 I know we're starting a bit late. There was someone in the audience who, this slide is not directly related to my talk, but someone I spoke to on the first evening said something about, are there really volcanoes in Africa or are they dangerous? There's not much activity in Africa, is there? And I always, I thought, well, there's an opportunity to educate this as a school. In this diagram here, all of the boxes in clothes confirmed, either eruptions or intrusions that reach near the surface in a relatively large volume. Intrusions in all of those boxes, and the one on the left, this near-a-gongo, Yamuragira, there's actually, I'll show you a picture, I think, at one point, in the preparation for a laboratory this afternoon, two volcanoes, 15 kilometers apart, both with lava lakes, that have really exciting activity as well. Lava lakes, the longest-lived lava lake on Earth is the Earth-Aulea Volcano, or continuous Earth-Aulea. And many of these are volcanic islands in the Southern Red Sea. And there are more volcanoes showing signs of unrest or high pressure per unit length along the East African Rift than there are along the Andes. So yes, magnetism is a critically important process in that part of the world. And it's what draws me back to field studies as time goes on and on. Now, I feel really privileged to have the opportunity to speak to you here. Unlike the other speakers that we've had so far at the meeting, I'm not a theoretician. I work with theoreticians and trying to acquire the most critical pieces of information and the data sets, and to integrate then and work closely. I think that what I'm trying to present to you is that taking the theory, doing sensitivity analyses, and knowing exactly what observations are critical to a problem, are what motivate my research. And I hope it's part of the reason that I'm presenting here at the meeting. Now, what I wanted to talk about, I've been working, what I'm presenting now are volcano and magmatic intrusion and fluids, fluid movements within the cross in areas that are in extension. And at the very end, I'll comment about magmatic systems in compressional environments. But I want to try to emphasize the point that's been made elsewhere, particularly with Torsten's talk, that there's also a tectonic driving stress that may or may not be a coincidence with the body force is induced by the magnetism itself, its density contrasts, and the surface topography. And so knowing the state of stress is a really important part of setting up any problem in these areas. So let me continue on. So I have two parts. I want to talk and continue on from what Torsten was presenting and talk about some ongoing problems with dicontrusions and challenges we face in trying to apply rate state friction. I also wanted to say that I'm trying to work with really smart people who are trying to extract information on the dimensions and distribution of melt and its pathways, not just in the chambers and directly beneath the volcanoes or in these larger volume dicontrusions, but actually from the mantle to the surface. And some of the physical property information and the differences in the different methods are starting to show promise in being able to understand those aspects. So I'll touch on those. I'm covering a lot with the methods. Because of your diverse backgrounds, I won't say too much about the methods. I hope you trust that what I've chosen are published papers with really reliable data sets and interpretations. So I'm getting used to these. So what I'll do is because Torsten just presented so much on dicontrusions and still simply and not completely is take a diconturn at sideways, obviously its relation to the free surface is quite different. But in many ways, we can think about the stress concentrations along the edges then of these zones where the intrusions are occurring. I'm going to talk about detecting melt and fluid pathways and then briefly talk on separating out source effects in terms of the volcano sources and fault magma interactions versus seismic waves passing through fractured material that may have gas or even aqueous fluids within them and how that will change then and how we interpret signals from volcanoes. And that leads then to what we'll do this afternoon. And this afternoon is more looking and having you evaluate some data sets from a very active volcano from Nirogonga volcano. So where am I going? The other part that I want to talk about as well is the presence or the clues that we gain from active process, obviously, seeing a particular volcanic eruption in real time. We gain a large insight, but understanding the time average strain patterns is an additional constraint we can put on these problems are the conditions that we're seeing at any point then. Changing with, are they relevant over longer time periods and how does that change the density distribution in a structure within the plate zones? So here's a really simple, in a way, it's a cartoon. And it's just meant to hit on a variety of processes. And I think we all know that we can have volcanoes and lava and exclusive lovis at the surface. We can have complex systems within the crust. But there's other things beneath the crust. And we're generating the melt, in most cases, through adiabatic decompression melting. Enhanced temperatures in some areas then will increase these problems or through the addition of volatiles in the subduction zone setting. But I'm drawing the rift zone example, because it's the one I'm most familiar with. And most of my examples will come from there. I want to point out then that we're introducing fluids from beneath the plate and they're rising through the plate, sometimes stalling and freezing, and then maybe being reheated again. In other cases, then, rising very quickly from the mantle to the surface. A kimberlite is an example of a lava that's filled with us, the CO2 gas that basically bursts up through the plate and maybe a week or two. There are mellotites that may come up in a couple of days from the base of the crust as well. So really fast, depending on the volatile content. So volatiles are really important in the process. The changing boundary conditions within the tectonic situation though can also become important in terms of fluxing and changes in the distribution of the surface topography or even migrating volcanoes. We've talked about the sisters before in terms of local stresses, but in many cases that the boundary conditions are going to change over time periods of hundreds of thousands of years or a million years. And so the volcano may be in a more or less favorable situation. So I'm not going to worry about the time evolution of systems. I'm just trying to place it in context. But one process I really wanted to emphasize is that we know that in former subduction zones that subduction process itself may