 All right. Good morning, everybody. Two logistics announcements. Most importantly, after this morning session at 10 a.m., we're going to meet outside for group photo. And Louise Kellogg has asked me to announce that the people who are going to share her at the concluding panel discussion should meet up with her, I guess, after the morning break, shortly after the photo. And so welcome, everybody. It's my pleasure to chair the first session of what I think is an exciting workshop. We heard a lot about the challenges, and in particular, you know, coupling involves around trying to understand the constraints on the surface process laws and trying to figure out how to implement those surface processes efficiently in parallel algorithms for supercomputer applications. Our first two speakers this morning have worked for a long time on those two subjects. We're going to kick it off with Brian Unidis on reconciling landscape models with reality, a spectrum of success. So I would talk this morning about what happens when we take sort of these predictions of the elements of them and sort of apply the predictions and some of the implications of them, how they sort of scale to sort of these natural systems where we can actually go with the data and better understand how well we're actually, or what we can actually learn with that sound. So what I'm talking about this morning is sort of what locations, spatial scales, time scales are the models that we implement actually reveal information on the experts and how sort of, you know, if they don't work, you know, or they don't work as well as we hope, right? So I'm going to start with, there are complexities in natural systems that limit the applicability of our simplified models, or potentially require some sort of, you know, innovative ways to think about how we can just better capture physics with how these lands are connected. And finally, I also want to point out that, you know, a big sort of point is that there's not going to be a single equation approach that can survive all rivers, focus on better up rivers, unless all landscapes, all environments, on a climatic around the world, right? So we have to sort of have this thought or this sort of framework to allow us to be flexible to think about what I want to start. So we've heard a lot about this stream power model. I recognize not everybody in the room might know where rooted in works. I'm going to start with some quantitative background and field sort of that model that's just used so ubiquitously. But the root of this is really what I like to call a proxy model, mostly because it might contest the trip over. You didn't say it right now. I knew I would screw that up. That is perhaps rooted in some physical that the rate of erosion of a river system is equal to the sheer stress of the river possibly raised some power times some erodibility, erosion. Right. And this is sometimes we add a threshold, right? Non-bron sort of pointed out that's not always happening. And it's kind of covering my figure. But there are some physical reasons to start with sort of this posture, right? You can imagine that the sheer, I'll keep going. I can just talk. The sheer stress of a river, we might think, is a strong control on the river's ability to take a jointed block of rock and sort of pivot it or slide it out of place. That's a fundamental process. Also, sheer stress is known to strongly control sediment transport. The more sheer stress, the more sediment you can use. You can move the bigger sediment you can move. And that's through the river system to abrade the sort of sand blasts of the bedrock or bounce along the bed and increase damage into the rock. There's some physical sort of reasons to start with the sheer stress. We don't know very well. Double sheer stress, does that really mean you double erosion rate? That's dependent on the process as well as what this sort of K is. So what goes into calculating sheer stress in a river system? I like to tell my students that are just learning that. You can think of this as a slow parallel component in the weight of the water in the river channel. The density and gravity, I think we know those pretty well. End term is a frictional coefficient. You know how rough the channel is. It can be important, but for the sake of simplicity, we're going to ignore today. A few here is our water discharge. How much water is flowing through your channel, direct connection to climate. Our W is our channel width. If you sort of narrow your channel, you're already knows those horses over smaller areas. You have a higher sheer stress. And then finally, slope, which we've seen a lot of so far because slope is the gradient, the river, and then you integrate it up and that's what you do here. So to get to the generalized stream power model, we make some sort of assumptions. We say, well, look at river systems. They tend to get wider as you go down a stream. So we're going to scale width, width, drainage area. Water discharge tends to get bigger as you go down a stream. So we're just going to put these parameters in there. Likely, you know, function of precipitation, hydrology, stuff like that. And we're going to lump all these parameters with these parameters and get this equation that we've seen a lot of. So we start with some physical theory and we end up with this nice powerful simplified model. What I want to talk about is, are we doing this sort of water discharge? Let's channel width, rock erosion efficiency, how we translate sheer stress to erosion. We're doing that right in our models. I do want to point out, I'm not trying to throw a wrench in the machine. I look at more as taking that wrench, sort of cleaning some of these knobs, things that might be a little loose in our system. Anybody that's worked with machinery knows, if you tighten that knob that was loose, the whole system can now work with potential for that to happen. So my outline, I'm going to start with talking about the rock type wrench and erosion efficiency. Talk about climate as well as channel width. I'm going