 Good morning. Thanks. I want to say first of all that I'm really honored to be giving a talk here at CSDMS and I'm going to be talking about some land lab modeling eventually. Full disclosure, this model was running first at 4.45 on Friday. So there's some sort of preliminary results here but we are conserving water which is a good thing. I think maybe perhaps especially for this context of bridging boundaries, the more useful, hopefully useful thing that I'm going to talk about is really about the the gains that this group I'm working with had to make in terms of our conceptual models in order to do this work. And so as a person with a very traditional geology and geomorphology background, I was carrying around some conceptual models that were just just too simple, not quite sufficient for doing this. And so I'm going to spend some time on these aha moments that we had as a group which if you're coming from hydrology might not really be aha moments to you. They might be pretty pretty obvious to you. Okay so I'm going to tell a story about this group of collaborators and I who've been working on how landscapes evolve following continental glaciation in the central lowlands, which is just a fancy term for the Midwest. Okay so acknowledging my co-authors, Jing Tao Lai, who's here he's going to actually talk about some of this work tomorrow in Cecilia Cullen who's the student behind the model that I'm going to show you at the end of this talk. I've also got collaborators Peter Moore Bradley Miller and Josh McDonnell at Iowa State and then Karen Gran and Brian Sockness at University of Minnesota Duluth. Okay so our story of post-glacial landscape escapes starts before the glaciation of this region and I just want to remind you that before glaciation there were actually well-developed drainage networks in this area and they had things like big and sized bedrock valleys. Okay so we've got, this is an example, this is the Taiz Muhammad system so kind of a Ohio River scale sort of thing with bedrock valleys in size 50 meters or so and if you know the the river networks of the Midwest this doesn't really correspond with the modern river valley. Okay and that's because these valleys got filled up with sediment during the set of glaciation. Now what's hard to see on this slide maybe is this is the valley I'm talking about. Okay so this bedrock valley the Taiz Muhammad Valley is like 10 kilometers wide here in Champaign County and it's filled up with a stack of pre Illinois episodes so greater than half a million year old glacial outwash and then till from the penultimate glaciation and the most recent glaciation we see inside that big valley a stack of penultimate glaciation outwash Wisconsin episode outwash and then here is the modern Sangamon River River Valley in set into that set that stack. Okay now it's a it's a different question for a different time why we have sort of superimposed all these valleys at different scales but what I want to point out is we had a we had a big big drainage network it's all filled up and now we have no topographic expression of that. Okay now this slide is really courtesy of Andy Wicker who helped me see this clearly for the first time but essentially the drainage network for the whole Mississippi was really really different when the ice was here. Okay so we can look at this reconstruction of the ice sheet here's the centers of accumulation and drainage glacial divide there. Okay so all of this stuff is going this way. Okay here's the modern Mississippi basin on there so we got hundreds of kilometers more to the north of us all that water is coming out through these river networks. Okay so this this makes it sort of not surprising that we're going to be doing lots of sediment transport down these down this out this southern margin and we're actually going to be having erosive conditions sometimes too. So this is a really really famous example of this enormous proglacial lake glacial lake agacy during the retreat of the Laurentide ice sheet which several times catastrophically drains down the Minnesota River and carves this big big valley all the way down to Pedrock. Okay but this pattern of incising river valleys that toward the end of the glaciation happens in a lot of places. Okay so the Ohio River is like this too the Illinois River the Wisconsin River the Wabash River the Sangamon River okay there are all these really really big valleys in size toward the end of the most recent glaciation and now there are the major rivers of the Midwest living down in the bottom of those valleys with these flood plains meandering around between these valley walls. Okay so all of this leaves us with the landscape that we have in this region which has some positive release features like enmarines wildly vertically exaggerated here the University of Illinois sits on an enmarine big enmarine it's about six kilometers wide and 20 meters tall. Okay that's the big positive relief feature and then we have these negative relief features which are these in-sized valleys. Okay they can be incised the Sangamon's incised about 15 meters okay the Wabash maybe 70 so sort of tens of meters incised big negative relief features and then we have this which we refer to as the uplands I'm not sure that that means the same thing in other parts of the world we talk about that as the upland. Okay so this flat surface and a key feature of this surface is that it's got a lot of closed depressions in it. Okay so while the water associated with this melting of the Laurentide ice sheet carves these big huge valleys it doesn't really effectively drain most of the landscape. Okay