 Our next speaker is John Palachiar. He's going to be talking about modeling impacts of vegetation changes on landscape pollution. I want to thank James for the invitation to come and speak. Thank you. I'm going to be provocative in my first slide and propose that the first low-primary morphologists have often taken climate change to be synonymous with changes in runoff, destroyed precipitation, some combination of those things, i.e. the driving, shear stresses of the landscape. And I'm going to show you some examples and hopefully convince you, at least in this subset of cases, that common vegetation cover changes can lead to changes in erosion rates and topographic activities that are in order of mounting larger changes in driving shear stresses. So the resisting strength of the landscape, the anchoring effect of vegetation in particular, can be much more important than those driving stresses, which in my sense is typically emphasized. So just some examples of motivating data from the last century from real literature. This is Mark Melton, excuse me. This is Mark Melton from 1957, looking at drainage density primarily in Arizona and New Mexico. There's some variation in precipitation, you know precipitation, in these regions. But the primary impact on drainage density, as you go from relatively forest-covered environments to relatively bare environments, there's a two-order magnitude increase in drainage density. This is an example of antibiotic change from done in 1979, working in Africa. So he showed that gotta give a meaning in a runoff, antibiotic changes in vegetation cover can lead to order of magnitude changes in certain fields. So vegetation is clearly important. And some of the things I want to touch on in this talk follow upon an NSF-sponsored workshop that was held in Tucson a few years ago. And there's a publication that came from that that fed many of the people who were here in this room, who are also present then. And I thank you for your participation. I just want to highlight a couple of things about this particular thing. So the question we posed was about forecasting the response of those surfaced to future climatic and land use changes. This is a review. In particular, we spent a lot of time on what ended up being a supplement, but a very long and comprehensive supplement that goes past this zone, but past this zone, and demonstrates what we think is our current capability to make forecasts and also what the knowledge gaps are. So if you read the paper, there's actually a supplement that has a lot more detailed information. We like the interest again. And this is just some key points from the 10,000 foot view from the large scale view that we wanted to emphasize in this paper. Number one, that the response is to vegetation cover is a key knowledge gap for all the process zones. So whether you're working on hillsides or sand dunes, it's important. And it's, in some cases, the vegetation cover effect is dominant. Number two, generating reliable forecasts for future requires of models and validated, successful fine casts that are essential, obviously. And so for a future talk, I'm actually going to spend a lot of time in the distant past in this particular talk. And I hope that's okay, but it's coming from this motivation. And finally, genealogists should collaborate more with their system models. And that's sort of an obvious thing, but it requires that we move to a larger scale than we are typically used to working at, working at as a community. So I'm going to try to touch on and provide some examples of all three of these things. I'm going to draw up my own work. I apologize in advance for that, but it's worth what I know best. So I'm not suggesting that other work is a little too good. So in terms of vegetation changes that have impacted the landscape in a major way, one example is the glacial, renegotiable transition of the Pleistocene to Holocene transition in the southwestern United States. Forests were at 800 meters above sea level, at LGM, and they're probably at 1,800 meters above sea level. So there's been a shift to one kilometer in the biomes in the southwest United States. Elevation is king. You can go from Mahali to Sonora to Chihuahua and you get slight differences in those elevations, but these are the key numbers and they impact a huge proportion of the landscape. So I took in my home state of Arizona as an example. This is the Grand Canyon. Everything below, everything that's a blue is below 800 meters above sea level, everything that's an orange is above 800 meters. So two thirds of the state or more is in this region. It's on the back and forth between forested and non-forested conditions, roughly 20,000 in the paternity. This is important as you morph it through. How do I know this? Well, we have a really nice database of elevation-specific paleo vegetation in the southwestern United States and other areas around the world. These are pack-rat buildings. They are ingesting seeds of plants within their range, which is about 100 meters from their den, and they're building up these these are deposits and ECUs there that are associated with stratigraphy. They are radio carbon datable back to 50,000 years and they're elevation-specific. So rather than a lake core or some other geologic record that's innovating over a large airshad, this is specific to this one place, and that becomes very valuable. So in terms of modeling, my model framework that I'm going to analyze a lot of examples that I want to talk about today looks something like this. I'm defining erosion as being an erosion rate as being positive