 Welcome everybody to day three of our CCDMS natural hazards meeting. We still have bunches of keynotes, clinics today and so it's like a full day and we'll end the day with panel discussion which is sort of a new item that we'll be doing. I've been looking around and seeing people being very engaged and enthusiastic at the posters etc. Thank you for that. We do have a request to do a little bit of homework in the sense that nobody, maybe this is the fault of the clinic leaders who were supposed to give out links to surveys at the end of the clinic and have people fill in like what they felt was the good parts and bad parts of these clinics and sort of get some data on how those are done. Lynn has not received very many responses and she sent out to everybody those links and so we highly encourage people to fill those in even now and then today at the clinics the clinic leaders will like prompt you again for their own clinics. So it's my pleasure to introduce now to you our first keynote speaker of the morning, Jenny Zuckbill. The real pronunciation is to call it but it's fine. When you like look up her profile and try to find what her science is all about you'll find that it's about the Navier Stokes equation in general and that she works on many topics and it goes from hazards like volcanoes and the Antarctic ice sheet and that's I think how we have met before. But what I loved was that it had like this little article that said it's the equation that keeps on giving and that's how she frames her work. Right thank you so much for this introduction so maybe the background story to that quote is that when I started studying physics the department had wrote down four equations and you said you have five years to learn those once you can explain those to me. You're good and as I was at the time I thought oh that can't be this difficult. Turns out I'm still working on the first one which is the Navier Stokes equation. So the work of our group has a slightly different twist than some other research groups you might have seen. We're interested in understanding the fundamental physics that connects a lot of different natural systems. So we're interested in understanding multiphase instabilities and how these can lead to really dramatic changes in the system scale behavior. And we apply this to mainly four different systems. So we have four projects in ice, fire, water, and rock. What you're seeing here is the shear margin of an Antarctic ice sheet. So that's the ice picture and that's the one I will mostly talk about. But we're also interested in understanding volcanic eruptions. That's the picture on the right where you see Stromboli erupting. And then we're interested in understanding induced seismicity for example in Oklahoma has a lot of earthquakes that come from dumping a lot of wastewater into the subsurface. And finally the near shore run up of coastal hazards like tsunamis and hurricanes. So at first sight these systems might be might look completely different and obviously they are but they share a lot of similarities and I think it is useful to study them jointly and try to understand why all of these systems have the common behavior of basically completely changing their dynamics over a very short amount of time. So we are not interested in understanding everything about these systems. We're specifically interested in understanding the extreme shift that all of these systems exhibit. So for example if you think of a volcano it just sits there happily for thousands and thousands of years. And then within a few seconds it creates this massive eruptions. Gas emission rates change by five six orders of magnitude. So why does that happen? And that happens in not just volcanoes but it actually happens in most natural systems. So I can think of these as four different systems or I can think of these as different realization of multiphase instability. So how do I think of an extreme event? I think of it as a system behavior that changes abruptly over multiple orders of magnitude. And the fact that it changes over multiple orders of magnitude I think has a lot of information in it. It basically means that we're switching the physics behind it. We're switching the governing process from say a flow behavior to a slipping behavior or something like that. And I think studying them jointly can help us understand, can help us create new constraints because they're all different realizations on honest diagram, the regime diagram of less to more solid and lamina to turbulent flow. In addition they also have thermal processes being very important. So you can see on this diagram that volcanoes probably exhibit the largest range here. They're also the most difficult to observe. So it's almost impossible to observe the interior workings of a volcano. So I think there is a value in studying systems that are similar but not identical that I can observe better. So for example you can think about degassing from sedimentary stack in a marine basin. I get very similar degassing channels that actually look quite similar. They're much much easier to observe. They're also a little simpler in terms of the physics. There's less sort of thermodynamic coupling to the fluid dynamics behind it. That being said today I want to present a specific project that I thought was of the most interest to this community and that's our project in ice dynamics. And ice dynamics I think is a really exciting field right now partly because I think we're really at the forefront of revising our fundamental understanding of ice sheets. And I think that process was started by satellite data like this one. So what you see here is insidata of the speed with which the ice is moving towards the