 So this is an open GitHub repository. It's a CSDMS Espin repository. These notebooks live there for people in the community to pull from too as examples. Some of them may like undergo like a little bit of improvement still, but the teams will explain to you what their contributions are. So teams, no flow. Yeah. Well, thank you so much. And I guess I'll start off by thanking the CSDMS team as well as Espin who really delivered a great learning experience for all of us and really helped us through. And also thank you to Jian who's model we use and she was also incredibly helpful. So we are the teams, no flow. So we are kind of like this Pokemon up here. So in this group, I am Kartike and I'm going to start this off. Other team members in the team, we have Carlos, Alex, Lauren and Katelyn. And so basically when we started off this model or what was assigned to us sort of was to assess kind of the flood risk from the overland flow model. But just like my PhD, we started off with something and sort of ended up with something else. So what we ended up doing was kind of include another input into the overland flow model and which was to include the snowmelt. And to calculate that snowmelt factor, we wanted to do it as a function of temperature. So we included a degree day factor, DDF, as a general temperature over the entire basin and then calculate the amount of snowmelt that will be generated and contribute to the overland flow. So we have a number of steps here that we used or that we kind of followed. And I'm now going to ask my fellow member Alex to come forward and speak through the steps. Yeah, so the first step, like with all programming is to initialize your stuff. What you do, we use an API key to get a digital elevation model from Open Topography. This is a nice open source, open access like LiDAR dataset. Here are our libraries, you can read those later if you want. Here's the folders we made. So the first main, so the actual first step is to actually download a dataset from Open Topography. We looked at a little valley near Boulder, south of a city called Jamestown, Colorado. It had a fire a few years ago. So the original notebook was looking at how that postfire, let's look at postfire landslides and overland flow. Here it is. And now next we have to generate the snowmelt. Thank you. Hi everyone, my name is Lauren. So one of the first things that we had to do as a team was determine snowmelt. So our team looked at a number of equations online related to positive degree days, no melting. And we found this equation that is within our notebook, melt rate. So that's DDF times temperature. DDF typically ranges from about three to five millimeters per day Kelvin. We decided to pick four right in the middle. Can't go wrong with that. And so we had to convert temperature. So again, this is an educational Jupyter notebook to be used by undergraduates, for example. So I won't go into the details, but essentially we made a function that shows students how to convert and obtain that melt rate factor. Yeah, so more steps related to that. I will add that for this particular notebook we use three cases of melt. So in order to show students what that would look like in an array, it prints out an array at the end as long as they follow those steps. So that is it for us. I'll add in Carlos, who's going to give us a few comments. Thank you. So a couple of comments about this model. So the idea of this overland flow is to you have the option to add some precipitation rate. So we consider also the how the snow is melting, how this precipitation rate changed. So we are actually, we are adding the effect of the rain and the snow melt. So it's a very simple equation for calculating the snow melt. A couple of comments here. Obviously the question comes from energy surface balance where it is calculating using the net radiation, the lighting heat flows, but the more important factor after reading a couple of pages in a book, we find out these, the latent heat flows is the more important factor to determine this melt rate. And they provided us very simple equation. Actually it's only like in a factor of lowering its plane and you need to select or choose a temperature. That is a temperature there above the zero Celsius degrees. So after we do all that, generate all the snow melt, we have to figure out actually how to get a watershed for the snow to flow through. Which we do in this step, you can see a little like post-fire picture of the area. Just set up a raster model grid and then we do flow routing. And our watershed will come out of this point down here. It's also the largest watershed in the region. But you can see map of here and actually just have to calculate overland flow. Hi, I'm Caitlin and I'm gonna talk about the simulations that we did with overland flow. So now that we have our watershed and we have several different melt rates and we also added in a bonus simulation where we said what if snow was melting but there was also precipitation happening at the same time. So with that we created five different scenarios. Each of these scenarios runs for 200 minutes and for the entire duration of those 200 minutes there is either a heavy rainstorm happening. There's a heavy rainstorm with snow melt. So we take the snow melt calculated above I think with like the 0.5 degree Celsius case and we add that melt rate to the precipitation rate. And so that's our combined case, the second one. And then we did three cases where it's a sunny day there's no precipitation but we just have the melt rates that we computed above. So those were like 45 millimeter, 46 millimeter