with materials being removed. And so the erosion rate is equal to the divergence of a volume metric in a seven flux. This is a corubial term. This is a corubial term. So when I'm in the channels, this is channel flow, and when I'm up on the slope, this is overland the real flow. That's this component right here. So there are some typographic terms. So typically when they're doing corubial sediment transport, we have some type of non-linear diffusion model, there's a diffusivity sitting out in front, and in the transport limited case, some of the weaver that in most soil mantle landscapes, it depends on soil texture, will believe that most soil marrow landscapes are in fact transport women, which is a little bit more difficult to model computationally, but I think it's worth doing in most cases. So the divergence of the previous sediment flux looks something like this. I try to resolve the channel width in one way or another in my models. Otherwise, this is just a same power model that is specific to the unit area. And so I'm actually not going to couple a hydrologic model to an ecosystems demographic model to a geomorphic model. And I understand that that is the idea of systems that I support that kind of work. But what I'm going to show today is the examples of what I would call maybe implicit coupling. That is, rather than an explicit coupling, I'm going to make this diffusivity a function of vegetation cover, and that could be tight or a percent barrier area, it could be multiple aspects of it. And I'm going to calibrate this value to data and look at how, look at transient natural experiments to see how things vary. So, and I'm just going to emphasize it, although I'm not looking at coupled models, so I can't do all the feedbacks, I do claim that the calibration of these values locally to these particular case studies is going to be tight. It's not that hot. I'm not changing things kind of willy-nilly. I have a small number of parameters, but I know exactly what they are. So, this is a diffusivity. This is a trivial transport efficiency, if you will, for a given unit area and given for given slope, how much material is capable of moving. It's a function of grain sizes, a function of storm distributions, a function of many things, including vegetation cover on the hill slope. So, in general, what I want to propose and show examples of is that diffusivity goes up, whereas the plant cover goes up because more plants see this faster by activation. And I'm going to argue that in most cases, this is an order one. We, this is kind of the changes that I'm already talking about lead to changes of, you know, a factor of two, something like that. I think what's more important is the fact that there's an inverse relationship between the transport co-efficiency and the vegetation cover such that fewer plants, because more one after the same rainfall and even more importantly, more bare area acting as sediment sources. And this, in examples that Michelle is larger by about an order of magnitude relative to this effect. So, I'm taking the wrong approach to cultural experimental watershed where a nice natural experiment has been set up in the late Coliseum. This is, the upper half of the watershed is grassland, the lower half of the watershed below 1,430 meters above sea level is a shrubland today. We know from news, however, that this portion of the landscape was a grassland until 2,000 years ago. So, we have a case where we have a portion of the landscape that is grassland throughout the Coliseum. We have a late Coliseum transition from this to something new that has a much larger percent bare area. So, we've got a natural experiment set up where there's similar uplift in the leaf. So, the tectonic setting is similar between the active faults in cutting and creating large changes from here. But, and the precipitation is slightly lower at lower elevation, which is typically the case for these elevation gradients in the west. But it's a small difference, 10 percent lower. The sediment yield in the erosion rate is 30 times higher in the shrubland than it is in the grassland, which isn't too surprising. When we cover grass all over the landscape, it's hard to get soil off of the landscape. And you go, it can change down to a shrubland-type landscape. You're going to change drainage density. You're going to start removing soil. And in fact, we see that the AED horizon is well developed in all of this landscape is completely stripped from this landscape. So, there are soil indicators that indicate how much erosion is taking place. And we see also significant changes in drainage density and also stream tail cohabitation. So, I'm going to go into this a little bit more detail. The essence of this is the vegetation change. This is our control setting, if you will. And this is our transient experiment. Another nice thing about roller cultures is that my friends Mary and Mark and their colleagues went back in time and built a state-of-the-art homes that are actually measuring every grain of sand despite the extreme episodicity of sediment transport in this environment. They're actually trapping everything and measuring those volumetric sediment fluxes in a way that's very useful for me as a modeler. And also, grain size information. So, there's a lot of information here. And I'm just spotting the total volumetric sediment flux in this case as a function of drainage area. And so, I'm going to be calibrating a sediment transport model, wanting to do area and slope if you want. But the key thing is that because they have a range of drainage areas that have these