's our network grid and then we transport sediment along the network grid and we take the results from the network grid and pass them to a finer downstream floodplain where we simulate the infiltration from that storm event. So here are the learning objectives that we addressed with our simulation and we have two sets of learning objectives. The one is the software engineering and the other is the fluvial geomorphology problems. So our software engineering objectives are all within Python and we built this in an Egyptian notebook and so here we're looking at coupling different types of grids, the rectilinear grid with a network grid and we take a moment to point out that that is a novel contribution to the land lab ecosystem, that type of coupling. Did not yet exist. And then communicating the output between these types of grids and then we coupled existing land lab components on that coupled grid framework. And we use that to address our fluvial geomorphology objectives. So here we're looking at the effects of rainfall intensity, rainfall distribution, sediment supply and then the infiltration properties downstream. So this is just kind of a brief overview on how we did that first step of grid coupling. We used the same sample catchment DEM that you saw Brooke present in for team one. And from that we extracted the channel network and then laid a network grid along that channel network identified the nodes and made sure that it was all communicating smoothly. We then took that output and fed it down to this finer grid in the low lands where we were able to adjust for the cross sectional topography of the floodplain as well as variable hydrologic conductivity within the floodplain and then these are also parameters that students could change within the code in order to explore questions. So here's just one example of that kind of parameter exploration. So learners can do this within the code changing things directly or we can have things like this slider where students don't have to have a lot of coding experience and they could change for instance this threshold which controls the density of channel generation when you extract the network grid from that upstream Raspard grid. And here are the results when you run the whole simulation we have rainfall in the upland and it transports down through that network grid. Oh, I went too fast. Sorry about that. And then you can see the sediment transport and the histogram as sediment is moved downstream and then the results from the upland grid feed into the floodplain grid and you can see the inundation of the floodplain on the right. So we were able to accomplish that goal of coupling these different types and sizes of grids. And so I wanted to share with you kind of the next steps we have in mind. This is still a work in progress. We've been communicating since the end of ESPN and having Zoom meetings and we're continuing to work on the automation of linking that upland catchment to the lowland floodplain. We're working on different types of visualization including that sediment transport. And notice we didn't have a video showing the sediment parcels moving downstream. We're still working on that. We have the beginnings of questions to deliver this as a lab and we're continuing to develop those as well as building a tutorial to help other people who might want to couple a raster network, sorry, a raster grid to a network grid. And so here is the link to our GitHub repository if anyone would like to go and see what we've been up to. And hopefully we have a few minutes for questions. I'm gonna stop sharing and hand over question fielding to my teammates. Go for it with questions. It looks like everybody can just unmute themselves and ask questions if they want to. Excellent, Rachel. Oh and credit to Shelby for the artwork which is amazing. Can I ask you a question? I will just monopolize all the time if you let me. So this is really cool and it's really interesting that you were able to couple things that had never been coupled before. So I'm guessing that took some massaging. So when you made your notebook is this more like designed for developers so that they can use, I don't know who to look, they can use these or is your notebook more geared towards students? I think it might be geared more towards, I'm thinking of graduate students who are thinking of using, trying to couple components that maybe use different grid types but there could be grid types other than the ones that we choose that could be helpful, this knowledge. And Shelby and Hima have been working for a while trying to figure out how to do this. It's not a simple thing. Yeah, we were hoping to maybe make that grid coupling between the raster and the network, sort of an importable module. It's sort of at an awkward place in between where it's all in the notebook but it could be packaged into a Python script. I've got a quick question if I could. This is really, really impressive. It's super, super cool. I'm curious, what were you using to get grains to transport into that river network? Like which of the modules, probably up there at some point, I just missed it but which one did you use to get that hill slope diffusion into the river network and decide grain sizes and everything? So the network sediment transporter is the component doing the heavy lifting of getting the sediment through. We actually don't have communication between the evolution of the hill slopes. That would be advanced. Currently the coupling happens from the overland flow sends water levels to the network sediment transporter and that feeds a sheer stress that then mobilizes the existing. We prescribe a sediment thickness and that has a grain