simulations. So we ran all those in a for loop. Yeah, right. So these values we have our three cases. We have just rain, rain and the 45.58 melt rate and then the three different melt rates that would occur at those different temperatures. And then we can visualize our results. So in this first plot you can see just the rain, the blue line and then rain with the additional snow melt. And you can see at the end of the 200 minutes it starts to reach like a steady state. So we're like turning on the faucet of melt and precipitation and then over time that like reaches like a steady state flow at the outlet. And then our second plot compares our different melt scenarios on discharge. And so I think our main findings here was that although increasing the temperature does increase the amount of discharge, it's not by much like having something like precipitation and a warm day is much more likely to result in a flood than having just a slightly warmer day. Okay, and then we thought, okay, well having a warm afternoon that could happen but usually like during spring melts you get like many hours in a row that are above freezing. So we decided to run the model for 24 hours with 12 hours of melt. And these videos don't embed on GitHub but I tried this and if it doesn't work we'll just let it go. I don't know how to play it, that was fun. So you can see a plot similar to those first three hours where you get that initial peak and then it reaches steady state. It melts all day at that steady state and then as it cools off in the evening the discharge plummets. And we had a bunch of other, this one that I'm showing is for the 45 millimeter per hour melt rate. So like just above freezing day but we ran a bunch of other different ones with much higher melt rates and you can find those in our GitHub repository. So just to kind of dive in around that. So we kind of modeled under the snow melt contribution to the overland flow under different conditions. On the shorter one we show that adding snow melt to the rainfall significantly increases the peak discharge. It also reduces the lag time highlighted by the left shift that we saw in the hydrograph. We also isolated only the snow melt component as well and we showed that the temperature didn't really change that much like in terms of the peak discharge. The slight changes with the higher temperature kind of highlights I think the likelihood of generating more melt quicker with higher temperatures. Sort of to preempt the questions that we're gonna get from the room. We added what we could do. So we do understand that there can be special variations especially with differences in temperature and elevation over the basin. There would also be temporal variations especially over the longer model runs. We can switch to an hexagonal grid. We'll probably increase the accuracy. And our original kind of aim to do flood vulnerability may be next time around when we come to see the MF we have that, thank you. Yeah, I can take that one. So that is a really high amount of real melting but that is the amount like during a really heavy, really heavy rainstorm. In the model that we borrowed from Tian she had 55.2 as her like heavy rainstorm amount and in her simulation she was only doing that for 10 minutes so that's like a deluge for 10 minutes. But we thought what if that happened for a whole entire hour so that's 55 millimeters right there. And then in our snow melt model if we have a temperature of like just above freezing that would generate a melt rate of 45 millimeters per hour. So we just added 45 to 55. So you will see from these presentations that like the people coming in actually with quiet variable backgrounds and disciplinary knowledge as well. So the next team will be talking about landslides and yeah, please come on up. I'm a Shawnee Long lead and I need to ask for a project. But before, sorry, but before I begin I wanna thank the CSDMS team on all our mentors throughout the week in Espen. Sorry, we're having issues. Okay, so our project is on landslides susceptibility calculations in Tuscany, Italy. I want to first mention that or landslide susceptibility project is modified from Tien's existing landslide susceptibility use case. So it starts off with the very basics. So what is landslides susceptibility? Landslides susceptibility is the likelihood that a landslide will occur based on certain conditions. And the conditions we chose to focus on was root cohesion and also soil saturation values. Here we present a Jupiter notebook that uses similar concepts to the Snowmelt team where we download topographic and soil data from the CSDMS data components. And our next step is to calculate landslide susceptibility with root cohesion values. And then to calculate landslide susceptibility with root cohesion value and also variation in soil saturation scenarios. So why Tuscany, why Italy? We chose to focus on Tuscany, Italy because one of our group members is doing research in Tuscany, Italy. And we chose to apply this model to a real life scenario of heavy rainfall period in September of 2017. Hello, everyone. My name's Selena. I'm at the University of Oregon. Next, as you heard from the previous group that, these are labs, so we focused on learning objections and key concepts. And again, like Ashani said, we focused on changing root cohesion using values we got from literature. And then the second part of that, we varied both and we plotted some really nice susceptibility maps. And then from there, we want to focus on students really learning how to use these data components from CSCMS, LandLab, all that good stuff. And then, like I