to show a mix of some of my work and my students' works as well as other folks, including some people in the room here, that sort of provide field examples of their tests of these. So first, lethology. If you really want to understand the erosion surface, if you ever look at a geologic map, I think it's safe to say we need to understand how different rock types. If I went and handed this bucket to any few more colleges in the room and told me what's the different cave alleys, I probably get a lot of trying to pass it off. It's a big question, but it's a big issue. I want to sort of motivate it with small changes in mythology. You can have really big propagating effects. And this is a schematic from Adam Forte's paper years ago. I'm going to show some model results of this, but it sort of outlines a big part of the problem here. So we're looking at time, different river profiles through time. We're starting off in a soft rock, and then we erode through the contact and we expose a hard rock. The river system is going to, the hard rock has to pass the river. The erosion rate is going to go lower because it's harder rock and the river is going to start to sink. Well, that lowering of erosion rate, the river upstream of the soft rock sees that lowering of erosion rate, and then all of a sudden this starts propagating through the landscape. So at any given time while this is happening, you get this sort of spatial pattern of steady state rivers eroding the rate of rock uplift and a decrease that propagates upstream of the contact and then goes back to, you could think of the other case where you have hard rocks over soft rocks, so there's a soft rocks. You get really rapid erosion that sort of propagates and reaches upstream of that contact. So we can see this in landscape models. I'm going to show you this movie from Adam's paper. Up here is our erodibility map, our K, different Ks in the stream power equation. You can think of this as the evolving geologic map. This is our erosion rate. Rock up, the rate is one millimeter per year, the red. So as we tip down to the blues, we go down an order of magnitude, and then we'll see the elevation. So we can see we go from one millimeter per year, drop down to a tenth of a millimeter per year, and then come back to a millimeter per year, all in a few millions. These are big landscape effects for only a five-fold difference. Remember from Arthur's talk yesterday, his sort of calibrated K factors were varied by two orders of magnitude, right? We're getting a tenfold change for a five-fold, five-fold difference in K. Our gain, as Jean-Bron sort of pointed out, is greater than one, right? Lethology is important. It matters. Change your lethology of big consequences. Similarly, if we have soft over hard over soft, so we're going to start with hard rock, and we're going to expose soft rock. You see, you have a little bit more of a localized difference in your erosion rate, but again, pretty significant as you change. Five-fold difference in that K goes from six kilometer high origin. We have this challenge in sort of implementing these models and sort of these natural systems, and very hard to find a place on earth where that has only one rock type. Want to really understand the dynamics of these landscapes is that rotability is going to have really big populating impacts. We have to be able to go to the field and be able to figure out what these values are. This is just an example of a river profiling high space in central Arizona where we can see that effect that Adam showed in his modeling of a hard granite layer hooking up between a softer pylite system, and we can see that the pylite has very different steepness because it's this propagating impact. These differences in erosion efficiency have a pretty big to attempt to shed some light and show you how we can potentially implement some of these models in real scenarios to shed light on this rock type problem. We're going to go to central Idaho where the Salmon River Gorge is cut through the central part of the steep. What's sort of special about this natural experiment is that it cuts through a number of different rock types. We have the Idaho Bathlet, we have the Columbia River Resolve, we have different Nices. As this system sort of propagates through, sets off nick points and transients and incisions into different river tributaries, and they're pretty much maintained within one rock type. It can isolate the effects of different rock types on the same sort of thing. We can look at these profiles here as a topography versus distance and an ortho nice, and we have this relic topography and it dips over into our mix zone and then it's just sort of new adjusted. Here it is for a river underlain by basalt. You can get these much more discrete nick points, nice as well as the ranch. You can help constrain this sort of modeling effort they're going to take by going out in the field and actually measuring erosion rates with causing different replies. We find the relic landscape is eroding about three times lower than this new really adjusted or not the transient part, but the part that has seen this propagated. It's about a three fold difference between the landscape in terms of erosion. What we can do is take a simple stream power model and we can run it through parameter space, searching for sort of best fits. Here's an example of a pretty good fit of a stream profile as it feels that three fold increase in this case in rock got plastic also be a three fold increase. We're able to do a really good job reconstructing the modern river line with our model by just by taking that relic landscape part, projecting it back to the surface as our starting point and then searching through parameter space to find those erosion rates matching up and photography. So I put this plot down there same same river system and the absolute best fit model calibrated model we get a value of k, a value of m and a value of n. Sort of interesting is something that we have to think about in this room is it's not just k's ends vary by