size distribution to it. And it's a Lagrangian transporter that moves parcels down. So that's why you're seeing the full distribution at the output. Cool, thank you. Awesome. I love that there's lively discussion but I'm also keeping on time a little bit. So like please type your questions in the chat and like maybe the team can like field those questions back and forth while we move to the next team. Rachel Allen will be presenting and this team worked on Lagrangian particle transport through a tidal estuary. Hi, thanks. Let's get this up. Share, screen, full screen. Great, can you guys see this? Yes, we can. Awesome, thank you. Let me make this go away. Okay, fantastic. Thank you so much. We were project team three and we were working on Lagrangian particle transport through a tidal estuary. So we used three different tools for doing this. The first one was based in LandLab. It was a tidal flow component that was built based on a course, Coast Morpho 2D model. The second was a random grid generator, GS tools. And then the third was a particle tracking component, Dorado, that Jay, one of my teammates actually wrote. So we had a lot of inside knowledge here. So the overview of what I'm gonna talk about today is or the outline of what I'm gonna talk about today is an overview of these modeling tools and then go through the Jupyter Notebook that we set up to walk somebody through actually using these and then some examples from like the kinds of things that you could test with what we set up. All right, so that first model, the tidal flow component is built based on this Coast Morpho 2D model, which is in a paper by Moriarty and Mershid in 2018. So that model is a 2D model for morphological evolution of a tidal inlet. And it's built with a simplified version of the full Navier-Stokes equations and it ignores unsettiness, infection, spatial variation in the water level. And so by making all those assumptions, they can solve for velocity based on an assumed form for the temporal variation of the water surface. And when you do that, you're gonna get column two in this figure here. But like through tidal inlets, inertia can still be important. And so you can do a momentum correction by, and you compute that by balancing bed friction with the inertia term alone. And when you do that and add that on top, you get column three. And so with that full momentum correction, the Coast Morpho 2D model matches up pretty well with full Navier-Stokes solvers like Delf3D, which is column one here. So that's the tidal component. This model is partially integrated to LandLab. The tidal model flow model functions. It's not yet a core LandLab component. So you have to do some separate insulation. But sediment transport and morphology is not yet included. And so the thing that I'm presenting is actually just one of the pieces that our group worked on. Another was trying to get some of the morphological components into the tidal model. So I'm not gonna talk about it, but there is information on it if you're interested. All right, so that was one of the components. The second is GS tools. And this allows you to create a random field with specific properties. So you can use like a Gaussian model and give it a length scale, and it'll give you the size of vegetated and un-vegetated patches. So that's like what I've got in the top right here. You've got other model options like exponential or return. I don't even know how to say that. You can also do things like specify anisotropy and rotation, and when you do that, you get something that looks like this. And then the third was the particle tracking component. And so this simulates the transport of passive particles in a flow field in a Lagrangian fashion. That was built by Jay. And so we combine these pieces and that allows us to do particle tracking through a random field in tidal flows. All right, so here's the Jupyter notebook. The first thing you gotta do is install some basic libraries. And then we're gonna set a number of parameters that are needed for this model. And in particular, I want you to note the roughnesses. And so the low roughness will be the channel regions and the high roughness is the vegetated regions. Then we generate a random field. So this is the GS tools component. And so when we set the length scale to 20 here, that gives us a vegetated region in yellow and a channel region in blue that we can then use to spit into the model. Then we take that random field generated by GS tools and turn it into a grid for land lab. And so there's a little bit of conversion here. And here is where we set the roughnesses between just the zeros and ones that the GS tools spits out and actually assign a roughness value. And so then when we've got the land lab component for the tidal flow calculator, we instantiate it and run it. And then finally, once we've got the land lab model up and running, we add the particles and run them. And so then we can run this through a number of time steps in order to look at the output. And so the visualizations for this, we can look at the velocity magnitude on the ebb tide and flood tide, as well as how the particles are moving through the system. And so I'm gonna show you a couple examples of this and I wanna just call out the roughnesses and then the length scale as well as a couple other parameters. And so I just did a couple examples to show you how it looks. And so for this one, I varied the length scale between different runs. Oh, and so flow is allowed through the bottom and the top boundaries and the left and the right boundaries are solid walls here. And so our tidal flow is moving up and down here. And so the blue regions, like I said, our channels, the