things, and because they have them in the shrubland and in the grassland and they notice this huge change, we can look at vegetation cover and we can also look at how erosion rates there is a function of spatial scale. This is unique data in the world, at least for semi-arid environments. So, the drainage density varies a great deal. This is just a wide argument and they assess showing up particularly well here. But these are these instrumented watersheds in the Lucky Hills area. This is the Kendall area. There's less instrumentation over here because less is going on, the one thing. But if you just take that wide IDM and you extract the channel network, you can get a drainage density or if you prefer to think about it, the mean distance from the topographic divide to the first valley head. It's about 15 meters in the shrublands. That is the previous system is much more fine than dissecting the landscape compared to about 16 meters in the grasslands. That's one of the main features. So, what I've done in a recent paper here with Mark and Mary, this is Easter 2016, is I have modeled drainage density as a balance between rubial excavation of the valley head and perlubial infilling of the valley head. So, there's a competition between those two things. One can set up a mass balance equation, set up equations. And so, at valley heads, the perlubial erosion rate must exceed the collubial deposition rate, because the road crossing is positive in that place, must exceed the collubial deposition rate by and not equal to the net erosion rate of the landscape. So, there's a collubial erosion rate, there's a perlubial erosion term, there's a collubial deposition term, and one models the channel head or the valley head in detail, including the channel width. And one can predict the drainage density and drainage density variations in time following this transition from grassland to shrubland. And so, what I've got here is just a map. This is a function of distance from the divide, and I've got the magnitude of collubial erosion, collubial deposition, and I've got the total erosion rate down here. I'm not going to take you in detail to all the curves, they just want to know that one can predict with some accuracy using, I think, a model that has no free parameters at all. It's taken couple of using the available data at Walnut Gorge, one can predict this length scale of 15 meters for the shrubland and 60 to 65 meters for the grassland. So, this is a prediction of drainage density as a function of vegetation cover and time. Just another example that I want to touch on, this is if you go to high elevation of Tibetan plateau, you see actually a broadly similar landscape just on steroids. This is deeply dissected landscape. If you've ever been to the area around Lhasa or other places to the west along the Yolongsangpo Valley, all of the landscape is absolutely hammered with gullies. And so, this is what it looks like. And these gullies basically penetrate partly at the landscape, they're vertically walls. It does look like recent incisions. So, a number of years ago, I went to the fan deposits for there, the base of these gullies, did OSL dating, and we were able to bracket this decision event on between 5,000 years before president and 1,000 years before president. And this coincides with the onset of astralism in Tibet. So, and there's no known climate change or vegetation that changed at this time that would lead to this kind of event. So, we think and we propose in this paper that the increase in drainage density and the increase in stream channel concavity was a function of overgrazing due to this anthropogenic effect. And so, we produce some amount of models at the time that would show that a transient increase in K, that angular rotability coefficient, can reproduce both the long, long profile and the map view changes geomorphically. I want to come back to the southwestern west and think a little bit more about the long time scale. So, here I'm plotting up all of the backup reading data for the place to see the Holocene transition in the central Mojave Desert, which is the richest source of leading data in North America. And what I'm going to take advantage of is the fact that the forests, as they go from lower elevation and retreat to higher elevation during warmer time periods, it is a time transgressive retreat. And it's one that is fairly slow, so from about 15 or 17,000 years before president to about 9,000 years before president, the forest area was extended by 215 years when they went out much faster. Now clearly, there are multiple ways. This is juniper is abundant and this is juniper is absent, so I'm using the pinion juniper forest as my transition point between the shrubland and the forest. There are other ways to track the biome shift to higher elevations, but this is what I've chosen here is I've integrated this curve into a numerical model because what I want to do is predict, I want to test this hypothesis by predicting aggregation on fans downstream from source engines that are forced in this way. So this is a map that makes a prediction everywhere in the Mojave Desert for the timing of onset of aggregation due to place to see the Holocene transitions up so far in that location. So it's using essentially a full routing algorithm combined with the data that I just showed you to make a prediction about the onset. And there's some details about, I made some assumptions about exactly when the source region was undergoing a transition to cause the triggering. And there's some details in the paper that we have to talk more about it, but in