we're left with a whole bunch of area that's that's stranded it's not part of these external drainage networks. Okay so we have lots and lots of closed depressions lakes and potholes and so on but just in central Illinois it's just wet prairie. Okay they're not even particularly clear topographic expressions it's just the water table is high and there's nowhere there's nowhere for water to go. Okay so what we know about these landscapes is that this condition of having all this non contributing area doesn't last forever. Okay we can take advantage of the fact that we have surfaces that were glaciated at really different times most recently across the Midwest and do the classic geology space for time swap and think about how must development proceed. Okay so again this black area is all glaciated more than half a million years ago most recently the dark gray here in southern Illinois about a hundred and thirty thousand years ago mostly most recently glaciated and then all the light gray areas during the most recent glaciation although there's considerable variability in the timing between the different lobes and then you have the retreat of those lobes giving us surfaces of different ages. So going way back to Rui in 1956 he's making an observation that as you go off of the Des Moines lobe off of this recently glaciated surface onto the much older surface there's a pretty market increase in drainage density. Okay so channels are growing in to drain those uplands. So one of the big efforts of this collaboration has been to try to look at this from a different angle. It's really difficult to reconstruct drainage density in the Midwest because there's a lot of drainage ditches. Okay so we see like a 300% increase in channel length in our county that's that's not natural. So instead of looking at drainage we've been looking using soils data to try to identify areas that were closed depressions or were at least very poorly drained prior to agriculture. Okay so we can look at those and we can use more detailed chronologies of the ages of these surfaces and make plots like this which are showing us the percentage of non-contributing area so this is how much of the land surface is this poorly drained disconnected stuff versus the age of those surfaces and you could see decreases over time. Okay so we're losing non-contributing area over time. Several students are working on expanding this to much bigger regions. Okay and trying to constrain here so here we're looking at in general as we get to darker red we're getting to older surfaces and in general as we get to lighter orange we're getting to less and less non-contributing area so it's a pretty good relationship. We're working on now splitting this out according to landform type so can we find glacial lake plains on all these different surfaces or can we find look at marines on all these different surfaces and so on. So this is a bit preliminary but I think we're gonna we're gonna see that these patterns hold. Okay so we're gaining drainage density we're losing non-contributing area but how? This is what these landscapes look like and there's not a whole lot to work with here. Okay we don't we don't have a lot of slope and we it's really hard to see how you organize the water to get it together. This picture for example this is actually the high relief part of the upland this is the these are the heady heights of the champagne marine here. Okay so that's that's the big action place that we have. Alright so yeah so the slopes we have are actually on the edges of these in sized valleys we have these valley walls we have those slopes so we think let's treat this like essentially like the same problem as passive margin great escarpment retreat right we've got a you've got to grow channels into this flat plateau away from this escarpment slope along these valley walls and to do that we got to give them water. So so we went back to Tom Dunn how do you get water into channels right how do you get runoff generated so this is his classic figure telling us about how not all the water in the channel came from flowing over the surface right a lot of it probably came from this shallow subsurface flow or maybe from groundwater flow okay so let's think about how we get water in the channel and that's let's think about what you know what would Tom Dunn has seen if he was working in a landscape like this where it's really flat it becomes really unclear that those surface water divides which you can still define right you can still define surface water divides but it becomes really unclear that those are very important in driving the water flow okay so essentially this question boils down to our drainage divides our watersheds really good indicators of how much discharge we should expect okay so here's like the first paradigm that I had a lot a lot of trouble with breaking okay because what is this I mean my kids are being taught this in elementary school right now okay you live in a watershed and that's it's all going down there right and that's true in Boulder but not so clearly true all the time in Illinois so you start feeling like you're on thin ice when you say maybe maybe watersheds aren't real okay so in particular like we have these closed depressions and they all have watershed boundaries they have contributing areas they're not very well drawn right here oh as an aside this is actually a geopark in Denmark that is dedicated to the glories of the postglacial postcontinental glaciation landscape and it is amazing and if anybody wants to help me do this I would love to have a Midwest geopark I think that would be excellent but yeah they