yellow regions are vegetated. And after 50 tidal cycles, the particles move from the blue location where they started to the red location at the end of this period. And so you can see when we allow the length scale to increase, we've got more dispersion through the same time period with the same forcing. Something else you could try with it is varying the roughnesses. And so that one that I just showed you is here down in the lower right, if you move above it, we've decreased the roughness in the vegetated region. So like thinner plants, as we move over to the left, ooh, as we move over to the left, there's a lower roughness in the channel region. So that's saying like it's a smoother channel. And then down at the bottom is a smoother channel but the same roughness in the vegetated regions. And you can see here that when you just try these runs, actually allowing a smoother channel will allow for more dispersion, but changing the roughness between less rough, less vegetated, or more vegetated doesn't actually change dispersion. So it's really tied to how rough the channel is, which is kind of a nice test. And then one more example is allowing anisotropy to vary. So in this first region, there's no anisotropy. We've just got a length scale of 10. In this region, we've now added some anisotropy. So there's a length scale of 10 in one direction and three in the other. And then here we're allowing some rotation. So we've changed the direction of the anisotropy to pi over nine here, or pi over two in the bottom right. And so that rotation angle defines the direction of dispersion. You can see it's sort of north-south directed in this first figure, sort of along the diagonal in this figure. And here it's a little more evenly spread but almost along that direction of rotation. So with this model, you could look at things like the impact of oil spills or like seed and larval dispersal or fine like non-settling sediment transport. You'd think about how the vegetation scale impacts particle spread. Like what's the role of vegetation density? You could think about region connectivity. So what parts of an estuary are easily connected or not at all connected? And if you wanna do any of this, all our stuff is on GitHub for our coastal team. So it's got the repository and the documentation. That's everything I got. Thank you, Rachel, for a really impressive set of simulations and like new capabilities that are like generated by this team. I'm gonna open it up for one question and then have other questions in the chat to give the last two teams a little bit more time to like get their stories out to you. Hopefully there's a person who wants to like chime in with one question. I see a question from Christian in the chat about do the particles have inertia and gravity? They don't have gravity, they're passive particles. I think they also don't have inertia. They're going along with the flow. So I think it's just like Lagrangian tracking of the particles rather than Eulerian. But Jay who built this could really talk more about it. And I think you can probably get in touch with him. Thank you, Rachel. I'm hoping that you guys keep monitoring the chat a little bit. I see there's a few different... Oh, Greg still has a question. But we'll move on to the project team four. It goes to a completely different scale of geology and time. Gustav Palisgard Olsen will be presenting. He's at Arhus University. So like phoned in from Europe or zoomed in from Europe. And their topic is using landlabs to model tectonic activities in the landscape evolution model. Yep. So I'll see if I can make this work. Okay, so I have a couple of screens. Does it seems to work? Yeah, we're seeing in direct. Okay, cool. Yeah, so my name is Gustav. And I'm a PhD student at the University of... At Ols University in Denmark. And the co-authors for this project was Liang Xiuhe and Krishian, Xiaoni Hu, Eyal Marder and myself. And our project was called using landlabs to model tectonic activities in a landscape evolution model. So, yeah. So our overall goal was really not to answer a specific research question was more to, from the beginning to create a tutorial for undergraduate students. So we're looking at how to model different simple tectonic activities in a landscape using landlab and how to analyze some key geomorphological features with landlab tools. And like a basic outline of this tutorial goes through these four sections. So we started out with creating a kind of a basic landscape evolution model which we use in the other sections. And then we go into investigating a little bit lithospheric flexure. And in the third section, we add faulting and different pathological domains to the model. And then in the last section, we use some of the built-in terrain analysis tools. We have in landlab to analyze a uplift event and also using the same basic model. So in the first section, we have a model setup where we have a 20 kilometer by 20 kilometer grid. We just start out with a random landscape. Then we use the, we use steam power, stream power erosion for the fluvial incision. And then we have just regular old diffusion erosion for the rest of the landscape. And then we have close boundaries in the West, East and North. And then we run them all through time. So this is an image of how it looks in 300,000 years. And to the right is just an image of sediment discharge or average erosion during this time where we see the nick points migrating reaching the drainage headwaters and then slowly eroding to stay in state. So that's the kind of basic setup for the rest of the sections. Yeah, and then we go have the students investigate lithospheric air flexure. So a very simple setup where