have rehydrated the mantle lithosphere. So we may have an additional contribution of volatiles, particularly water and CO2 that are able to participate in the magnetic process in addition to the magnetic fluids that are generated down here. So this process of metasomatism, metasomatized mantle can also happen by reinjection or the addition of fluids through mantle plume processes. And so we know ocean island basalts are enriched. And many of these mantle plume provinces have unusually large amounts of CO2 gases. So we have more gas than just the volcanoes themselves in many settings. And I'm going to talk about an exceptional one, and that's the edge of the Archaean craton in Tanzania, where you'll see in a minute, CO2 degassing and water degassing as well are important components of the process and the failure process. So they're changing the status. So down below on the bottom part of the diagram is a diagram that actually Michael Manga had. He had also stresses against it. I've just put the process versus the timescales over which they are operative. In any one situation, though, we can have intersections or a superposition of these different processes and timescales. So full slip happens like this most of the time. We have a seismic slip that happens over periods of days. Diking processes, again, over periods of days. And I added this from Michael that over periods of thousands of years or so, the glacial loading. And as he pointed out, it's actually a very large force. And then the magma buildup and stress buildup within a magma chamber, though, is on the timescales where tectonic processes are definitely going to intersect and be important. So I just wanted to add that. Now, I may joke about this a lot. I am just joking. I really don't. Anyway, I'm exaggerating just for the point. Volcanoes are not cones on top of straws that come down to some pressurized cup down below. And we're well aware that that's not the case. Obviously, to develop analytical models for many situations, we start there. And then we add the complexity when we understand the first order parts of the problem and the system. I'm going to take, though, and I love this. I think this is a very beautiful volcano. This is O'Doniel and Guy, and it was erupting. This is just a little burp one day in 2008. It was doing this for about eight months before the caldera collapsed. But this is actually O'Doniel and Guy, right? So you see this, and you think, wow, that's a cool volcano erupting. It's just like this. No, whoa, what's this? Another volcano, another volcano, another volcano, another volcano. What are these things? Really big faults, and there's a big base and adjacent to it. So you can't isolate, in this case, a single volcano without trying to understand the context. And I'm going to come back to this area. You should also talk to Sara J. Oliva about some of these processes. But you'll see this little circle right here in this volcano. This is an area where we had, in 2008, I'll explain the sequence in a second, we had a very complex sequence of a fault episode, a dicontrusion, and a volcanic eruption. So I'll come back to the question that we had before. I'll start off with stress-triggering volcanic systems. Now, towards the end, I'm going to talk about a point out, new avenues, new directions, without going in any detail with any method, how we can try to understand the pathways through. Now, this is, from a pathologist's perspective, this is Anand and Steve Sparks. I'm not sure who else, the other co-op? Hm? And Bloody, right. Looking at heating and intrusion down here in the bottom, it's the percent melt present within the crust and their hot zones within the base of the crust, stacked systems within, and multiple pathways leading up to the surface. And so intermediate storage and then the eruption. Why I put this in, okay, there's also, there's no cumulates anywhere in the system or no discussion or real identification of cumulates and how they, what we do with them after we extract the materials as well. And so that's like kind of a new part of all of the systems that we need to be considering as well. But what I wanted to say is that we have multiple pathways up through and we're altering the rock on the way. And trying to understand that part and what types of cracks and fluids we have along the way become important in determining the rheology of the rocks, the time dependent behavior. And that's one of the biggest limitations when we, we could develop semi-analytical or numerical models for situations but we really know little the deeper we go about the compositions in that and the time dependent behavior of those rocks. So let me start off with three separate examples of indications where volcanic eruption or unrest is triggered by a dike intrusion in this case. So we can talk about distant earthquakes or dynamic triggering of volcanic eruptions but we can have stress triggering then from nearby interactions or even faulting episodes. And again, many cases it's hard to see how the faulting that's probably induced by injection in one part of the system then may trigger or change the stress state and cause flow of material that then leads to these eruptions. So the first one I'll talk about is in, so in many cases, sorry, let me step back very quickly. Crapla eruptions in Iceland, a whole sequence informed the world of dike intrusions. We know that they happen on mid-ocean ridges but we saw the sequence and knew that the central reservoir in the center of a mid-ocean ridge segment was shooting dikes and along the link to the rift in both directions. And we've seen a diagram, Torsten had a diagram showing that situation. What I'm talking about here that's separate is having a magma, a central segment magma chamber dikes coming off that system and then interacting with other volcanoes in the system. Okay, so in, whoops, in 2005, sequence of events occurred, so I'll just explain. But unfortunately, there were very few instruments working operative at this point in time, so most of what we know about the very first and the largest dike injection that happened in 2005 come from global records. So there were gas emissions, Simon Karn at Michigan Tech detected large SO2 emissions in this area at about the same time that there were these massive earthquakes warms and then groups from our great long-term colleague at the Geophysical Observatory in Addis Ababa alerted us and said, listen, this is our opportunity in Ethiopia to try to build infrastructure because there was this long three-day period of almost a continuous earthquake activity in the Afar. At that point, we didn't really know what had happened. I'm just gonna quickly show you, but many of us came because we went with instruments not knowing what was happening, partly driven by curiosity and partly driven to try to turn this into an