said, it's very much organized in a lab setting, but we go through getting API keys so they can access the data components for open topography, excuse me. And then get them to load up their packages, great folders, where to save things. And yeah, and now on to the next. Thank you. Hi, my name is Sharad. So we kind of, this is not what we prepared. Yeah. This is not what we prepared. This is from a paper like we found, but this is the actual area. And we kind of selected from like somewhere, somewhere up here in the north side of the Kuscany. We used whole like open topography data for like downloading the digital elevation model. We calculated, these are the all the details of related to DM. And then the DM TIF file has been brought up into like Raster Grid model from the LandLab. So all the further kind of, so this is in the LandLab and we also, we downloaded like soil depth data from soil grids, which is at 250 meter resolution. And then we calculated slope, slope grid, et cetera. Yeah. Yeah. And then because all the data components are from different spatial resolutions. So we used ESPY, like this is the weather related repository, Python package, which kind of provides a regridder function to kind of regrid all the data to a specific same resolution. So we, our DM is at 30 meter resolution. So all the data component, all the data has been re-sampled to 30 meter using this re-gridding function. Yeah. Now, Estefania, I will talk about it. Yeah. I'm Estefania. So the next part is the core of like what the students should be looking at that it's lens looking susceptibility. So first of all, we defined what the factor of safety is. So we used like the formulation by Bill 1995, but doesn't have like these clumped cohesion that, and it separates root cohesion from soil cohesion. And we're doing that because we want to see, like use these root cohesion as a proxy for vegetation cover. So yeah, like overall, like this explains to students how to do that. And the susceptibility is defined here as one over the safety, the factor of safety. Yeah, we have had comments about that. And like numerically, like if we only do the factor of safety, we get a lot of errors. So we're using the susceptibility as these. I might be confusing because the students are learning like factor of safety as resisting forces over driving forces. So it's between zero and one. And then the susceptibility is exactly the opposite, like the numbers. But anyway, keep in mind we're looking at susceptibility. So we are varying root cohesion. Here we're just defining the plot thing. Here we just defined these ranges of values for root cohesion that we found in the Tuscany area between 3,000 and 12,000. The units are pascals times kilograms over meter squared seconds. And yeah, we just, we're just changing different root cohesion values. And the results are these like for really low root cohesion is just only changing root cohesion. We see that there's a lot like within this landscape, there's a lot of susceptibility for most of the area. And as root cohesion increases, it starts to obviously decrease and the landscape is more stable. And finally we get like to the highest value we get for root cohesion that that would mean like a pretty solid vegetation cover. And we only have like really high areas that probably it's driven by slope. And yeah, now Sara. Hey, I'm Sara, I'm from Vanderbilt. So the next part we were kind of thinking in a pedagogical way like obviously from that equation above there's multiple variables playing with one another. So we can do a really basic kind of showing how two variables might be competing or working together. So we're varying root cohesion with soil saturation. And we keep it in a really simple for loop just stepping through a saturation of zero saturation of 50% and then a full saturation where your water depth is the same as your soil depth just to show that root cohesion can't actually overcome different parameters. So it's in a simple for loop and also here kind of doing a lot nicer plotting to show students like an example of how you can do kind of a pretty complicated plot. So here we have the rows are root cohesion values and the columns are saturation values. And you can pretty nicely see there's a competition where so low saturation and low root cohesion very little areas are experiencing high susceptibility but as we increase saturation a lot of the landscape will be susceptible especially at those higher slopes. And then as we increase saturation and increase root cohesion that root cohesion is balancing with the saturation but we still have increased area that has a high susceptibility. Yeah, okay. And then we just wanted to kind of make it easier for people to see relative to the actual landscape. So we added in our raster hill shading kind of doing the same thing again this time just using a different type of grid. And here we see the actual relief with those areas mapped on top. Yeah, and similar but we didn't put it in the gridding space. And also like there is an option like we have not run it we have kept it like kind of student experience like later on, experiment later on. So we can also download the era five data and we are using like total moisture there and soil moisture at different depth. So from this, and at last like the data is 0.25 degree plus 0.25 degrees. So we need to read this data and just need to run the code above to kind of get the susceptibility in terms of in case of real time like range of conditions. Yeah. So these are the references and just for the sake of students what we have created like just to see that okay there