the student gets to place a point load on the topography from before from the basic model and see the subsidence in the system. So it's a very basic 2D elastic model. And then we go into the third section. Here we add faulting to the same model. And also we have two different lithological domains with different rotability. So we can see, oh, so you can see in this image where we run the model for 100,000 years, we see kind of a slanted line here, a tilted line. So that's the fault. So it starts somewhere up here and then it migrates south. And then we also have this very clear change in lithology or rotability. It's also, I mean, maybe slightly visible over here. Not really good. So when the model runs, we see the fault propagating and then also there's a low line created down here. And then the student gets to look at these through the different terrain analysis tools or the use of these terrain analysis tools like a CHI index, Deepness Index is kind of explained in this section. And then in the next section they get to play around with that themselves. So here we have a, again, the basic model. So it's run for these 300,000 years. Then we have chosen just some channel we're looking at. And then a spontaneous uplift of 50 meters happens and we run the model until it reaches steady state. So the student gets to look at different stuff like CHI index, Deepness Index, some channel slope drainage area plots to kind of analyze the evolution of the landscape through time. So we see here, it's very simple, nick point migration. So it starts with a steady state and then through some different time slices we see here in the CHI index, it's a nick point migrating through the system. And then after 200,000 years or so, we have, again, just a steady state landscape. So that was kind of how we built this tutorial. And then we of course have the students write a little bit of code themselves or kind of modifying the existing code and also analyze some of the, some plots of some of these CHI index, Deepness Index graphs and also create some of them themselves. And then in the end, the meaning of the exercises that they can themselves try to create some scenario, a tectonic scenario, and then just copy some of the things from the different sections and try to create their own model and analyze it themselves. Yeah, so I guess thank you. Thank you Gustav. I can certainly see faculty being excited about having this as a tutorial that they can tap into or TAs who are like using these to do things with people. Is there a question for Gustav and the team that works on tectonics? I have a little question. So I agree. This is all very nice and excellent teaching materials. So thanks a lot to all of you. Gustav, if you plot sediment discharge, is that actual sediment discharge measured at an outlet or is it average erosion over the catchment? Yeah, it was just average erosion really. So we have a couple of different plots and we're getting the students to plot them. But yeah, it's really just average erosion. Yeah. We'll take, if somebody has an additional question, they can type it in a chat and we'll leave it to the team members who are here to like help answer those questions. And we'll move to the last team that is a small team. And it's Zenming Wu we'll be presenting on land geomorphology evolution over continuous permafrost region. So a bit more specific to the Arctic applications or Tibetan plateau applications. Yeah, can you see my screen? Yes, we can. Okay, thank you. You can, yeah, excellent. Yeah, thank you. That's great. So this is the last presentation. It's made by Fen and Mei. So it's, we have very small team just to us. So this is about the permafrost erosion over a very small catchment by coupling Q-Model and the diffusion model. So this is overview of what we have done. So the key point here is to, to do some permafrost erosion projection by applying Q-Model and the diffusion model. So we made a Jupy notebook, but we still didn't upload it to the GitHub response. We're sorry. This is the main contribution of we have done. It's, we made the Q-Model to be a 2D model. The previous example is only one day. I will explain it later. So this is the whole process chain of the coupling of Q-Model and the diffusion model. Let me give a brief introduction about the starting area. So we selected a very small catchment in the non-CBRI where it's underlaid by, sorry, where it's underlaid by continuous permafrost. We can see the, we can see it from the left figure and the right figure shows the top graphic details of the starting area. So this is the flow chart of the Q-Model application. So we need to get the active layer thickness first because an active layer thickness is one of key characteristics of the permafrost. It can be used to indicate the state of permafrost. To run the Q-Model, we need temperature, amplitude of temperature and the snow depth and other some information. So here we used the one climate station in the starting area to interplot all other informations of temperature and amplitude temperature, the snow depth of the each pixel. So we know previous Q-Model example only use one set of parameter over the whole area. But in this study, we assume the starting area as a roster so which is made by many, many pixels. Then the active layer thickness of each pixel is calculated by applying Q-Model with specific input parameters. So this is after we get the active layer thickness by applying the Q-Model. So then we need to input them into the diffusion model. But the main problem here is we need to think how much of soil is erudible over the permafrost region. So the left figure is the structure of the permafrost. We can see the top