opportunity to build infrastructure for hazard mitigation in Ethiopia. So, and it was through actualized vision. What we knew after the fact then was that there was a central magma chamber that with top at about six kilometers, so more than twice the depth of the Iceland magma chambers. So a deeper depth, a different system in fact that fed dikes that went in both directions, immediately after it, but then at that point, within the two-day period, a 60 kilometer long segment then was lit up with earthquakes and these dike contrusions. One of the dike contrusions came up and curved into Dabahu volcano, which is this thing sitting up in the, right about the mid-second segment, looking up along the length of the rip, and there's a broad volcano at the northern end. So the dike came up, interacted with it, went past it over on the eastern side, went by some cooling, highly fractionated, small volume magma chamber causing an explosion and the eruption of these ashen pentellerites from this vent. And this is the northernmost segment of it and it stops about at this magma chamber or at this volcano called Dabahu up at the top. So she's blasted through loads of things along the way and it took us a while to figure all this out because there were so many interactions and this wasn't the story that this eruption here was relatively minor and accidental or because it interacted along the way. This is an interferogram spanning the time period of the earthquakes that are shown on top. So this was after we got out there, this wasn't the main earthquake sequence. And like I said, there were only a few stations and so the earthquake locations are poor and we can't build a really nice story from that. But what you can say from the interferograms is, oh, from these interference fringes, we can see in this case there were some subsidence at the center of the segment, roughly where we have the magma sources. This is Dabahu volcano that went through a period of inflation and then deflation afterward. A lot of earthquakes happened, it did not erupt but it was in a period of unrest and probably near critical. This is the shape of the dike that came up and curved and interacted. This is one of Sandy's beautiful dipping planes that come down, this defines a nice dipping plane. You can talk about that afterward. And then this is the segment of the dike that moved northward, this is where the eruption site was. And then there was magma drained or there must have been a source that was tapped with material that moved into the dike or moved away from a magma chamber here because there was rapid subsidence beneath this volcano. So all of this happened within a period of about six months after the original dike intrusion. So magma source from several different areas, whether it joined the flow of the intrusion and whether what and how that happened remains unclear. All right, because subsequent dikes did not do that. They were just sourced from the center and none of them traveled as far. But going back to this question about arresting dikes and asparities, I'll show you an example of one of those, we've had multiple examples. But staying with this idea of stress triggering, this is a paper that was just submitted I'm working with Carolina Pegley and Derek Keer and Seng Ho-Yoon, where in this top diagram, looking at the aftermath of this sequence with many multiple dike intrusions added, how the stress state will then change if you think in terms of Coulomb stress, that we inject a lot of material open as our own. What will happen then at the tips of the area then that were where these intrusions occurred and the cumulative opening is more than eight meters along the entire length of that segment. So the left is the first invariant of the strain rate tensor and here is the maximum strain rate and basically intruding that 60 kilometer long zone causes stress concentrations at the tip, two lobes at the northern end and it's where we see the seismicity. So the aftershock sequences and then induced seismicity by the stress concentrations at the northern tips, we see a large zone of earthquake activity here and up to the northern end and then enabling connection with the next segment and then enabling some strain continuity across. This would be Djibouti just right here or the acile rift is just over here and with a connection zone across the two. Superposed on top of this diagram are predicted earthquake or simulated earthquake epicenters from a stress concentration, a dog bone shaped stress concentration between the two. So the dyke itself induces seismicity after through this large scale or tectonic stressing across the zone. We hope that we'd have had, I think, I know Paul's well aware, we'd hope that we'd have great state-friction in first models of this sequence to be able to present at this meeting but one of our colleagues has been delayed for an extended period of time. So we don't have that, but hopefully these will progress in your future. I think also, yeah, nevermind, let's keep going. A third example is not proved. It's speculated in this beautiful paper or a title of a paper that I love. Colin Wilson, a student of Colin Wilson's has a paper, the invisible hand tectonic triggering and modulation of a rhyolitic super eruption and their speculation based on the very different lava compositions that they find at the edge of this magma zone. So we're in Taupo, this is New Zealand, we're in a back arc spreading situation, subducting slab going down over here, Taupo, several large volcanoes and at 27,000 years ago, there was a super eruption, about 530 cubic kilometers of lava were erupted and they showed that there had to have been some chemical mixing from another source immediately prior to that eruption. And so they say that it was a dyke intrusion along the length of the rift, something like the Dabahu example that triggered then the super volcano eruption. So it would be as if the Dabahu volcano then had gone into an eruption and had a much larger magma chamber, more fractionated material and larger gas content and cause the super volcano eruption. So it's possible, but here's one another example of the potential for this. Another example of induced stress triggering of a dyke intrusion then by faulting processes and then later a volcanic eruption comes in 2008 in an area where we've subsequently done quite a lot of work. And I'll tell you what we feel is probably the case now that we know the subsurface structure. In 2000, and sorry, this was published in 2008. In 2007, there were a series of large magnitude earthquakes. So that instead of a main shock after shock sequence there were actually several large magnitude magnitude 5.9 earthquakes and then a dyke intrusion. And fortuitously, there was a, what's it called? The satellite. Anyway, in sorry, I can't remember which satellite it was. I'm