is a fact of regretting. So we have created a repository like files showing that okay if you will use such like this is your original data and if you are kind of using regretting different kind of so what will be the effect of regretting on the final outcomes. This is just that and our kind of repositories is very much complete. It contains like literally everything starting from key concepts, literature, study the description properties and like even inventory of the landslide locations and thank you. Our next team, two teams, they were all interested in effects of vegetation and ecology on landscapes but they were a very large team and they had different interests so they broke into sort of two like split into two branches and so we'll hear first from the team that works on like river or worked on river meanders and vegetation effects. This one for sure. Okay. Oh, this is loud. Hi everyone. My name is Nicole Hookay. I'm from the University of Idaho. Hello everyone. I'm Noxian from Penn State. I'm Nick Korak from Wake Forest University. I'm doing my PhD at the University of Oxford. So first I wanna congratulate the first two groups that came before us. Those are really, really awesome work. Ours is a little bit shorter. So a little bit humbled. We are the migrators. We chose this group name because we were interested in the effects of vegetation and river migration and mandering. So just like as a brief introduction to what we have been doing is that we know that channel migration is controlled by so many factors such as channel morphology, flow conditions, sediment transport, but it also is controlled by bank resistance. And so plants, vegetation densities make the banks more resistant, hence harder to erode. So we wanted to demonstrate this through a numerical model using a Python module called Manderpy authored by Zoltan Sylvester. I'm sorry if I'm mispronouncing that, which was based on this very simple model from the paper of Howard and Knudsen in 1984, which basically does a very simple link of the curvature of the river. So the original morphology and the migration rate. So first I'm gonna give a demonstration of the original Manderpy module. These are the input parameters. It has a ton of assumptions such as like constant width, the depth and it only changes the migration rates laterally. So first we're gonna do a example of a low vegetation density. And the way that we wanted to account for this vegetation is through the friction factor in this case, the Chizzy coefficient. Because we know they're not directly related but we know that when we have higher roughness they're gonna be slower velocities. And so erosion rates are going to be lower. So for a low vegetation density we're assuming that we're gonna have higher rotability. So a higher Chizzy factor. And this is how we initialize the model, we run it and this is one of the results that we get. So we can see that it has a lot of cut-offs and a lot of oxpose which represent basically where the river has been historically. So if you're on the same one changing the Chizzy factor now to a higher vegetation density or a lower erodibility if you will. We can see now that we have much less oxpose and a more stable, or not stable channel but you can see what I mean. Anyways. So in order to show how this meandering rivers moves across the landscape we have prepared a small video. So here you can see the blue dark line is the actual current active river and the light blue river actually light blue parts are actually the cut-offs made by the river. So here we have used a higher Chizzy faction that are higher factor values that indicates higher erodibility. That means low vegetation and low stability of the bank. So that generates lots of cut-offs as you can see in both sides. So for the last couple of examples we've had one Chizzy factor. So our first mission was originally going to be create like a varied landscape and have multiple types of factors throughout the grid. But first we needed to start with the same if we could implement two different Chizzy factors. One on the North Bank and one on the South Bank. And so we began this by forking the repository and then creating our own branches and started editing code. And here we have two different types of vegetation densities on the North and South Bank. So we have two different Chizzy factors and after running this model simulation you can see we have lower vegetation density and therefore more erodibility on the North Bank. And so you can see a lot more cut-offs than these blue markings here. And then on the South side of the river we had a higher vegetation density and so there were less cut-offs. And we wanted to maybe make this even more varied but we'll get to those issues in a little bit. So for right now we can just consider everything on the North Bank one unified type of vegetation and on the South Bank a different kind of vegetation. Thank you. Yeah, so and looking at that plot I'm sure we are not the only people in this room who find it very unsatisfactory to say, oh yeah, the North Bank looks kind of different than the South Bank, we like numbers. And so we thought about how can we kind of use metrics to compare the North Bank with the lower vegetation density to the South Bank. And a very obvious one that we created this little cell for is to just calculate the number of cut-offs that we have in the North versus the cut-offs in the South. And so what we're doing is we're iterating through the cut-off objects and looking at the coordinates and if the coordinates were mainly in the North and then those would be calculated for the North