layer is active layer and the middle one is permafrost which is now which is generous frozen and quite stable. And so here we take, we take, so in the active layer is regarded as erudible. So after solve this problem, we need to think about Dm which is which is a necessary parameter for the diffusion model to get the slope information. So we also need other, sorry, so we also need other information to run the diffusion model such as sorry, generation and diffusivity. Okay, this is the result of active layer estimation of this small catchment. So we can see the over the high end code region in the east part of the study error. We can see the quite shallow active layer and in the south this area is lower warm and quite flat. So we can see the quite deep active layer estimation. So after we get the active layer thickness, then we put them into the diffusion model. So we get the error map of the study error so this one is the error map. And if you look at the river back of this error map, we can see the sedimentation in the rear beds. Okay, we also did some further investigation about the effect of vegetative coverage or the error rate. So we can see the left two figures is about the vegetation coverage from Google map not 80s and the 2020. And the red error is in the error map, we can see the change of the error rate. And this is the simple analysis about the active layer thickness and aspect. But to be honest, there's no correlation between them. And the active layer thickness, we estimate in this study area is more than one meter. And this is also a simple analysis of error rate and aspect. But we can see statistically the slope error with aspect between zero to 50 degrees shows a slight accumulation. So this is our conclusions. The first one is POMPRA study rate error estimation process with 2D Q-mode and the diffusion model was built. And also in the future, if you want to use those two couple of models, you can consider to use the multi-process tail in Python. It's because this 2D Q-mode we built is quite computation consuming. It's very slow for the 2D calculation. And the third one is you can input more accurate parameters. For example, temperature and the snow depth to you to get accurate result. The temperature data and the snow depth data we used here is only from one climate station, station which is not accurate. So which may bias the real result we get. So thanks for your listening. And if you've got any question about all this presentation, please contact us. Thank you guys. Thank you, Xenming. I wanted to compliment the two of you. You are like the smallest team of our, of our all our teams in the Espen summary institutes for like making actually like a real new contribution to this set of modeling by making everything spatially variable and starting to test that against the real case. So like you are like maybe too modest for, in your presentation about your accomplishments. I'm opening it up for a question or two. Hey, Irina. Yes. I have a slightly broader question. Yes, go for it. And this is kind of coming from ignorance. Is there a place, a journal, where these could be published? There is a journal of educational open source resources. Oh, right. Isn't that like hope? Yeah, right. It's like Josh, but then it's for educational too. Right, Jose or something, yeah. Yeah. So maybe. Yeah. I mean, or at the very least like when, when everyone's finished, you know, we should make sure that everyone has a DOI on their repository so that, you know, all the Espenites can take credit for their work and have a sightable reference. Yeah, and I also want to emphasize to everyone, like make sure you, like if we build this into the CSDMS Wiki website, that we have your names on there because it's an educational resource that you build and you can point to it. And if you, if somebody asks you ever like, oh, what did you ever work on teaching material or whatever, then you can say like, well, here it is. And it's being used by a larger community. But that DOI idea is a good idea. And I can definitely send to everyone the link to that journal site. It's new, and yeah, I don't have much experience with it. So I can't vouch for it or anything, but it might be a good idea to scope out. Espenite. Josie is saying we should be, like all of you are Espenites now. Well, Mark used the term. I just think it's more than anything. Okay. Are there any last more general comments that people wanted to make? I wanna keep to time because I know like many of you like have like zooms back to back to back. I see some really complimentary comments. Amazing work. People are very impressed. And that's true for like our entire instructor team too. It really made a big jump still between us finishing up in August and now a few weeks later, a few not even a couple months, but like a few months. Yeah, I just wanna acknowledge that we had a really intense six days of teaching, but still the products I think went way beyond those days of teaching. So there was a lot of work and learning on their own. So this is really inspiring. I think for what our community can and should be doing. Cool. With that, I'm gonna conclude. I wanted to thank the teams for all the work that they put in. I wanted to thank the presenters explicitly for stepping up and being the voice for their teams. And I wanted to thank everybody who's like listening in and we're like hopeful that this is either an inspiration to you to like partake in Espen in the next year or it's an inspiration to use some of these tools or it is a resource that you can pull from and using classes or in teaching moments that you have yourselves. So thank you everyone. We were quite a number of people here.