drawing a blank, sorry. Somebody help me if they remember. But here are the interferograms from the first pass immediately after this earthquake and there's indications of a fault slip. On the southern end of this volcano called Gelai, there's Oldonulengai, okay? So it's in between the two of them. And this was the 5.9 earthquake and it was simulated by a shallow rupture to the surface along a 45 degree dipping fault plane but well away from border faults within the basin system. And then immediately afterward, two days afterward, three days afterward, there was a dyke intrusion that in the absence of any information, anyone would say, well, you've got a big volcano there. It's on the southern flank. So it came from a chamber underneath and propagated. Southward beneath and then a month later, well, without earthquake recordings, activity started at Oldonulengai and it went into a nine month period of eruption and then an explosion and then called there a collapse within the middle. It's a relatively small. So here's another example of these situations and what we now know from a whole series of experiments, I'm only showing you the tip of the iceberg. In terms of data, there are many groups involved. From the US, France, Tanzania and Kenya, we worked across borders at instruments on both sides. These are the volcanoes. Again, I'm coming back to, see this is a nice circle and then beneath it another almost circle of earthquakes and a cloud of earthquakes that plot then in this way on top of a tomography. This is just one slice from a tomography model that's the joint inversion of ambient noise tomography, PNS wave arrival times and gravity that provides a constraint on the cross mantle boundary. This is not very well-image. Ambient noise gives you a lot of resolution in the upper crust. The PNS arrival times throughout the middle crust and the lower crust isn't very well determined. The gravity is a nice additional constraint. And Steve just sent me a couple of days ago. So here, let me take you through this. We have a large number of earthquakes that are happening in this area. This is where the 2007 dye contusion occurred. We know the spatial position. We just don't know the earthquake locations particularly well. This is the mechanism for the main earthquakes in 2007. It's located here. We think it probably should be right here. And what we're interpreting is a stack sill complex between two volcanoes and above a magma body and the lower crust. And here's a persistent zernic seismicity that we suggest is a zone of feeding zone rising up to this dike sequence. Here's Odonia langdai over here that erupts carbonate to the clavis. There's the big fault system bound in the base and crystalline crust in through here. So high velocity against low velocity, as you can expect, across this zone. So we think that what happened in 2007 is that the sill was inflating and it faulted on the side. That's why it's a relatively shallow earthquake ruptured to near the surface or to the surface. And then the dye contusion emanated from the sill and propagated northward into the volcanic complex. Just a little bit went a little bit northward up into the southern edge of this area and through here. We can't prove that, but we're working now reinterpreting the INSAR with Christelle Wotier. So that should provide some really nice insights on these interactions within this layered complex. I think for the interest of time, I'll just skip through this. This is more time average deformation and showing similar sequences. What we did in this case is a numerical simulation, putting high density bodies in bedding high density and lower density continental crust to look to see where we see stress concentrations along a segmented rift system. And then changing the densities and making them hot, put them in a phase then of activity or replenishment and seeing that they have a tendency to grow or propagate along the length of the rift. So stress concentrations at the tips of the segments, though, are quite important. But I don't want to get into that. I don't think we have time. Let me move ahead because, right. Now I want to talk a bit more about the sequence in AFAR because it goes back then to what Torsten was presenting about the diapropagation. And I'll then say some things about the whole Iran. I'll make some comparisons because there's some big differences between the two of them as well. This is just a summary diagram showing all of the dikes that occurred after the main sequence. Some of them, they emanated from this largely asysmic zone in the middle from a magma chamber with top at about six kilometers. They propagated northward in some instances and they propagated southward in some instances as well. But they started from here. Fissure eruptions came up in the largely asysmic zone. There were three fissure eruptions that came to the surface here from southward propagating dike intrusions. And this is just the zone. And through here, there's the Dabahu volcano I was telling you about that was activated in the first stage. Subsequent dikes never made it that far. And then this sequence here is exactly an example of a stalled dike sequence where it stopped and then restarted again. This is a summary then of the propagations and rates. And I'll show you, there were questions about the magnitudes versus time for dikes. I'll give you an energy release as related to the propagation. I'll show you diagrams of that in just a second. But this is a summary from Manolo Bellatru's PhD. This is the center of the rift segment where beneath the lavage. This is where the fissure eruptions were. You contract those across. They're largely asysmic. So these dikes would propagate through the fissure eruption or the area where some of the dikes rose to the surface and continue southward at different rates. And Eleonora is going to explain why the curve, there's not a straight line, why they're fit with curve lines. She's going to explain more about that or not. Yeah, I know I did. He did talk about the pressure behind. Well, Eleonora, sorry, Eleonora did this work. It's based on a brilliant paper of hers. So, okay. I must have been inserting slides at that moment in time. The other point about the dike intrusions is that in this case, there were very large fault scarps created. And so there were hundreds of fault scarps that we mapped from the air. The problem was there were so many earthquakes. It's hard to know whether multiple earthquakes caused the fault slip we see at the surface. But this is a derrick here who's about two full meters. So if you use him as a scale, you can see that the displacement, so this is where the ground surface used to be. But this hasn't been sandblasted by the catabatic winds. And this is basalt. So this is a solid rock that ruptured. And there, then, is that it's well over two