have these same sort of closed depressions so what happens to that water does it do any geomorphic work okay is this if this is of any importance to us well one thing we know it does is we know that water within those depressions contributes to slope wash that sort of fills them slowly over time we see this most clearly when we look at what's happening after agriculture okay so here's an example this is near Purdue there's a basin here you can see this is kind of hundreds of meters in scale there's a basin of a closed depression and what they've done is map out this is what the soil surface is like now and there's a paleosol they can define that's the pre agricultural surface okay so the swales getting filled up with stuff if we if we think about the time before agriculture we still see like thickened a horizons and so on in the hole in the bottoms of these depression so probably even without agriculture we're filling in these depressions slowly over time essentially making it flatter but that's it's helping us with our problem of getting water out of the landscape but let's think about other ways that we could get water out of the landscape not just by filling up these depressions with sediment but maybe we could just fill them up with water okay we know that happens that sometimes the closed depressions get filled up to their spill points and they spill over and the water goes other places okay here's a an example of a really these are really big closed depressions really deep ones from the prairie potholes region and here's their conceptual model of how as the water level rises in the different depressions they spill over into one another and and maybe eventually into some external networks and so here's an example of mapping all the flow paths between all these different closed depressions all right so Jing Tao will talk tomorrow about what happens if you make a numerical model of how that changes landscape evolution what if that water from those closed depressions actually did contribute to headword growth of tributaries and I don't want to wreck everything for him I want to not steal his thunder too much but I think it's safe to say that that you will see that the rate of evolution and the channel morphology itself is really different if we make the water come out of those closed depressions versus keeping it keeping it in there okay so more on that tomorrow but my other student Cecilia was working on what happens if we are moving groundwater to feed growing tributaries and I wanted to show you this picture again if you're a hydrologist maybe this is really boring to you but to me this was like I can't believe it groundwater divides don't have to be the same as surface water divides okay and in this case I'm showing you they're actually really really different okay so this is in central Illinois here's the Mackinaw River you can see the rivers are going northeast to southwest right and here's the groundwater divide which is basically orthogonal to that okay so the groundwater is not doing what the surface water divides tell it to do this is an extreme example this is a it's a pretty shallow sand and gravel aquifer but you know this is a kind of an extreme example what I'm thinking about more is is simply that those really subtle surface topographic divides are not the major force for this okay groundwater between these big and size river valleys is gonna move toward those big and size valleys even if it has to go over a half a meter I mean even if there's a half a meter of relief between there okay so we we know that when we look around toward the headboard end of some of these streams and this is this one's Minnesota this is Iowa I think that's Minnesota as well we see springs of groundwater coming out toward these channel heads here it's running on the top of bedrock but in these two it's this one I think it's running on kind of a lag of sand and gravel on top of a till sheet and here it's coming out at the interface between loss and till so there's a lot of variability in the permeability within these systems and there's places where the groundwater kind of tends to come out so Cecilia developed this sort of block diagram of a kind of a conceptual model of what we think is happening so we're imagining now this this steep slope there's gonna be a huge river you know this is like the Ohio River coming in and out of the page and here's a little tributary trying to grow up into the upland and now let's say we have groundwater coming from over here flowing through the system even though we might have surface water divides within there okay so you know again here's our this is our growing stream here's the cliff the elevation that 15 meter valley wall at the edge of the river and then we're imagining groundwater is flowing toward there okay so we started with the simplest thing we could we could do okay so we're not thinking about the veto zone okay we're not thinking about details of what is this aquifer like we're just saying there's water it's it's moving we're just gonna use Darcy flow to say it depends on some conductivity and it depends on the head gradient okay so we're gonna we're gonna make this model we're gonna say I'm gonna fix a flux in through this boundary here's my river valley on this side I'm gonna fix the head there at at zero it's coming out to the surface along this boundary all the time and I'm gonna just say there's no flow across the top and bottom boundaries okay so we got this and now we're gonna try to put this in Landlab so what do we do Landlab has lots is really set up to do this in a great way okay so Landlab already could handle