general basically you get earlier predicted deposition when the source region upstream from you is at lower elevation and later deposition when it's higher elevation. And so this is a comparison to optically stimulated luminescence ages, that are rarely what I call the Q3A stratigraphy, that is the stratigraphy that is the primary depositional event during the Pleistocene Holocene transition. And the open circles are the mild predictions and the black ones are the LSL ages at the same site. So there's a nice general trend and there's some differences and some of the ages are just the maximum ages, not because of the stratigraphic context that they're in. But there's a pretty good trend towards later deposition onset at higher elevations. And I also want to make one more point without getting too much in the weeds in this particular example. When you look on fans, it's interesting we can actually get two terraces from one side of the pulse system. So I want to emphasize that when you look at Pleistocene Holocene transition terraces in the southwestern United States, we actually see two. There's a Q3A and a Q3B. The Q3A is the timing that I just documented. The Q3B is inset into that deposit and occurs much later. It's a near photosyn age typically deposit. And so the modeling that I've done can predict both the Q3A time and the Q3B time. So the point here is that the decrease in vegetation color, the increase in percent bare ground, triggers an increase in drainage density, the pulse of sediments. So what's happening is that the global system is expanding, all those restoring global sediment are now forced to other channels. The flux in the sediment, like fire hose, down on these fan systems, they're causing a pulse of sediment to the fans, overwhelming fan systems, causing a vulture. And when the sediment supply is increasing, that's when fan building takes place, when the sediment supply is declining, the fan heading trenches, the channel narrows like crazy. And it actually reworks a lot of sediment from the primary zone into the distal zone. So these are two terraces from a long response, okay? And so we always see the Q3B inset into the Q3A. So there's a very specific map relationship that we can use to test this model. And also we can make predictions about the timing of deposition and abandonment of the surface and this one as well, the one that's inset that represents the waning phase of the sediment pulse. So I'm now testing this in other areas around the southwestern United States. Just an unsolved problem that I'd like to throw out to some ambitious graduate student. This is a lower Colorado river, okay? It's integrating a very large watershed obviously, so they're interested in questions about what might be happening. But the statistician is pretty well known, and I just want to point out that there are three time periods in the last five million years where more than a thousand cubic kilometers of sediment has been stored along the lower Colorado River, okay? So these are huge avidation events. Why? Because, well, this is 200 meters vertically. It's 500 kilometers long. It's 30 kilometers wide. I mean, this is a huge area that's undergoing major avidation. There's no literature on this problem whatsoever. There's no interpretation. I don't think there's tech products going on. It's significant. Some people argue for the pyrogenic uplift of the Colorado Rockies potentially driving something like this. I think it's related to climate change. I have my own ideas and I've been testing these things, but I think this is a good modeling target, one that is ambitious obviously because of the scale of the problem, but it's so important. Basically, this is the Colorado River upgrading 300 meters coming back down to its modern level, re-aggrading, re-ensizing, re-aggrading again multiple times, and we're talking a huge volume of sediment here. The stratigraphy is well known. The dating is well known. There's no interpretation. So moving to shorter time scales. I want to talk to you a little bit about wildfire. And so wildfire is of course one mechanism of removing vegetation catastrophically. We're seeing big wildfires around the western United States and I've been interested through the CZO program on looking at the impressive wildfire landscapes. And one of the things we found is that when you measure all the origin rates that are wildfire affected and that are non-wildfire affected, so this is a high-severity wildfire less-conscious 2011, these are non-wildfire affected regions before the fire and because we're integrating these lines in your tidal sediments and we do other long-term erosion rate studies, you can run the numbers and you can show that wildfire is responsible for 99% of all sediment exported from these watersheds over geologic time scales. So essentially nothing is happening in the absence of wildfire in the few years following high-severity wildfire. That's where everything takes place demorphically. So this is an idea of hot moments. Things are extremely episodic. I'm not saying that this is true everywhere in the world, this is specific to the various caldera area. But in high elevation forests of watersheds where there's non-wild sediments being exported in the absence of disturbance, disturbance is critically important. A few years ago I started working at the Global School and it was actually really open. It really opened my eyes to the fact that if you work at the Global School, people care about what you're doing. Usually I write a paper and nobody reads it, nobody cites it, right? And so this was a case where I was