coast. And right away you notice that this speed distribution looks very counterintuitive. So if I thought of Antarctica as just a huge ice cube sitting in a warming ocean I would expect that most of the melting happens around the edges. When you look at this map you see that that's not necessarily true. So look at these red zones these artery like flow routes. Those transport ice from the center of Antarctica right to the coast. And those are the ones that we're particularly concerned about when it comes to ice loss because these so-called ice streams and also the outlet glaciers we've had contribute 90% of the mass we lose from one of these. So I would argue that this is a process we really need to understand. When you show a map it's always really difficult to get a sense of scale here and one thing that's always mind-blowing to me when I think about ice specifically in Antarctica is the scale. So here is a zoom into the cycle coast which is one of the fast moving zones and that's a map of New Hampshire. So these are not small rivers these are not minor trickles these are massive state size chunks of ice that are moving extremely fast. What is so puzzling about these ice streams? The first one is that contrary to glaciers ice streams are not controlled by bottom topography. So let me remove the data layer that shows the speed and let's look at the subsurface. This is bad elevation from radar for the same zone I just showed you. I bet had I shown you this image before you wouldn't have guessed where those fast flow routes are because it's just not controlled by topography. That is very puzzling I have ice moving down a plane and for some reason it channelizes in these very rapidly moving zones that neighbor ridges that are basically stagnant. And take a look at the speed scale here at the color scale this is not a minor change in speed. The blue zone is about a meter per year the red zone is a thousand meters per year so I'm changing by three orders of magnitude which is completely puzzling. The number itself is also puzzling when you think of ice as a fluid because ice of course is very viscous and moves slowly and you can see this here in a photo that a friend of mine took off an outlet glacier and he hammered nails into this poor ice sheet here or this poor ice zone and came back a year later and it moved but it moved basically by a bunch of centimeters right. So ice is moving generally as a fluid is moving extremely slow so a meter per year is reasonable but how would I get to a thousand meters per year with a fluid this viscous and of course the answer to that is it's not the ice right it is the subsurface that is creating this movement. So when you look into the subsurface again here's a field photo you can see that there's really extreme deformation happening at the base of these fast moving zones of ice so you can see this very intense sheer deformation here in the sedimentary layers and that is really interesting when you think of it from a modeling point of view because it basically tells us that if we just quote unquote look at mass momentum for ice we're solving the wrong equations right so in a way these very fast moving ice zones are a reflection of basal processes right so if I want to model these accurately in a in a model that I have I need to think about the subsurface I need to think about the sedimentary dynamics and I need to think about meltwater so that's adding a lot of complexity to our models right there now complex models are great but it's always really helpful I think when you have a new puzzling phenomenon to start simple so let's start simple on the simplest thing I could come up with this force balance so here are our streams the dark gray zone is the slipping portion so it's not flowing it's slipping right that gives you this enormous speed increase and it's slipping because the sediments underneath are failing so sediments are have such an interesting material property right that in if I change the water content of sediments it can get them from being solid to fluid and that transition happens pretty abruptly which is really quite unique so it's a very novel in your material and that non-linearity is really I think at the heart of the dynamics that we're observing here so if I just do a simple force balance of gravity driving the stream and basal resistance acting against it you will notice that these are not balanced by any stretch so that could suggest that we're simply in an unstable regime or could suggest that we've missed part of the picture I would argue we've missed part of the picture here and the picture we've missed is the margins what if label tears margins which are basically the zone we get transition from the slipping to the slow moving ice so let's add this in so I have a lateral stress that happens throughout the ice column and this lateral stress for a steady state would need to balance the difference between the other two integrated over the half width of the stream you're already noticing that when I think about this lateral stress it depends on the width and the height because I'm integrating these stresses over those length scales and that means that this lateral stress will actually depend on what the width and the height are so let's think about this a little more carefully because if this margin really plays a critical role in the force dynamics I will create a lot of shear and with that I will create a lot of shear heating that is important because as ice gets warmer we all know that it gets much weaker right that's why ice cream is delicious it's the perfect mixture between the solid and the fluid it's not an ice cube right it's still frozen but it's sort of this mushy delicious