Bank and if they were mainly in the South they would be calculating for the South Bank. And so looking at that plot again, so for this particular plot what we got was 126 cut-offs in the North versus 27 in the South which is I guess pretty distinctive difference. And then as Nick has already said we would actually like this to be a bit more complicated even because it's not very realistic I guess that we have one kind of vegetation on the North Bank and one kind of vegetation on the South Bank that are very distinctive and not varied at all. So the next step, I'm not sure if we're going to continue on this project but if we ever did we would want to create some kind of more realistic vegetation with maybe random vegetation patches of say three different kinds for example, shrubs versus trees versus grasses. And we've already started to create nice grits but we haven't managed to implement that into the model. Yeah and I think that's all from us if you have any questions we're ready to answer them. So that was one of the difficulties is like can we evolve the Shezzie factor to move with the, we were unable to implement that so it's kind of relative to what was existent in the grid on the North half and the South half. That's correct. Yeah the Shezzie factor it was implemented just as a kind of a constant value throughout and so now we've had at least the spatial component to it so the next group of S-men maybe can build on that. So the other vegetation group will be presenting their codes and you'll see that they took a very different approach so it will be something very different for students to look at. Good morning everybody. I'm Madoche and my group will be presenting vegetation dynamics. You know, normally or traditionally how we model flow resistance it's just assigning a constant vegetation but in some decades there is a lot of effort to see how we can actually capture the impact of vegetation and flow resistance and there are some models like XBH of DELF-3D that is applying the approach of Baptists where you have two flow wedging submerged and after that you have a more logarithmic profile when the water depth it's higher than the vegetation height. The main characteristics that we use to module vegetation it's as a cylinder so we take stem density number of stem per square meter, stem height and stem diameter and you may see that this is the expression from Baptists this is the value of GAC based on vegetation dynamics and here this is the three parameters it is vegetation height, density and some kind of drug coefficient also and you may see we'll have like two value so the second hand of that equation if that value of the logarithmic it's lower than one we'll have negative value so we need to separate that formula in let's say two size depends on if the vegetation height it's lower or higher we'll be using this later on in the model in the let's say implementation of that Baptists or Mula I will explain that. So the main objective was to see how using a constant roughness and using the varied roughness impacts the discharge at a given outlet or overall water depth and for our study we first input our we imported a bunch of libraries and then set up our mortal roster and decided to keep this simple and use a test basin that's provided land lab and for the simulation we closed all the boundaries but introduced one node where it would have discharge flowing out from the simulated area. So for the constant monies we defined a function that takes one value which is the monies and value so for each vegetation type or simulating types of vegetation, grass, shrubs and trees the user can define the monies and based on literature and in this function we redefine the greed and user can enter the end. One other thing that we introduced was a rainfall intensity so there's a constant rainfall intensity for the simulation and within the same function we have two lists that store the hydrograph data so the time and the discharge at the end of the simulation which runs for 500 seconds we also get the overall water depth at the end of the simulation so at 500 seconds we calculate the water depth. Okay, so this is the Baptist formula we just built our class so like I mentioned you and that equation we saw above if the second part of the right hand side it's being negative we don't use it and we put a conditional statement if the vegetation height it's higher than the water depth we will just be using only the first part of the second hand of that equation because if we don't do that we would have the case of negative chase it. So we use that approach and we will have our spatial temporal viability of the roofness which varies in time with the water depth and with varies also and space depends on the conditions. Here I have to say that I just we just only assume one vegetation density for the world domain but in a real world case we would have to make out of discretization depends on each let's say polygons of vegetation to define those kind of characteristics. All right, so we next built a function that would implement this formula of allowing for a variable roughness based on the vegetation and also the water depth. So it starts pretty much the same as our previous function we set up some lists to store our data we set up the time step and then we set up our grid and the same but then we add a roughness at the so we first add it at the node and then we translate it to links and then when we initialize the overland flow model we use that field of links to define our roughness and then we loop through our time step but then we also loop through every link and we use