meters of displacement on a single fault. There are several other fault systems with even more. So up to five meters of slip, massive piles of rockfall. And if you think about the global catalog of surface ruptures for normal faults, I think prior to this episode, they'd been, if you exclude mid-ocean ridges, there'd been probably 17, 20 well-documented normal fault systems associated with a single earthquake and aftershock sequence. And yet we have hundreds in this case, and we just wish we could do a bit more. But what we do know is, in terms of scaling, that they're very short faults, so large displacement short fault systems that form above the dice. So short meaning, maybe 2,000. The mean is 2 to 2.5 kilometer, 2,000 to 2,500 meters in length. And they tip into monoclines at the tips of these fault systems. So interesting surface faulting patterns and new, but largely along pre-existing fractures. This is obviously a pre-existing fracture. There was displacement prior to this. Okay, so, yeah. And you can see the lava flows that they, these tips into lava flows at the northern end. These are the full-moment tensor solutions for the dice sequences. This is from a second paper, but from Amalotus thesis. I have diagrams, I'll just quickly show you the magnitude versus time of the rate. But this is a distance along the length of the rift from the two largest of the dice intrusions. This November dike propagated south where it stopped for about 12, I think it's 12 hours. And then restarted and moved southward. But its first path rose to near the surface. The second segment never reached higher than about six kilometers subsurface. So it remained a deep, much deeper dike than the first part of it, which is a bit confusing after this stall period. So you can see it moved faster at the beginning and there was a slower rate of propagation after the fact. These red are earthquakes with much lower frequency content. China's, and this was something that perplexed us for a while, but we now feel pretty confident about why we have low-frequency earthquakes during, we have some low-frequency earthquakes and some more tectonic earthquakes with a typical spectral content. How much of it is source and how much is path and how much has to do with rupture to the surface is something that a new student, Gabrielle Tepp, just teased out from the data set. Now, just keeping to time in this case, earthquakes, this is the earthquake sequence that just recently happened and I'm sure everybody here has talked to Elias now and who, I can't remember the names of the Cambridge students who've been working on this sequence. Wait, what's your name? Sorry. Hmm? Okay, Jane. Oh, yeah, okay. Wait, your last name's Greenway, right? No, what's your last? Oh, well, never mind, we'll talk later. Okay, anyway, this is from a recent page. Sorry, Freistine Siegmundsson is at all, hmm? It's okay. Oh, yeah, okay. Yeah, why not? I don't know how that happened. That's straight from their page. You can see I downloaded, no. This is a simple elastic model showing the distribution of opening along the length of the rift with seismicity superposed on top of it. So here's opening in meters and this is based on modeling then of the interferograms and with additional constraints imposed by the earthquake pattern that was interpreted in parallel with the analyses of the interferograms. So you can see that their patches are much more opening along the length of the rift and it's segmented and there's a, but the most important thing that you should notice in this case is that, oh, sorry, I should have said before that there are some diagrams that have depth but almost all of these earthquakes are occurring at the top of the dike. So in the upper two kilometers of the crust and very few of them are deeper than that. In Iceland, they're beneath the dike. So the stress concentrations and the brittle deformation are actually in the viscoelastic zone beneath the dike where you wouldn't expect to nuclear earthquakes. So rapid stressing is the most likely explanation for them but there's not actually a large moment release beneath the dike zone and there's virtually no earthquakes at the top. Yeah, there were surface ruptures. So I gave you one example from now far where almost all the seismic energy is up at the top of the dike and released during, in both cases, the energy is released during the propagation phase but brittle deformation below versus brittle deformation above, surface faulting with lots of seismic energy release, brittle surface faulting with very little. Almost all of the earthquakes in this sequence were strike slip whereas they, as I just showed you, they were normal earthquakes with a dilatational component in the Afar example. Much slower opening rates and a deeper source body are some of the contributions but it's not fully explained and requires considerable much further modeling. And in terms of seismic moment release versus time, this is, these are the Afar dikes and they're off the top of the page because there are so many. I'll just take one of these. This is the October 2008, there's the November segmented dike. You can see that there are low frequency earthquakes throughout the sequence, well, barely any during the second, deeper part and that's important. And then here, almost all of the energy occurs during the propagation. So almost all the earthquakes happen as you're moving the dike as Torsten explained that that's where the stress concentrations are along the edges of these blade-shaped dike systems and that the, you can see there are some probably aftershocks then to the surface faulting at the same latitude but this is as they propagate along. So almost all the energy is released during the propagation phase and you can see the magnitudes of the earthquakes then they're all scaled to magnitude on this diagram. Now I said something about low frequency earthquakes. I'll spend a little bit of time talking about this in just a second because this links back. One of the, how many of you have heard about low frequency earthquakes on volcanoes? How many of you study low frequency earthquakes on volcanoes? Oh, it's a couple. Okay, but you all are aware and then, okay, so what's the, what are the low frequency earthquakes caused by? It's not a rhetorical question. What have you been taught? What are you, what, if you know about them, what's the association you make? Fluid. Fluid. Fluid? Fluid. Fluid. Right, so it's faulting with fluids involved causing a slower, potentially slower slip, slower rupture of durations, source time functions that are causing them this frequency reduction along the, within the zone. Now here's a dike. We're not in the volcanic edifice. It's a slightly simpler problem in some ways in terms of the rupture. So I'm gonna take this apart. And in this case in the alfar, what we had, we