rock type as well. So here is a river profile it's in high space sorry for the switch pathology here but elevation versus high space and here's our sort of stretched out nigg zones in the granite system and here's our basalt system you get an n of 0.67 versus it might not be seem significant but as Jean-Bron pointed out ends control sort of can have a big influence on sort of that landscape response style and kind of so blanketing a landscape with a single n may not be and then trying to fit different k's may not be appropriate if you have and we find this is consistent for essentially all nicest and all granites always less than one over 30 profiles in basalts always greater than one usually around 1.5 which is both of those values of sort of 0.6 and 1.5 are significant go back to that original physical sort of justification for using shear stress. Helen did a nice sort of little scaling argument that suggested that if you're dominated by jointing you might have an n whereas if you're dominated by sediment transport systems maybe these these kind of can bounce around depending on the assumptions you make for scaling but they're consistent with having different sort of n-member ends based on your process of actual erosion. We see this process in the field when we go up the nice channels are nice to go up because they give us a nice a lot of information because the jointing is controlled by the foliation you actually see the shape of the pluck blocks and we can find pluck blocks with the foliation right orientation and match up we see it all over the place we see plucky evidence plucky but if we go to the basalt and find the parts where the river channels hung up and it's limiting the incision we get these very discontinuous joint jointing not a lot of plucking and very smooth sort of abrasives tend to suggest that we might be more dominated by the sediment transport right so this rock type controls this scaling between erosion even slow all right so rock type variability is the first order influence on the rate and style landscape evolution we're starting to unravel how to parameterize this lots of work to be done so I want to point out that variation rock type is not just differences okay the exponents matter because rock type dictates the dominant erosion process that we're trying to capture next I want to move to one of to the climate part so if you go to google scholar you'll see something like this on that a coursing of erosion of landscape and tectonics versus dominance of tectonics over climate in Malaya coupling of erosion of precipitation couples very spatial variations in precipitation minimal climatic control on erosion rates climatic control on better rock river incision rates or we can just you know throw the wrench in the machine and just fall on chemical weathering as the reason why climate people this is sort of two takeaways of this you want to get a paper in a short format high impact journal either find or not find climatic control second of all this is sort of a dog's breakfast when we try to reconcile this cue that we scale up to drainage area our stream power right so dog got confused and made a mess of things so worked up about um safe to say we need some serious adjustment how john hit on this beautifully yesterday this talk fact that it's weather that arose right stream in arizona water weather event so i'm going to show an example of where this might play out next and i'm also going to talk about another aspect of climate that was also brought up yesterday and the discussions climate changes for many reasons one that i'm going to talk about today coupling so i'm going to show a couple slides here from rovin db osse's work in the sand gave me a tell-in where this sort of been in the santa andreas what's up this nice national experiment in mcabriol's and they went and measured erosion rates so it caused a few slides and found a pattern of erosion rates so great test now we can look at the rivers and you know calibrate this you know tectonic geomorphic uh stream power model really stuck on your system and this is what they found we have erosion rate the approximate this and it does what we expect you get more and more higher and higher rates of uplift but then that relationship starts to sort of fall out fall over right it starts to bend over and their argument and their way of explaining this was done hit it on or talked about yesterday and it's that you have to actually consider the weather events explain the sensitivities right if we don't we're gonna we're gonna predict them but what i want to focus on or talk about is something we haven't really just we change somewhat motivated by chris polson's work he took sort of the modern central andes and drunk them down put speeds and down to different elevations and global climate models regional climate models over them found that when the andes got to be 70 and 90 percent there was a threshold change in the atmosphere so it started kicking off a lot more convection increase change the distribution and so to attack this uh when student bridge at lunch along with chris polson he's working to couple weather research forecasting model there's a component called work hydro that run it over the landscape into landlabs and we can sort of allow this feedback to occur based off um so quickly um here's an example from the work hydro page this is the front range you can see the 2013 storm you can see that the map is going to start being increased in rainfall and then we can see the river channels obligated we're linking this sort of infrastructure into very ac so what we've been doing is utilizing the andes not necessarily to recreate the andes but because then we can use all the model boundary conditions um and it's sort of created this sort of synthetic topography run along the latitude of the andes we have sort of five different landscapes that we bring worth down to a five kilometer scale model but the weather obligated system that we can then landlap and this is a very much a number of different projects that are that are