our two formulas to calculate a new roughness and then we add that to the field and it goes on and on. Yeah, so we ran three different scenarios well actually six one with the constant M and one with the variable Manning's N and we chose grass, shrubs and trees and this is the input vegetation information so you'll notice it shows the same Manning's N for shrubs and trees. So here is how we ran our... Hi, I'm Dominique, let's show you the results. So the first thing that we wanted to do was put like a hydrograph of a final output point on our artificial DEM so here you can see we've got grass, shrubs and trees but you can only see two lines and that's like Ira said we only use the same Manning's N for shrubs and trees which was just sake of the like and actually ran with trees. The next thing that we did was put again but using our dynamic flow resistance so using the Shazie number which was sped to Manning's N and we've got a slightly different graph so you can see we still have the grass and the shrubs so the shrubs are rougher so they're resulting in a slower discharge but if you have a look at trees trees are doing something weird and trees are doing something weird because we plotted them as or we inputted them as 10 meter elevation and we were like, oh, that'll be fine but as you'll see in a minute when we plot the water surface height we actually have water depth of 20 meters so these are really rough features which you can see in the instability at the beginning of the graph so yeah, so and just to plot them alongside each other you can see that there's a bit of a reduction in discharge rate with our varying Manning's N so yeah, so just to sort of like explore that weirdness of the trees a little bit more we plotted the discharge rate against the Manning's N and so the first thing that's wrong with this graph is that the Manning's N value is really, really large for all of them, especially trees but what's right with this graph if you look at the grass and will ignore trees for now but the grass and the shrubs there is a negative relationship between Manning's N and the discharge, which suggests like as water water elevations are increasing we're getting less flow resistance which is what we'd expect, which is good so yeah, so then we've just got a plot on the water depth if we had a little bit more time this is just for the constant N for grass we would have also liked to have plot constant N for shrubs and trees and then also done the varying N and then subtracted those so we could have got a difference but we sort of just run out of time but something that we will improve for them and then kind of yeah, just to finish off like so what like this is just some figures though yeah, like it's basically just some figures showing that the Mike Snips revived the internet showing that we're really interested in roughness like why have we done a why have we done like a workbook on this it's so like students can understand like the importance of like varying roughness there's a huge amount of research in like how like vegetation can attenuate waves and like how it can change like bank erosion like the other overland playgroup and yeah, so yeah yeah, thank you very much any questions so this year we had a few people interested in glaciology and glacier systems and so we quickly assemble the team glaciers that's not a traditional a traditional like set of notebooks that they could pull off like very quickly so they had to be a little bit creative with what kind of problems they wanted to start on and work on and I'll let you tell them thank you, Arita but yeah, just to echo that we're like one of the first glacier groups and so we like to thank Arina and Ethan for guiding us toward something feasible and also encouraging us to do something really cool and fun and so this is something targeted toward grad students or also undergrads and understanding how meltwater will change over different kinds of surface topography and what the resulting meltwater production and for presence is like and so I'd like our team members to introduce themselves and their glacier interests so my name is David I am interested on hydrology I am from Wake Forest University I am Emily I go to Columbia and I'm interested in ice shelf stability in Antarctica Hello, I'm Noreen I'm from SUNY Buffalo I am working on crevasse section on the Greenland Ice Sheet Hi, I'm Jonas I'm from Simon Fraser and I work on periglacial landforms so don't like glaciers that much actually but and I'm Mike Hala I'm from the University of Texas at Austin and I look at glacial sediment transport under glaciers and so to start us off I wanted to start with this image especially for those that might not think about glaciers all the time like us so this is a glacier in Banff Banff National Park and this is showing like just a large amount of water to charge on glacier surfaces and so why do we care about surface hydrology about a third of our entire population lives within glacier water resources and so it's super important to know when where and why these this water content will be released from glaciers and also if you haven't been convinced from Alex's talk yesterday it's super important to understand this very complex system especially looking at this figure and how much water is coming from glaciers and also just understanding climate and stability in the future and evolution of these glaciers so super important I hope you can agree to and so yeah a lot of our motivation is to use our approach to understand the different kinds of surface topography starting with something synthetic or idealized that our