interested in sediment fluxes and James Sylvester has worked on this for many, many years. So I produced a model that had, I think, some things that I felt were important, like slow texture and vegetation color and the river profile globally, every river in the world has a launch to a profile where sediment takes place along that profile. So this is just an example of sort of the baseline case for that anthropogenic effects. This is with anthropogenic effects included in agriculture basically. This shows a major increase in sediment yield by about a factor of 30 globally in a shift from the western United States to the eastern and midwestern United States due to that anthropogenic effect. So this is just an example but I produced this paper and started hearing from the Earth's system modeling community and got really excited about the fact that they cared about people who are asking people like me to follow just through asking questions at the global scale. So I started talking to them about Earth's system modeling and I said, no, you know, we always assume that soil, that's the bedrock, is two years everywhere in the world. And it doesn't that work? And I said, no, you know, first of all, really there's things like satellite below the soil that was used to some of them. And so basically I started a project to do something better than a two meter global average. I'm not sure that this is the best available data in the world, but it was my attempt over several years to create a map of depth to paralytic materials, so not necessarily fresh bedrock, but I make several layers in this database which is now available to James. And so this is the average soil uplands and you've got a few meters above bedrock. And then in lowlands, of course, you have large seminary pauses like the Ogallal Foundation, the lower Mississippi. So that was the integration of geologic data and models. Models were began in the Suzio program at the watershed scale. And within a few years, and within the time scale of this project, this data, these data are now used in the CLM. So this is an example of a Suzio product began, you know, in the mountains of Arizona. And over the course of a few years, the land global and CLM, the best land model in the world, was modified to accept this data. And this data is in CLM. So this is not an idea of how, you know, this is an example of how Suzio research, as one particular example, has a direct impact on the quality of the best soil system model in the world. I think I can say that because we're here in Boulder. Okay, so it's an example. And, but I think it was very exciting for me, so there's actually two papers. One is on the dataset. There's also another paper in the general climate that shows the impact on the results of CLM by doing variable depth of bedrock. So I just want to touch very briefly on the issue of the colonial transport terms. I've been focusing on the strength problem, and I've been focusing on the sediment yield in the fluvial component. So I just want to say very briefly that Leakwood Liar, from a student of mine, has been looking at cinder cones across an elevation gradient from Arizona to Oregon. And the cinder cones are nice because they're radiometrically dated, and they typically start off steep with a nice initial condition, they're like big anhyls, right? And so they're at the level of the pose, and over time they develop soils, so there's a soil component to the study, and they also evolve into a product primarily by collubial processes, okay, by dry gravel and by various landsliding processes. So what he was able to show, without going into all the details, is that looking at the development of asymmetry over time and relating that to the potential controlling factors, he was able to show that it's the vegetation density on the north-east of South Basin slope, not today, but actually at an interglacial climate that controls the aspect development of these columns. So if you're in a, if you're in a Pleistocene color like this, since we're 100,000 years old, most of its life was not spent in a Holiston climate, it was spent in a near-glacial climate because of the asymmetry of the global temperature curve. Most of the Pleistocene is not like today, it's like that out of time period, so when you look at vegetation cover as a function of elevation in Arizona, this is the current setup where there's more plants on the north-east of South, well at LGM that was actually reversed, and that's critically important for explaining the asymmetry. So we argue that the data, and it's the host of data, who's named slope aspect and also drainage density, can only be explained by increase in this diffusion coefficient with biomass. So some conclusions for you. Common vegetation cue changes can trigger order of magnitude increases in erosion rates and topographic metrics, and hill slope removal systems, and I should say that I have a version of this talk that has wind erosion, because I do wind erosion work as well, but I just focused on hill slope removal systems due to time constraints, and increase the trivial alertability curve, which usually produces the first order behavior, i.e., a transient increase in drainage density and channel convectity, with the resulting pulse, abrogation, and fans, which we can study and date. More vegetation cover can increase the Corvio 7 transport, or this seems to be a small magnitude in cases that I've studied anyway. In global scale, earth system models and dynamic vegetation models can be used to determine future hot spots of geomorphic change. Such models may require better