texture so what's the simplest thing we could come up with to estimate shear heating the simplest thing we could do is think of this simply as a crack the sliding portion would be the crack and if I do that I can leverage the classical techniques of vector mechanics to estimate shear heating analytically that was our starting point here and you see that this is the domain we're considering we're neglecting downstream variability for now which means we can reduce this basically to a 2d cross section because of symmetry I can cut it in half in the lower domain the dots indicate that the ice is now moving towards you and I have a slipping portion I have a non-slipping portion and between that I have a singularity now the fact that I have a singularity is interesting because singularities are giving basically extreme stress concentrations now I can end as I can estimate the shear heating if I do a J integral approach which is well established in in earthquake mechanics we do it a little less I think in fluid mechanics but the basic idea is that you try to integrate how much heating you experience around the singularity so you assume that out of the heating is due to the singularity I won't go into the details of this the bottom line is that this factor J tip which basically multiplies how much shear heating I have so that's the key factor here it captures how much I'm forcing the system depends on the lateral stress so I can use this to estimate how much lateral stress I get and how much lateral strain I get and if I then appreciate the fact that ice rheology is very non-linear so I can relate the strain rate to the stress itself but that's not a linear relationship that's a relationship to third order what I'll basically find is that the shear heating which is the tau epsilon dot term scales as width over the height to the fourth power so you just kind of might be wondering like why the heck am I going into this much detail about it well the answer is very simple shear heating scales with width to the fourth power that means if I disturb the width of the stream a tiny little bit my shear heating in the margin will basically go through the roof right away and that means my lateral stress will drop significantly but this lateral stress would now have to take up more of the force balance because I've widened the stream so I'm actually weakening the term that could give me balance here right so this weakening due to the shear heating basically means that I have a positive feedback you appear if I disturb the width of these streams a teeny tiny bit I would get a dramatic increase in shear heating which means I get further weakening of the ison of the stream which means I would keep widening and widening and widening and all of these ice streams should basically eat up all of them well that's kind of exciting but it's not what we see right so the data doesn't suggest this yes ice streams are variable in width and they're very dynamic but they don't widen without control right so I think the power of a simple analysis like this is you can see that the system could be prone to instability you also see that this instability is not the full story we've clearly neglected an important process here so let's do this numerically because it gives us the flexibility to relax some of our assumptions I think we've appreciated by now that we need both the mechanics and the thermal problem so we're doing an america model here that is simple and just a simple 1d stoke solve we're going to compare it to field data here for one of these ice streams I talked about bin chadler and the great thing about this ice stream is we have very detailed speed data across the shear margin and when you look at the plot here in the center you see that the speed localizes significantly as you go downstream right which is indicative of physical processes changing on the right hand side you can see our numerical solutions for temperature and I want to draw your attention to this brownish orangey zone which is the zone in which ice achieves temperate status which basically means it's at melting point right so that's very interesting because it basically means that in these shear margins actually start melting the margin at some point right so actually start melting the ice let me emphasize that this is not fluid this is just 1% water in the ice right but that weakens the ice dramatically and that's an important insight because it means water clearly is an important player here and that water will be generated in the margin so it's not the same as the water that's generated under from the frictional heating it's a marginal water that we care about so let's integrate that into our model so now we have a hydrology layer here between the ice and the tail why is that consequential you see that on the right hand side here the reason it's consequential is if this water channelizes and it likely would because I'm inputting a lot of water in a very localized position namely the margin the consequence of that is that the yield stress of these sediments will change dramatically because the water changes the pore pressure and the yield stress depends very sensitively on pore pressure which basically means if I have a channel I get an increase in the yield stress which is related to the fact that I'm excavating the water very efficiently in the channel so the water basically the sediment right next to the channel is very water-poor and that's really interesting because that gives me the opposite effect of what I've noticed before which is that if my width change creates enough water in the margin alter the hydrology then I can actually locally increase that strength and that can stabilize my margin so