spatially variable runoff rates you can just pass pass this field that's runoff rate okay so we we start by saying well let's let's have this unit runoff rate that's just equal to precipitation okay now we got to find where we think the groundwater is coming out into topography for now we just made up a rule and said anytime we erode down to 10 meters of elevation we're gonna put that water out on the surface okay so then we're gonna fix the head at that point it's essentially like making a new fixed fixed condition inside our model domain then we've got to iterate our groundwater model till it gets to a new steady state okay and then we're gonna calculate how much water would be coming out of all those seeps at that steady state we're gonna add that seep flux to the runoff rate field and then you just pass that over to existing modules in Landlab that are gonna route it by steepest descent in particular here we're not using any fill-and-spill closed-depression routing so we're just doing it's only gonna route the stuff out that's that's properly downhill and then we're gonna erode the topography as a function of the discharge of the slope so just to show an example of a simulation so here this 10 meter threshold would be the boundary between the kind of the orange and the brown and here's the head the steady state head that makes sense with that and you can see there's a little bit of of curving of those head gradients that happens okay so there's a little bit of focusing of water to those channel heads it's actually a little bit less than I would have expected but like I said I'm showing you the first six small runs that we made so so that this is probably a function of how conductive things are and so on okay so let's look at what happens we'll start with a case where we don't have any groundwater we're just gonna put precipitation on here equivalent to having a meter a year of rainfall and this is not exactly the initial condition but is is early in the simulation because the initial condition is so hard to interpret basically so here's our big valley here's our upland after 20,000 years we get this so channels are growing into the upland right here's what I'm gonna show you several graphs like this so this blue dash line is how much precipitation we're putting into the domain okay and the the solid line is how much of that water is making it out the left edge of the model okay I showed you a time in here if we evolve it a little more eventually we're getting all the precipitation out okay so it's all drained we're not putting in any groundwater but I'm going to show you green curves for groundwater in a second so we get a channel network we've got a river discharge in there and this is this is just reminding us what that topography looks like okay so I'm going to take 20% of that precipitation and say it's coming as groundwater instead so same total amount of water some of its groundwater some of its surface some of its precipitation okay and so what happens is my total discharge here in black is a combination of precipitation going out and the groundwater coming out and you can see that that still grows over time okay and we get a slightly different drainage network coming we're starting with the same initial random seed on top here but now discharge and drainage area aren't the same thing anymore in this model right that extra groundwater is focusing more into this stream than into other ones okay so we're having so a little bit of competition between these stream networks okay if I make 80% of the ground water come out of the ground and only a little bit over the surface now we're starting to see that this topography after 20,000 years is kind of different at this resolution it's really hard to even think of this as topography but what you're seeing here is that there's really not no area at these kind of intermediate elevations it's a really really steep scarf and we have more channels forming on the edge here and the evolution is actually a little bit slower and when we make it almost all groundwater and you can see that we're nowhere near kind of achieving getting all the water out of the domain yet okay so again drainage area and discharge are really different in this model but you can see some differences in the structure here we've got more more like six or seven big streams in this case instead of three or four okay so we can look through a whole suite of these things as we go from no groundwater to 80% groundwater if I showed you a hundred percent groundwater what happens the way our model is set up is all the water goes out that cliff edge that escarpment face and nowhere is it sufficient to actually start incising a channel on its own so you just have a line of seeps and nothing happens so I didn't show you the picture but we can see changes in the steepness of the escarpment changes in the sinuosity of the escarpment changes in the number of streams that grow so as we increase the fraction of the contribution from groundwater we get more steep parallel retreat of the scarf we get a slightly slower capture of drainage area there's more branching of the stream network with more groundwater and the the escarpment front I think becomes a little bit more sinuous at first and then less sinuous after that so I'm quite interested in this sweet spot of part groundwater and part surface water maybe driving some more variability in the stream network but early days here of course we see big changes in the stream profiles too so I'm showing here the profile of the longest stream evolving over time in these simulations and essentially they they're not uniformly