team created and then understanding it in a real glacier case and so Emily will start us off with a synthetic one and then we'll go into one case for a real glacier awesome thanks Michaela hi everyone so yeah of course we started with our imports the only part I'll talk about here is that we created a glacier surface flow class that we imported here that incorporates a few of the different land lab modules and then to start off we wanted to run our flow accumulator tool on a channel synthetic glacier so we tested this on a few different synthetic glaciers we made a on parabola we made a nidome but for the purposes of today we're just going to talk about our channel topography so this is what it looks like it's just a simple two-dimensional sloped topography and then we kind of incise this idealized channel through the center of the topography to make sure that our flow accumulator tool was working so the very first thing we did is just a simple plot of the drainage area without considering differential meltwater production so this tool works by assuming the glacier is melting kind of at the same rate homogenously everywhere and then this is a plot of the relative drainage area represented by the blue to yellow color bar in each grid cell so as expected you can see as we move from the high part of the glacier down to the low part it's getting yellower there's more meltwater accumulating and then also in the channel we can see that there's a lot more meltwater than elsewhere on the glacier and then the white areas are of course showing the flow direction and this was created with the flow director D8 algorithm which simply looks at each grid cell and then looks at the eight adjacent grid cells and determines which way the water will flow based on the direction of steepest descent so then to add a little bit of realism into our model we wanted to incorporate heterogeneous melt production and we started with this kind of very extreme case of okay what if there is this one random grid cell that was melting at a rate of 10 times faster than everywhere else on the glacier what would this mean for a relative flow accumulation and you can see here with the same channel topography it kind of worked as expected so we still have our channel with the most meltwater accumulation but then you can also see the downstream of where we're producing extra melt there's extra melt accumulation as you would expect and then one more step we wanted to take to try and make our model a little more realistic was to incorporate the presence of fern so for those that like Kayla said fern is essentially snow that has fallen on top of a glacier but hasn't had time to compact or densify into glacial ice yet so this means that there's a lot of air content within fern where meltwater can kind of percolate into and refreeze and this is significant for glacial flow because if there's fern present on a glacier it will slow meltwater accumulation because instead of flowing downstream the water will instead just sink into the sink beneath the surface so Jonas is going to talk about how we did this so just to kind of show in our idealised case why this might be important what we did was we created binary fern presence which is basically just above a certain elevation we said there is fern and below it there isn't so that's why you see here basically one at the high elevations no fern low and then the channel doesn't have any and we used the lossy flow accumulator which basically allows you to add a function that says that we're losing some amount of our discharge and just to, because we've really pumped up the numbers quite a bit we initially tried to like do some rough scaling based on Darcy's law and used that to get like a constant loss rate turns out with the numbers we chose for our synthetic case that meant we basically lose nothing so we just decided to add a couple of orders of magnitude and the picture then looks like this so we again get less but we still get most of ourselves water discharge in this channel that we created but this giant peak that we had initially you know actually see that downstream of that there isn't that much more meltwater accumulating because most of it would percolate into the fern here this kind of just shows why this is important and why people might want to like look into this so we now have a bit more realistic case so I'm going to tell you about the real case and we chose a glacier in the Andes it is a glacier on a strata volcano the name of this glacier is Chimorazzo and we downloaded the data for the digital elevation from the short radar topography mission and here it is it is a digital elevation model in a three-dimensional plot and then we just plotted it again in a two-dimensional plot to then match and identify the flow channels for this glacier and here you see the channels what we did is just max out all the flow channels that were sinking in the same spot and that was like those points with the range area of the 10% and below and then we tried to add production with the Aero5 and what we did to match the special resolution about is using the Gaussian filter here and that's something where actually things started going wrong because the model that we created before didn't show any changes in the topography of the real case and that's something that we have to work and improve for future projects and now Noreen will tell you the completion of next steps so in conclusion we explored how meltwater is routed across glaciers for synthetic case and a real-life glacier case we also visualized how differential meltwater production can impact glacial routing