going on with this i just wanted to share i think one of the most interesting results so far that is just simply taking two different n-member topographies you make them in landlab and so we don't have to worry about different drainage differences looking at a hundred meters total relief versus a four kilometer high mountain look at the cdf for the transition function of water dishes right so i'm kind of putting this line here for reference we'll look at these slides um and we can see how in an equatorial climate it should build 100 meter to a kilometer high origin how it changes the conditions of the starge year at six hour weeks um and we can see that there's a pretty um that there's a system gets wetter but it gets wetter because the base flows are essentially pushed to a higher higher the big floods you know your 10-year flood if you will your five-year flood don't really change those sort of geomorphically significant events aren't really impacted by very different conclusions if you look at the subtropics here's our 100 meter topography here's a 10 reference line we build that four kilometer the entire distribution shifts we go from this 10 meters per second especially never being simulated in the river system 70 meters which is a long time of in terms of geomorphology right usually think of these events that only last for today go to the mid latitude it's even different still and our shape of our distribution the base flows go up but also the way that the tails sort of fall off the distribution is changed as as we change the topography so simply stating latitude matters in this sort of so cleaning up the dog's breakfast is coming increasingly clear that a mean climate state is really going to be difficult to translate especially weather roads not climate and resulting weather patterns change over a range of time scales specifically climate topography coupling varies with latitude because of this we might very well expect a dog's breakfast we start going to try to find these sort of climatic controls on landscape very basically very different happens finally I want to get to something I've been working on and thinking about for a long time and that's the effect of channel geometry I think when I say channel geometry think of the width channel right we know that that you narrow the channel you can increase so far we've the entire workshop here is focused on models that only allow slope be the topographic parameter just but we know in sheer stress with and we see it change the landscape so this is limitation challenge of how do we maintain this power of simplicity while recognizing the river and it matters to highlight some field examples lava and oblox classic paper from 2001 where they use river terraces to construct rock uplift rate across the main frontal thrust in the layers and they find there's about a 10-fold increase in rock uplift as the river passes the segment but look what happens doesn't change 10-fold change in the order magnitude to rock uplift changes a little bit in the bouquets but not the 10-fold change you might expect from this slope having to accommodate right what's happening is the channel we look at distance downstream the channel as it approaches this high uplift zone is narrowing accommodating this extra rock uplift by narrowing right if you're going to try to use a stream power model only allow slope model this system right you're going to have a bad time not going to work at all right slope is not changing at all only have a 10-fold change sort of think how we handle river systems I mean the Himalayas you may have heard of it one of those origins that we find a very similar strange sort of effect in southern Taiwan if we look at these river systems so Taiwan is propagating from the north to the south so we have an age and landscapes and so we can go from a young just emerging landscape and then the more steady state adjusted landscape as we move to the north we could go and measure channel morphology and we see that slope does what we expect as you go from the southern tip to the north we see this five to six full steepening of the river system what you'd expect from the stream power model but we also see the channel widening five to six full channels getting wider as the river is steepening at about the same pace here stress isn't really changing right it's breaking a fundamental assumption of what goes into that stream power model when we make all those scaling arguments right something fundamentally different is going on in Taiwan this k term is more complex than just a uniform value in some landscape and what's going on wire rivers wider where they're steeper to be influenced sediments applying river systems build this relief start to kick off landslides right in a way as we add that landslide material now the rivers have to evacuate them eroding bedrock so in a way the hill slopes are dominating that sort of slope increases and with widening the river system I don't want to just say things don't work and then move on I want to come up with a few ways that we can approach problems so a typical 1d stream power model works like this where width just scales as you move downstream we need a system that allows dynamic variation with maybe of a propagating nick point maybe but use this principle of channel optimization go through this quickly but it is you can talk to me later about it about the details but the width we essentially only have to add one parameter not making this super complex trying to maintain this implicitly adding one parameter and we solve this set of equations three times one for the current channel width slightly wider one for slightly narrower that gives us the direction whatever one gives us the highest erosion right we assume is going to entrench the channel the most and that the system's going to sort of be attracted to that sort of state and move so we can do this and we can look at a dynamic river model with an evolving width essentially a 1d river profile model but that is a county in this system we're a sort of sediment dominated