component models and input data on geomorphology, and we, as geomorphologists, can continue to do that. Thank you very much. I'm just trying to follow up. So now you have the work on this and this, and how does that affect any accuracy? You could tell us this sort of and you deserve some of the vegetation we have. That's a good question, but also, or maybe more on that type of region, or more affecting the critical surface, are you thinking about it? Yeah, it's a good point. So I was doing the simplest part of the possible modulation of those in critical coefficients. I think in the cinder cone problem, it's sort of interesting in the only case study that I talked about that's really sensitive to the critical angle, is the cinder cone problem, the one that I talked about at the end, because they start off exactly at the angle we pose. It's a volcanic plastic deposit, it's a type of fallout, and so it is a sand pile, truly. And it does depend sensitively on the SFC value, but it takes quite a while, about 100,000 years for a forest to develop, because at the beginning, these cones are just absolutely porous. It's full abity drainage all the way through. So you need to have enough dust input to the surface to actually uniquely run off and actually support plants. And so that's just another component of the problem. I'm not just missing any concern. I'm just saying that at least over the first 100,000 years or so, there's not a lot of vegetation cover to speak of, because there's not a lot of snow development. So while they're sitting right at that sensitive slope, and they're sitting right here at SFC, and there's a lot of nonlinear colloidal sediment transport, I'm not sure that vegetation is going to control SFC, because there isn't a lot of vegetation present in the early time period it becomes. But your bladder point is going to take it. Yes. So you're interested in that? Yes. So you said forest won't take a long time to grow, but it depends on what the climate was exactly at the point, because the plants adapt, and probably we're not looking at this right, right? Because if you look at Oma yet in 2015, he actually solved this problem using child, like a landscape violation model, based on other things too, was in private where it starts about vegetation, and it's back on the soil erosion, and not facing slopes, how they differ on not slopes. You have episodic erosion, and you can look at it as a function of me, but I don't think bioturbation is the only reason for that. Maybe it's because you grow up the amount of, no, you can save more moisture on the north side, not facing slopes, so you grow forest, but at one point you see that it cannot sustain the weight, so it's just episodically question on that. Probably everyone knows this, but I'm just saying probably that's the reason why you see a bee as a function of me is fine. If you look at a long-term view, but on a shotgun view, it might show you the reverse, because there's continuous soil erosion on the south facing slope, because of radiation, there's no plants, there are no trees, let's see, so you have a career erosion. Do you have any insight? Well, yeah, you're going to come to my slope after the first shot, right? Yeah, I think you are coming, right? So yeah, no, I mean, you're so, yeah, so I mean, I think that I'm not sure I answered your question, but I would say that, yeah, I think we should talk after, you know, maybe about the specific details and the special time scale you're raising. I'm trying to, I don't know if this is one answer for all field studies, and so, you know, I think that the models that are produced are motivated by these specific field studies and patterns of data. I mean, I don't start with a model, I always start with data, what is the pattern, and then iteratively try to see, marry the two, in a way that is still, you know, a good test of the model. I'm not trying to force the modeling to the data, but it is always alluded. So if I consider a broader range of sites, and I come through a different group and crew. So as a follow-up to that, I'm curious, what you mentioned at the beginning, the value of this one data set at this section, and you're going to vegetate it, or the grassland or some sort of, what additional data could be collected that would be most impactful to help constrain the models? I think that's a good question. I mean, I think that in my answer is that I don't have a lot of good data on how the diffusivity, the diffusion coefficient, varies at one level specifically. I have tried to look at, you know, to do some study state and look at rich top curvature, which is a standard approach that comes from, you know, the UK group and whatnot. And I had a hard time finding the systematic results from that sort of thing. It could be that it's just not a study state. It could be that the landscape is also undergoing a tilt-style uplift, which is different from what we usually assume a landscape evolution model is a block-style uplift. So I think that there's a lot to be done in serious landscape evolution modeling of all my goals in particular. And I'm not going to do it all. I'm just going to be the first people to get there. And I tackle this particular drainage density problem using this vegetation change as a hook, but there's a whole lot more to do. It's a great place. And if you have data on channel width, if you're interested in channel width, I mean, they've measured to anything you want. They've probably measured it. And so it's a huge, it's a great resource that many people know. It's supposedly being written as you speak in us. Thank you.