what's interesting about the water here is it has this dual effect right so it destabilizes the stream when you start melting because you get the slippage because you're weakening the tail and the ice stream starts sliding but it also later stabilizes the ice motion when it collapses in terms of the hydrological systemic forms and I think that's a really important insight for this system but maybe more generally you said when we think about feedbacks I think it's important to think both about the positive and about the negative feedbacks and that helps us to identify fundamental tipping points in our systems so what I just said is summarized here in this picture so we're worried about these ice streams basically widening if they were to widen we would just dramatically increase the amount of ice we're transporting to the ocean in the regime where I have a thin water film I could get slow gradual widening if I have channels forming in the margin I would stabilize the margin so I would have a typical width and that's actually what we see in Antarctica most ice streams have a pretty typical width that correlates nicely with where I would start to see melting in the margin however if I go to a lot of water at the subsurface and I get these distributed cavities in the ice planes things again start changing very significantly because now the ice stream can basically jump from one location to the other and that is something we actually do see in the data so here's a very interesting case of an ice stream which is KMI stream and it suddenly died for some reason that we haven't quite figured out and we're thinking the water was critical to it but we're not quite certain and during the shutdown sequence the margin jumped and that's what you see here in the picture you have margin one and then you have margin two and these margins are separated by a few kilometers but in the ice record so in the radar data that you see at the bottom here which has been used to identify these margin locations there's no transition between those you kind of go from one location to the other so think about that you have a new hamster sized ice stream that just suddenly shifts location just continuously which is just mind blowing to me that that's even possible however if you think of it from a water percolation point of view it's not that surprising because water can definitely reroute on the order of a year and if the ice just mirrors the position of the water and the yield stress pattern that the water creates and I can discontinuously shift from one location to the other and that's something that we see in our model and we can reproduce this behavior in a forward way so we're not trying to match data we're trying to identify we're trying to use the data to identify fundamental physical feedback loops and I think that's an interesting lesson both for this case but also generally that I think we can use data more creatively than just validating our model I think sometimes it's really interesting to look at outlier information like this one because it helps you understand something some of the more maybe extreme dynamic potential in a system so what does this mean from a modeling point of view and sort of from a more fundamental point of view in terms of extreme processes what I'm basically arguing here is that meltwater percolation at the scale of individual sediment grains governs ice stream dynamics which in turn governs ice mass loss from Antarctica so we've made a link between grains that are 10 to the minus 4 meters to an ice mass that is thousands of kilometers large which I think is just a really mind-blowing insight to me just how profoundly profoundly non-linear these systems are so from a modeling point of view I think what we need we have a really dramatic multi-scale challenge here and the way we're approaching that is we have a granular scale model what you see here is a 3d GPU based computation that couples the ice motion on top so the blue zone is where the ice is moving slow the red zone is where the ice is moving fast but only visualizing the sediment grains here and it has water percolating into this and then you see this coupling between deformation and poor pressure dynamically emerging from the model we use that to develop a contained set of equations for the subglacial hydrology so we just take that physical insight and plug it into our ice stream models those ice stream models are free boundary problems so we're trying to solve for the position of these margins rather than imposing them and we do it based on a hydrological model we have developed at the granular scale and then we plug those into ice sheet models and that's something that we started very recently and we're thinking about these margins basically as boundary layers we have a boundary layer approach to this where we're just kind of using a different approximation in these very thin shear margins and to understand how they couple back to the large ice sheet dynamics and one of the interesting things for example here is in this granular scale model you can actually see the mechanism through which you form these channels which is basically that the input of meltwater locally increases my porosity because of shear dilation so I have a lot of shear so the sediments will tend to dilate at the same time I have a lot of meltwater influx in that precise location and that gives me the feedback I need to create these subglacial channels that we also see at the surface so there is a lot of evidence at least significant evidence that those are real so let me quickly summarize conclusions from our modeling point of view that I think go beyond this specific project is multi-phase interactions at the