and we found that it's important to know where meltwater production sources are originating from or if they're localized since we see in the synthetic glacier case that meltwater is occurring meltwater occurs in the south side of the channel and we also found that the presence of fern can slow surface water accumulation and finally we provide an initial framework of including meltwater and fern processes for glaciers including the in the land lab infrastructure for future steps we would like to explore flow accumulator how it works for ice sheet surfaces we would also like to incorporate a more robust representation of the fern layer and we also want to add our classes to the land lab framework thank you I wanted to give a quick shout out to Tian who like helps the students a lot by like providing these data components to make their cases a little bit more like unique like the chimerazo like pooling topography from a specific place that some team member has an affiliation with or that is their research area and so I don't know like two more like assistant professors told me this morning that that was valuable to them and maybe to their classwork or et cetera but I think these notebooks show some of that too that there is fun in illustrating against real time data or like real data that comes from these data components with that sidetracked I wanted to like pull up ask the last team to like come forward this is a team that was interested in a bit longer time skills and they've been exploring the effects of rainfall patterns on mountain ranges hey guys we're the last group and we wanted to look at the effect of the or graphic effect on topography so as a bit of background your graphic effect is how basically rainfall changes as you move across the range so why this happens is that when you get there's a prevailing wind that usually most the weather is coming in from and as the air is forced over the mountain range it cools and releases a lot of precipitation and this causes an effect where you have an increase in precipitation with elevation that's the first thing that we're going to show you guys about and the second thing is that as the air continues to move across the crest and it releases all its precipitation it dries out and warms up as it moves down and this creates a rain shadow or a lack of precipitation on the far side of the mountain range so to preempt some comments here the way that we did this bottling was I would call it a thought experiment as opposed to realistic and we just wanted to play around to see what sort of topographies we could come up with so we put together a notebook that demonstrates how we incorporated the or graphic effect into the linear diffusion component and stream power to facilitate eroder components and so we developed a relationship relating precipitation to elevation and for the fascade eroder component we included this relationship in the water influx field and for the linear diffusion component we came up with kind of a linear relationship between the diffusivity constant and then we applied that to precipitation so first we created a synthetic grid with randomized variation in topography and then we applied these two components for a 60,000 year time scale and so in this first example we're showing constant precipitation so this is the example with no or graphic effect we're developing channels and hillslopes with the highest point in topography accumulating around the middle of the grid and the highest elevation shifting through time as the landscape is eroded and so this plot shows how we applied a constant precipitation without an or graphic effect and so as we can see from the resulting landscape that the river profiles are overall pretty smooth and concave up and that the elevation of the final topography shows kind of a east to west increase in elevation so now we apply the or graphic effect to our grid and so in this scenario precipitation increases as elevation increases so as you can see the point of highest elevation is moving around the grid this is the result of this feedback loop between increased elevation due to increased precipitation eroding and as a result we see interesting river profiles with kind of steeper steeper cutoffs and we also see the development of these mesas below the bridge lines to the point of highest elevation so now when we apply the rain shadow to our grid that excludes the or graphic effect so in this case where yeah so to apply the rain shadow we jury rigged it to basically just cut precipitation to a very low value at the other side of the ridge crest so that's what you're seeing here where there's a pretty abrupt change from high precipitation to low precipitation and it creates some sort of interesting topographies here where we have a very different north slope and a lot more diffuse on the north slope right and we can see a huge difference in the river profiles between the north slope and the south slope here where these jagged ones are being affected by the rain shadow as you can see here and some extremely wacky swath profiles here in our last simulation we're applying both the rain shadow and our or graphic precipitation role actually but we'll see if there's a difference here um yeah and we see in all of the simulations with a rain shadow that the divide migrates towards the area with the rain shadow and that is something that we do see in this year in Nevada our ridge is pretty heavily towards that side and we see the development in terms of topography kind of this high in front with kind of this flatter mesotopography and then a decrease within the windward side