like Taiwan so as we increase rock uplift the sediment supply we saw we changed parameters space such that lots of bedrock exposure the limiting factor is actually catching the rock system the same exact equations are being solved as we increase the rate of rock uplift here our channel is narrow this has implications for sensitivity between increase in rock uplift the fractional increase in rock uplift black line is our 1d stream power model blue line is that same stream power model but allowing for a dynamic width it's less sensitive less change in photography for change rock uplift and then these other colored dots show even less sensitivity or sediment so we can see why this happened or the sort of different regimes erosional regimes that these river systems end up in we have a fractional change in the wideness of the channel how wide it is for given drainage area versus this change in rock uplift our detachment limited systems get narrower as we expect in a place like Idaho where we're limited by bedrock detachment whereas our sediment dominated systems get wider as we increase sediment supply perhaps like a place in Taiwan sediment transport dominates us so the spectrum of success and be able to reproduce these different river systems I think is because there's a spectrum of how river systems work right and we have to acknowledge that we have to be able to implement that into our our landscape right um so in some environments channel width can be a dominant control of erosion process it's just as dominant as hope maybe 1d generalized stream power models don't really catch our I think there's some simple modifications so back to my original points there are complexities and natural systems that can limit the applicability of our simplified systems or require some adjustments there's not really a single equation I think that's going to solve everything to be aware of that it's going to require flexibility a bit of situations like this like my CEO flexibility we have to have situational awareness where are we at you know where are we at in the world what latitude are we at what's the sediment supply relative to the rock strength and things like that in order to think about what sort of physics we want to make sure we capture right and then finally come up with these innovative solutions right um to to to to figure this out and with that I will wrap up and take questions thanks a lot Brian we got time for questions we got a question yeah so what's wrong with chemical weathering seems like at least for long-term uh climate fluctuations if you have summerineous dependency you would expect that to be a contribution at least sure um in general when we prepare physical and chemical weathering order of magnitude maybe 20 percent tend to throw it up um but it's part of the mass balance something that we should account for I've tried to talk with geochemists and say okay can you give me an acquaintance if I can give you a slope or uh physical erosion rate expectation to what we do we get kind of a bucket of rocks well there's lots of matter um I could defer to John John or Nicole on this too well an awesome talk um I just have one basic question with the Indian example for models are there diagnostic signatures in the landscape associated with those different PDFs of the CDF signal discharge pick out in the actual landscape test that tough because that I mean we don't have fully we're not going to have a couple we don't have we've just been doing these sort of n numbers and now analyzing discharge variations erosion variations uh thresholds yeah in those situations you see differences but in thinking about when they are fully coupled which their program takes the work the land lab for about an hour and then we go back to the work right and so the the the dynamic the full dynamic there we anticipate where where it's going to go I think that's the best I can do right now to anticipate that I think the diagnostic would be you might have different oral variations different lattice topographically I just wanted to follow up on that on that on that exact question because one topographic thing you might might see a different question I want to ask you with a guess maybe suggestion ahead it's a really cool prediction to get the CDF of flooding right there's a lot of hydrology you gotta yeah we think especially in soil water balance so it's worth going in one thing you look at is data on the screen do they follow those predictions haven't done that much easier way to find out that's true but it doesn't tell you about how it may have been all starting with that phase as simple as we can where that's there's all that is in the work hydro soil evaporation all that stuff we just make it the same at all attitudes we don't want to have to interpret results on top of that just first start with that topography process then we can start adding adding other layers of complexity right and more work that soil water evaporation that's well this yeah I mean fair enough I just can add a little bit to the discussion about the Andes I've looked at it between around 30 degrees and in the wet and dry side of the Andes and looked at the distributions as well I can say that for example it doesn't really matter what the distribution is looking like and if your average erosion happens at the average discharge that is not affected by the threshold so that's the case in the Andes that you use for well that doesn't tell us about the temperature like you said but that is one example where you can compare the two effects of mountain uplift plus the climate it seems like an important component of these investigations the values of the parameters even creation etc there's anything to say about uncertainty in these but we often report them as a number without uncertainty constraints I mean do you have a I'm talking about like the Idaho example so our end values 30 different nice might be one outside so yeah we could the salt has a little bit more variability on our calibrated models and we're allowing for uncertainties our erosion rate