had, this is an earthquake from roughly, within a kilometer, the same location received on the same station. So this earthquake, for this, this earthquake, then you'd say it must have a source effect. Well, there's also a shallow effect and surface rupture component of these. And so what, what, I'll give you the short version of this, that Gabrielle Tapp, who's now at the earthquake, Yeah. The magnitudes are comparable? I mean, let's show you, in the case, well, there's not a large difference. None of the earthquakes are very large, greatly in magnitude, in our magnitude, and I'm only showing you larger earthquakes within these sequences. I can't tell you exactly what these are. I can tell you after, but it's not a huge difference. That does not explain this, okay? So you're completely right. Frequency, the frequency of the earthquake is dependent on magnitude. So in order to have magnitude difference in the magnitude, scale will lead to a reduction in the dominant frequencies. These are, that is not the case, and I can show you an example after example. We carefully examined that effect and that's not at all what's happening here. Okay. So these are suggested ideas. I think that from what I understand, a lot of olcunology meetings in the past have had heated debates about what the cause of the low frequency earthquakes are and what very long period versus long period versus the different frequency contents and what they can possibly mean. And whether it's resonance within the chamber organ pipe modes or gas escape, and then also to ask very slow rupture velocities and for shallow earthquakes, that rupture in the surface with really slow velocity material, really loosely compacted material in the surface, they rupture very slowly and also have a low frequency content because of the slow rupture velocities in those cases. And if you're interested in those, read Chris Beans papers, there are several of them and they're really enlightening and they're both observations in numerical simulations of the rupture propagation. So in a way, so we looked at it is then that the ground motion or in turn the velocity, everyone knows from interest seismology that you have a source time function. You have a propagation or path effect because it has to go through that hot material with lots of fluid filled cracks to get to the station to be recorded. So how it gets to the station in many ways becomes really important in the Afar area. Kind of funny story. Got dropped by helicopter, didn't know what we had. One team went south and was supposed to walk until an hour, walk for an hour and a half as far as they got put a seismometer in the ground. The other team was supposed to walk in the opposite direction. My team couldn't find a good place to put the seismometer. So we went up a little fault scarf and put the seismometer there and we accidentally had one station on one side of the dike and another station on the opposite side of the dike intrusion when we figured out what had happened. And so we measured the largest ever attenuation values across the dike system by earthquakes from outside the system. So we have hugely attenuating material and either within these areas near the dike sounds and fracture and fluid filled particularly immediately after the intrusion of a dike. And then the station terms and those are things we can deal with but it's the trying to look at that part. So let me just show you, Gabrielle put this slide together for me. She made it really cool so that you can see all pieces together but I couldn't get it to work. What in a ways, so this is an earthquake and the spectral response of a low frequency earthquake, a hybrid earthquake and a high frequency earthquake within a particular area. And if you look at the spectra, you can say the spectra are different. But if I tell you what's the peak frequency? Hmm, what's the peak frequency of this one? What do you mean by peak? Which peak? How do you want to classify or characterize any of these earthquakes? One of the ways that you can make it so that a group from one institution can compare with another institution is to use a frequency index and just take the ratio of the high frequency component to the low frequency component after carefully assessing and finding examples to see whether you can classify or find subclasses or differences within your earthquake sequences. And I could point you to these two papers that Gabrielle's written or Michael West in a group at the Alaska Volcano Observatory had been using this for a while and I think other groups are adopting it. But it certainly enables comparison and reduces some of these debates that are based on the way that you treat your data. It's a uniform and clear way of treating data sets. As you can see then that we can classify these earthquake sequences. Okay, so the path effects, 3D velocity structure go around the low velocity zones along the path. Right passes will turn laterally too. We have scattering, attenuation, anisotropy and also, well, the VPVS, let me, that's for later. So these are normalized to the same scale but they're on the same station and different earthquakes from the same spots. So at the same areas, and these are from two different dice, the October Dike and the November Dike and just showing the differences in frequency content. Received at the same spot. And what Gabrielle was able to find out and so these are the different sequences characterized by the frequency index, okay? I'm just saying that we have the lowest frequency earthquakes during the dike propagation phase and they're from the largest magnitude earthquakes that are ruptured to the surface. So the low frequency earthquakes in this situation are in, sorry, and no bimodal distribution instead of continuum from the low frequency to the high frequency. So varying degrees of path effects then are changing the spectral content remarkably along the way, without this change. So the lowest frequency earthquakes rupture at the top or we assume they're rupture at the top of the dike, they're located at the same spot as the top of the dike and rupture to the surface. They may be slow slip, but I'll explain in a second. Sorry, I think that's, hopefully this is clear to you then where we are, but the most important piece here, these are the normalized earthquakes. So when they're this yellow color, they're a tectonic earthquake. When they're this color down here, when they have the frequency index, the negative frequency index, then they're very low frequency. And this part here, you can see these two, these are the two deepest dikes. And so the earthquake rupture depths are the deepest in the whole sequence and they have the highest frequency content. So we had the benefit of being able to compare multiple dike sequences with different characteristics on the same stations and be able to