concave right they get really kinked and especially as we put more and more groundwater in you get a real contrast between how steep they are a pie above the groundwater coming in and how steep they are below where the groundwater comes in okay so we've got a lot to do we we need to explore the sensitivity of the model to things like where we put the groundwater in what are the rules for for where it emerges onto the surface we're working on incorporating spatial heterogeneity into the conductivity of the subsurface so so that would just be sort of in the map view okay not in the not in the vertical we would like to consider the importance of filling of depressions over time we do use some diffusion in the model but the at the resolution we're at with the slopes we have it doesn't change the the depressions very much so we we're looking into that we would also like to include episodic fill and spill of surface water okay so again in lay lab that would be it would be easy to implement that we could just at certain time steps say okay go ahead and route all the water out that would be one way to approach that we could also think about more more subtle ways to do it say let's fill the shallow depressions till they spill and leave the deep ones disconnected finally we're working as group to document the rates and mechanisms of channel network expansion in the field so if you have any ideas about how to do that I'd be interested we're looking at some some places in Iowa where we can have we have Holocene alluvium of different ages that we can find at different points along valleys so to look for sort of headword expansion over time of those and we've also found some kind of interesting examples I'm going to just leave you with this one this is from the Minnesota DNR conservation magazine and it's a story about the mysterious disappearance of Miller Lake so this is a this is a lake this is actually a relatively high relief area this is a lake that essentially filled up to its spill point during a time when the ground was frozen okay and then there was a beaver dam holding it holding the the channel between the spill point and the next point down and essentially there was a catastrophic erosion event of that of that spillway which is again a process we don't we don't have in our model but looks like it happens at least sometimes this is one way that we can kind of get this strange network connected up and this lake drained which was remarkable to people who just went expecting to see a lake and it wasn't there anymore okay well so with that I will take any questions you have mm-hmm I see lots of questions we're gonna stick to one or two okay and then people can talk to Allison in a break a little bit more one in the back John yeah oh it's full of them oh okay yeah okay so I think we're so the question was how long does it take to drain aquifers feeding things yeah well some of these deep aquifers have very very slow movement toward them so there's very very old water in some of them what I'm imagining is important for this surface water connection are much shallower things okay so I'm thinking of these as aquifers that are actively recharging under the current climate so you'll notice that the precipitation in my model never interacts with the groundwater right so it's it's a it's a tricky it's a tricksy thing that we're doing right now because we really just want to highlight how does the morphology change if we change the rules about how we rub the water so it's not really realistic enough to deal with that question yeah yeah one more red that's the one in Florida right yeah yeah what's the difference um yeah so we're not thinking about special processes that the channel had that contribute to erosion and all we're saying we assume all this erosion is essentially stream power fluvial erosion so the only thing the groundwater is doing is contributing to energy for erosion right as to what the groundwater paths are so it's very dictated by the conditions of our of our model okay which is essentially if we're if we're trying to have these big fluxes of groundwater to be on the order of precipitation we got to push it through pretty hard and when we keep this no flow on the sides it becomes very strongly wanting to just be just flows you know straight toward the edge okay when the channel heads grow they they put another they kind of pinch it down to this zero head a little bit but it just turns out for the conditions that we've looked at so far it doesn't do all that much to bend the kind of contours of the head surface very much okay so the head surface stays really flat and so we're ending up with a relatively uniform feed of the groundwater out to these tributaries so right okay caveat being these are the first 10 models we front okay so I think that if we were to if we when we get to the point of being able to increase the resolution of the model or increase the size of the domain or perhaps allow for boundary conditions on the side that pass water between them I think it's possible that that that surface may become more sort of fluted okay that's what we were expecting to see essentially that if a if a stream could get in front of the others I kind of start warping the head surface toward itself that it would really win from doing that and we're we're not seeing that as a huge result right now we're seeing something like brown water appears to favor kind of a more uniform distribution of the water and so a more yeah not not channelizing so much so how to reconcile it yeah it's a great it's a great question we we need to figure out how to do that thank you others thanks