granular scale can really trigger system scale dynamics right so we often parametrize these small scale interactions and that's obviously the only thing you can do if you work with the larger model but I think it is valuable to think more deeply about what are the dynamic regimes right so what are the fundamental different types of behavior I can get out of the granular scale and how would that translate to a fundamental shift in the large scale dynamics the second insight that we get seeing over and over again in our projects is multi-phase flows is just profoundly non-linear so when you take a course in non-linear systems you sometimes get introduced to this butterfly effect where you have like you know a butterfly changing climate and I have to admit as a student I always thought that was kind of surreal because through this I guess it's possible but most butterflies don't alter climate right so it always seemed a little bit abstract this project was the first one where I thought like wow you know there is like a bunch of sediment grains in the sheer margins of Antarctica that actually could have a dramatic effect on the continental scale which is just surprising that was the first time I noticed something like this but it's just interesting just how non-linear these systems are and one thing I would encourage you to think about maybe also in your research is what are the positive and the negative I think sometimes we get carried away thinking about instability and yes that's super interesting of course but the question is sort of started once I started instability will it really propagate forever will it really just kind of grow out of bounds or is there something at some point that will actually stabilize it and I think that's something that certainly for Antarctica I think we need to think more carefully about and finally I think data plays multiple different roles and our challenge here is I think we don't have necessarily enough data I would love to get more data from Antarctica and we're actually going and getting some but still like you know I feel like we're really limited by data but I think also that means that data should go beyond validation I think it should should be we should maybe think a little bit more about outliers or unusual data sets or things that don't fit the normal picture and maybe there's an information in that that we've missed so that was obviously the main thing I wanted to talk about but I just quickly wanted to mention a few ramifications of our work on Antarctica that are not purely on the physical side and I'll just give a very quick teaser on this because we've worked over the last year very intensely together with stakeholders in debate to think about what are the ramifications ramifications of these insights or sea liberalized adaptation planning and this is the Bay Area it's beautiful I love it sure what's not to love but this is also the Bay Area so here you see flooding last year in San Jose you see flooding of the Imbacadero that's 2013 and school closures because of flooding in 2014 so we see flooding events routinely now in the Bay Area but I don't think we've thought enough about what the consequence of these flood events are and one of the key insights I think from an Arctic ice sheet progress we've made or progress we've made in understanding ice sheets is these ice sheets are very non-linear right which means that the uncertainty goes through the roof and there's been an attempt to quantify that so this is work from the rising seas report in California and you notice that these are the RCP are the usual climate models that I guess most of you have seen and then there's this H++ scenario which takes into account the some of the instabilities or profound linearities that we're worried about I should say that this is not the exact same instability that we've worked on there are a bunch of instabilities that could be important for Antarctica that's a different one but it's also a very important one and of course really what this plot wants to say is that we have this enormous spread of possible flooding notice that say in 2050 this uncertainty goes from basically a meter of sea level rise which I think would change the Bay Area as we know it very profound way to zero the uncertainty is mind-blowing here what do we do with that I would argue we can't wait until we know better because I'm not sure we ever will there is always this possibility for dramatic instabilities I think we need to start thinking a little bit more creatively about how we deal with this so we did a quick analysis of what the risk is and you will notice if you see so we're meant we're overlapping hazard exposure vulnerability to get to risk and you notice on the lower left plot that the difference between these two scenario RCP versus the H++ scenario are dramatic and that's only for the next 20 years right so we're not projecting far into the future at all and the costs of this will be burdened to the commercial and the residential sector obviously some costs will eventually be publicized but I think that's interesting and when you look a little bit more into in the residential sector who will be affected you will notice here so this is a key means clustering analysis that we did to get a better sense of who will be affected if you look at the fourth category of rent burdened families so the label renters isn't a great one it should really be rent burdened which means that you spend more than 30% of your income on rent the renters will be dramatically affected by this right so one of the lessons here is I think we could it might be time to think a little bit beyond just limiting physical effects like blood walls because if you