see these patterns. The other thing that we know from this area is that we must have lots of fluid build cracks as well. This is Derek Kear's work on a really beautiful paper looking at, he also inverts then for the fracture density. And these are from local earthquakes recorded on the station we temporarily hand carried out to the middle of the Rift. And when folks camped there for two weeks, recorded earthquakes. And he was able to measure the direction of anisotropy from shear wave splitting from little local earthquakes. And the amount of splitting is actually very large. So three to 6% anisotropy, they're all normal on the right hand side. They're normalized by the depth because you'll, so you have an earthquake. And if there's anisotropy then the two, you have a three component seismograph, you'll have a fast direction, a slow direction. You'll be able actually, sorry, looking at the shear waves then, they're passing through this media. You'll be able to see if there's a lag in a directional lag. And one of the things that we, he was able to see that there's actually large amounts of splitting and it's all Rift parallel. So these are fractures and they're superposed on the fault map as well. So there are some local variations that correlate very nicely with these fractures that we can see at the surface too. So a lot of this then must mean that there's gas and fluids along these paths as we already knew. And we can also get some idea using compliance techniques invert to see what the fracture density might be as well. So we're extracting more and more information from these well-planned experiments or from extract, taking detailed analyses of these earthquakes warms. I didn't mean to imply though that in all places they are, we have a continuum. This is near Gongo and it's, Hawaii is the same. If we take the frequency index of earthquakes, there's a recent paper by, I think it's problem at Tosa on, and another's, oh, I can't remember the authors now. I can tell you afterward, but they see a similar pattern on Hawaii, the low frequency earthquakes and the bimodal distribution of frequency index. And these low frequency earthquakes in near Gongo on this diagram, on one side are the SO2 gas emissions detected by satellite. Okay, so these are the earthquake, number of earthquakes on the right-hand side versus the SO2 emissions and the low frequency earthquakes happen when we have large amounts of SO2 gas escaping from this open conduit in near Gongo. So we can correlate these guys with probably a very distinct process. So trying to extract these, the summary here, dikonduced earthquakes with an enhanced low frequency content. Well, you can read this. We should see a broad continuum, but differences from place to place. So squeezing more information from studying the swarms and making comparisons across provides a lot of information. And again, the independent observations and collaboration with taking the information from other methods, the SO2 gas emissions, for example, talking to the gas folks, I've learned more, well, as much, I don't want to say too much, I learned as much as I do from talking in other seismologists. So it's very important to try the multidisciplinary approaches to volcano research. Now, the second part, I think that is relevant, particularly to what we've been talking about in Eleonora promise that I'd talk about lower crustal earthquakes. The lower crust, as you know, you've all been taught, right? The lower crust is granulites, probably maybe wet granulites and it's pretty weak, continental crust. In most places where we're studying the volcano and yet we can have earthquakes in the lower crust. And in some cases, I'll show you, when we're in extensional environments in particular, we can have some of the largest magnitude earthquakes in the area are in the lower crust. Okay, so are they, is it because we have a really strong lower crust? It's really cold, it's very mafic material and so it can stress us for a long period of time and so build to a large magnitude earthquake. Okay, that's one possibility. But how much of it has to do with rapid stressing from the release of fluids and the transport of fluids? And so I'll give you some examples. So I'm sure everybody's seen a Christmas tree diagram, right? Has anybody not seen a Christmas tree diagram? I don't see anything more. These are different curves for different compositions, different fluid contents. You know, it varies drastically by fluid content. So you can see in some cases in the lower crust here, you have virtually no strength depending on what the composition of the lower crust is. Okay. So if we pass a dike through the lower crust, so why am I worrying about the lower crust? Magma's rising up through the lower crust, right? We want to understand those pathways from the sore suns and the mantle up through to the surface. How can we extract that information? We're making progress is I think the answer, but we don't, or is where we stand, but I don't think we have answers. So one area, Julien Barique in France, she was in Brest and now working on shallow earthquakes elsewhere on hydraulic fracking, mapped out in this, this is a yield stress envelope for an area assuming that there's a very, that there's a very mafic lower crust in the area. Now I think our new observations, this was before we did this new experiment in the area, it's most likely not diabetes or wet diabetes in the lower crust, but this is a yield stress envelope that you predict to be able to explain the large number of lower crustal earthquakes, and these earthquakes are very well-determined and they match the patterns, so local earthquakes match the patterns of telecisems and their frequent telecysmic earthquakes in this area, it's the largest seismic energy release in all of East Africa, and most of the earthquakes then are occurring in depths of about 30 kilometers. And as Julien thought that it was the release then transfer of CO2 from the metasomatizing mantle rising up through and causing higher elevated poor pressures and that would then predict some interesting characteristics in the VPDS that we are testing. Now when we did our experiment in the area where I showed you where we have lots of magmatic centers just to the north, we also had loads and loads of deep earthquakes and there are two colors of diagrams because we relocated earthquakes using multiple methods, 3D velocity models, the entire data set, double difference and they all show no systematic variations in the depth distribution. We also know that there's an area in here where if we elevate temperature in the lower crust we can have some changes in the in the pyroxene that may cause an increase in velocity as well. These bars are just showing, oh those bars moved. The bars, that bar is supposed to be right there. That's the bar showing where the