have this much uncertainty it's difficult to plan a flood wall well to think maybe about interventions that would limit the vulnerability of groups that we now can already anticipate will be heavily affected by this that's something very much work in progress but I just wanted to start the discussion about what do we actually do with this kind of information well how does it actually translate to the ground and I think it can translate to immediate action but not in terms of providing a specific number but instead of providing more thinking on adaptive solutions on creating a portfolio of solutions rather than focusing on just a single one with that I'd like to thank you and finish on one of my favorite quotes which says uncertainty is an uncomfortable position but certainty is an absurd one thank you thank you Jenny for a great talk are there questions I'm closer with my microphone here in the front so I'll give it to Brad first thank you wondering about the the channelization is there a component of ice melt involved there is it purely channels getting created in the sediment an excellent question that's what we're working on right now so the classical so the existing hydrological models we have assume hard rock at the base so they assume that out of the carving will carve into the ice if you apply those models to the cycle coast which was the region I showed a couple of times they actually don't work because the creep closure would dominate over the melting that you create so you basically you wouldn't create enough melt necessarily to see for the relatively small slopes that we have in this region right so those if you carve into the ice there's good reason to think that those channels were close so if I think about this as a worthless burger night channel it's a stretch so I would argue that most of those channels start carving into the sediments and that's what we are trying to explore in our granular scan models right because I expect to see so there are two mechanisms that could give you that one is the pore pressure related failure of sediments which basically excavate a channel rapidly right away could also do a sheer erosion and that would likely be more important in the actual stream but I think there is a lot of opportunity to come up with a better subglacial model that really takes into account the fact that we have hill which basically means we have very fine grain sediments there so that's like two of my students are working on that right now because I feel like that's a critical piece so the original model I showed that assumed carving into the ice but it's just because that was the only model we had at the time I don't think that's the right thing to do necessarily that was a fantastic talk thank you so my question is in regard to the application of this to warm water ice shelves can you switch the mic a little bit is that better yeah that's great excellent all right so my question is how your results relate to warm water ice shelves because we know from the larson b collapse that the back stress from Antarctic ice shelves is significant for velocities and removal of an ice shelf can increase velocities by around 200 percent and we're seeing those velocity increases in warm water ice shelves in the months and see so I'm curious how how these if these results are specific to cold water ice shelves or if there's also an application that applies to the warmer excellent question so right now our models don't take into account the grounding line dynamics right so the grounding line is basically the piece where the ice becomes floating so it's the transition from land-borne ice to floating ice right now our models just assume you have an infinitely extended ice stream and we're just kind of looking at you know how would that vary in width and speed right now one of my students is working on coupling the two right because I entirely agree with you if you have the collapse of a shelf you could get a dramatic speed up of the stream behind it right and I think one important thing that we're worried about is how would that affect the sheer margin because you change basically the the the buffering completely right so you change the stresses that the shelf provides which means that you will certainly speed up the flow the question is will you also why it's not dramatically right so that's something that we're currently working on specifically in the context of swathes and pine island so we actually have a field campaign going out to swathes to measure the to assess the fragility of the margin that separates pine island and to those of you maybe just a quick primer the amundsen sea which is basically swathes and pine island or the piece of an arctic where we're losing the most ice mass right now and it's also thinning very rapidly so it seems to suggest that that could be in an unstable state right now and what we see in paleo data is that these two glaciers were merged once far in the past so the question is for example if you have a system like that where you take out the shelf would you see like a more dramatic collapse event that actually merges these two for example right so what we're working on right now is a linked free boundary model where you couple the grounding line dynamics to the sheer margin and to the ice in between but yeah I think that's definitely going to amplify this kind of behavior in a potentially very significant way right so I think we're just at the starting point of really understanding how dramatic some of those nonlinearities are and how those instabilities from the different components of the system can build up to a very dramatic potentially very dramatic disintegration event thank you Jenny thank you Jenny