telecisums, the telecisums were in here. This is an area without magnetism in Tanganica where Odla de Isillera is working with us. And I'll come back to that in a second, but even deeper earthquakes and bimodal distributions as well. So I think as we keep going, I know you're all antsy to go have your coffee, but I'm trying to keep to time. When we make cross sections of what we have from our earthquakes, we have border fault systems and in this case, so I'm in a rift zone, I have a volcano just out of the plane of this diagram so I'm skipping the volcano for the time being because I want to try and understand the system and the tectonic setting before I interpret the volcanoes. The volcanoes sitting, these are base of crust from receiver functions and I have numbers for VPVS. VPVS when it's higher than about one point, well in this area, the mean or the background is about 1.7 and so anything elevated then may indicate mafic material in the lower crust and the presence of fluids that reduces S wave velocities more than the P waves. Okay, so you have high P waves but reduce S wave velocities and Poisson's ratio is directly proportional to the VPVS squared. Okay, but in this case, really clever folks. Toby Fisher and two students went out, James Muirhead and Hyun-Woo Lee, went out and made gas measurements all along the area that we were studying, let me go back. Do I have a diagram now? Okay, anyway, they went all up along away from the volcanoes just to see what kind of gas emissions they were measuring and measured some of the largest ever soil gas measurements of CO2 even more than they have been determined and can't be flagged directly above a volcano coming up these fault systems. We also have lower crustal earthquakes and we see that many of those earthquakes are directly at the tip of where we project the border faults at depth and we're suspecting then that we have gas release from magma intrusions. We know that it's magmatic CO2 so we're catching the magma intrusions from the gas emissions and we know that it's happening and then we have lower crustal earthquakes in the same places so we're associating the two of them and then using VPVS and crustal velocity variations in these areas as well, these are where the same earthquakes are and we see this low velocity zone over here, it's a magma chamber, this is associated with the basin. These are VS, you can't just directly translate this because you have compositional variations as well. I'd be happy to talk about those. So these are the gas measurements of the areas, they went along the fault systems and found very large fluxes in these areas, these are similar to what's been measured in comfy flaring here and their mental source fluids, the enhancement from the metasomatism and these numbers then, if we extrapolate them over the Eastern Rift or 11% of the global budget, I'm sure that this area is in a high early stage in a flush so this is probably an overestimate but you know, anyway, microsisems and saline fluids maintain permeable pathways for these fluids along the fault systems as well and we have active degassing that's enhancing the process. And just to put it in context, the biggest polluters of CO2, Niergongo, Mount Etna, Nakajima here and oh look, number four is actually a basin away from the volcano so large volumes of magma being intruded during this early stage rift zone. I don't think I have too much time, this is from the receiver function work and just saying that we are seeing high VPBS and the lower crust in these areas, reduced velocities in the upper mantle and actually getting reflections from within the mantle system. This is Christelle T. Barilla and Matthew Plassman that's just been submitted. I don't have a lot of time except to say that know your stresses in this case that I was talking about in Northern Tanzania. We have a, this is the predicted plate opening vector from geodetic data, this is the earthquake, this is a summation or cost of summation of all the mechanisms that we have in this sector at sub East West and there's a, see there's a large bound on the plate opening. There's few geodetic continuous GPS in East Africa. So North and South we have roughly parallel to the plate opening and we have a 90 degree rotation within the area where we have the sill complex, the magnetic systems and you can see in these mechanisms and through here and that tele-sizons from the dike intrusion show us exactly the same thing. We're working on the hypothesis that we have magma intrusion behind the fault systems as predicted by McAferry at all and so coming up behind it's driven by the basin system and the pressure differential and the body force is induced by the surface topographic loading and the intrusive bodies is contributing to the stress rotation in this area. It's a working hypothesis and requires some complex modeling, okay. So I just said that I don't have too much time just to say that the presence of fluids like CO2 or gases along fault zones, this is an example from shallow levels. You can see that it reduces the plates, the strength significantly. Finally then, I just wanted to say that we're suspecting that when we go to areas without magnetism we see lower crustal earthquakes and the similar types of patterns, high VPVS ratios and the lower crust that in fact the amagnetic Western Rift maybe in an instance where the magma intrusion is just starting it's at lower crustal levels and it may be a way to detect those. I think I just want, yeah. So I was just gonna say some things about VPVS. Let me leave then with this. Nobody else, nobody has said anything about MT, there's no one else here. I just want to emphasize that there's also a very important point in trying to take multiple methods and magnetotellurics are well, very well suited to detect fluids and the nature of those fluids whether it's aqueous or magmatic the conductivity variations are very strong within these areas. This is the Dabahu area with earthquakes here, the high conductivity zones, some of the highest measurements that predict more than 10% melt within these magma chambers and a large volume within the area. And these are comparisons then with receiver functions and waveforms providing some comparisons of where we would place the base of the crust versus the magma bodies and helping us to understand just as folks have done in the Andes with large volume magma chambers and trying to make comparisons between the multiple methods. So I'm just showing that the MT, blow up the MT seismic gravity and then inside derived models this is where they place the body as well. We need to place these together so that we can get past highly non-unique models that can explain the inside of the surface deformation patterns alone to try to get at what's down there and how it's moving and changing over time. Thank you. Thank you.