 All right, so now we're going to move on to a panel structure where we have two speakers who are going to speak for about 15 minutes on scientific opportunities and challenges in studying precursory phenomena. After each talk, there's gonna be five minutes for questions about that individual talk and then 10 minutes for discussion after that. And so we're going to start with Sergio Ruiz from the University of Chile. Sergio, are you there? Yes, I am here. Good morning. Oh, terrific. Thank you. So Sergio is going to be talking about some precursory slow slip signatures in Chilean subduction. So Sergio, I'll hand it over to you. Thank you so much for joining us. Thank you for this invitation to show some of the precursors that we are observing, the Chilean subduction. Here is my presentation. See you, my presentation. Yes, we can say it. My presentation is mostly about the Iquique arc, the event that was previously introduced by Emily. And also, one of the arcades that happened in central Chile was a smaller than Iquique, was a magnitude 6.9. But also, this event has some clear precursor. In general, in Chile, the mostly of the large earthquake that happened in the megatrust, in the contact between NASA and South America, are preceded by torture. For example, when we read the historical events, we read it wrong. For me, it's only an anecdote. The event that happened in 1730, this large event that happened in central Chile and event magnitude 9. And in this period, this event was very destructive, but fortunately, almost any person died. Why? Because when the earthquake occurred, mostly of the people were sleeping in the yard. Because previous to this large earthquake in central Chile were many four-chops. And at this time, the seismic culture for the Chilean people was that when there is swarms of the large events, could be happened a bigger event. And in general, for us, mostly of the events that happened in the megatrust are four-chops. But one of the problems that I mentioned is in Chile, the most destructive events that happened in Saitanascap Lake, around 100 kilometers. And apparently, this event, we don't observe some four-chops. OK, we see other on the last years, in Chile, we have recorded several events. But our network is working relatively well, only in the last few years. And for example, during the Mauler, an event, 8.8, also was recorded some more four-chops. But at this time, our network is not well. And we are not sure this big earthquake was preceded by some clear swarm of events. There is only one idea that some events preceded this large earthquake. Maquique was recorded, that was introduced by Emily, and I speak after. Other two were like Iyapel, that was an event, magnitude 8.3. In the zone that happened, that is in northern central Chile, there is not many stations closed. Then there is not clear four-chops because of this event. Also the same problem we have for Chilean events. And Valparaiso was a smaller event. But fortunately, we have some stations just in front of the epicenter of this earthquake. And recently happened another interplay to the Maillard death earthquake that was some destruction in the cities that are closed. But we don't observe for truth, probably because this kind of event is different that happened in the mega-flash. OK, Iquique happened in the northern Chile. It was an, we expect this event because the largest that happened more than 100 years ago was an event, probably magnitude 9 with a large tsunami. For many times there is not big earthquake in all this zone that is around 500 kilometers. But like Emily introduced, this earthquake was clear four-chops two weeks before. But I can show this observation. I don't have a physical interpretation of what, but I think it is very certain that the Iquique earthquake occurred here. And if we observe these dots are the epicenter of event magnitude larger than 6, we can see that mostly of the interplay event occur in the last 10 years or after 2005. And what happened in 2005? 2005 we have an added interplay to the Maillard death earthquake that happened in front Iquique. It was Tarapacá event. This is the epicenter. It was a magnitude 618 that happened 100 kilometer depth. And after this interplay of Maillard death earthquake, if we observe time series that we have in Iquique city and in this point in front of after Iquique earthquake, if we can see the time series, we can see that before of Tarapacá, that is a blue line, for example, for the east-west movement of this GPS antenna, we have some trend. Here, unfortunately, there is a gap in the data. But after of 2005, in this zone in Iquique, the movement of the GPS changed his velocity. There is a change in the trend that we think that happened because occurred this event in 2005. That is here, the interplay of Maillard death. This introduced some change in the coupling of the megatrass or some physical things that at the moment, I think we don't understand well that happened. But this is the observation. Also, this increased the event magnitude 6 at this period of time. Also, at this moment, we can start in 2005. We record many, many swarms that happened in the boundaries of the root of Iquique earthquake. This swarm started detected around 2005. Some people, like Cato, look in these swarms by the repeating events. There is many repeating events during several years before Iquique earthquake. And also, we observed that this start around 2007, 2005. But here, the problem is at this period, also start the instrumentation. Then here, I'm not sure that the observation of the swarms is really a start after the interplay intermediate death. Anyway, we have this observation also and so we study GPS time series. And she and her groups, they propose that a slow-bath event start at least eight months before. And like Emily chose, we propose that also an slow-bath event increase is moving two weeks before the Iquique. Then, this is for Iquique. There is clear foreshocks two weeks before, probably a slow-bath event also occurred two weeks before. And maybe his start previously, for example, and so can propose that this start, this move and start eight months before, and probably also happened after a start-up Iquique was in 2005. The other observation that I can show you is an event for us for the slow-bath never maybe not so big. It's an event magnitude 619 that happened in Central Chile. It's Valparaiso. This is Santiago. This is our capital. And there is many seismicity in this zone. And in this case, we have here in this picture, maybe it's not very good, but we have these inverted triangles are the statuums, broadband and some GPS statuums. And then, for example, here we have also a GPS statuum. All these red dots are the foreshocks and the aftershocks of this airport. And in this case, only two days before of the main truck that was in magnitude 6.9, occur an event magnitude 8.4. And during two days, we have many events. Magnitude target around five. Then we have and swarms of the event magnitude five. That if we look in this picture, this is the peak of time. Here happens the main truck. These are two days of seismicity. And the dots are the events. But the blue score are the repeating events. Then we look for the repeating. And we observed that two days before, many of these foreshocks also has the criteria of the repeaters event. But also, like I mentioned, you ask. Also, we look of this event, where they happen. Then we study his tensor moment. And we think that mostly happens in the interface between NASCAR plate and South America. And also, here in front of all this seismicity, we have GPS station. And if we look at the time series of this GPS station, this is our days. This is from 1st of January. And here are the two days before, we observe that the GPS station starts to move here toward the ecosystem. This is the main truck. And this also, we think that maybe this was a slope slip that happened during these two days before of the main truck. Here is a sum of this sum. But here is the movement that is around 10 millimeters. Here we have, this is the previous movement. The movement that happened in the GPS station two days before. And if we did an inversion for interpret this nucleation phase, these two days, we have that the slip that reproduced the movement in the GPS station is equivalent and seismic moment of around 6.5, 6.6. And the largest four truck that we recorded during this two days was only a 5.9. Then mostly of the movement that we need to do an interpretation of this movement should be a seismic. Well, after we did the slip inversion for the main truck. And this is dynamic inversion, but not many. Okay, then in this summary picture, I told you the events that was repeating events that happened during the four truck sequence. Then these are events that happened the two days before of the main truck. The rupture of the main truck is this gray zone. And this continuous line could be the seismic moment that we obtained considering the movement of the GPS station during these two days previous to the main truck of the Valparaiso Earth. And then and summary and other ideas of these circuits. Well, we have two events at least that we think that has clear precursors. They are has clear a four trucks and some swarms that and these swarms that also are repeated events. And probably in both case, Iquique and Valparaiso, this movement was associated to an insulover slip. Also for other earthquakes dimension, for example, Maul air for others, we don't have very clear four trucks probably because our magnitude completeness is not enough at this moment. It's in Chile at this moment, the completeness magnitude is around three. Some years ago, we are around four. Then like Emily said, I think that at this moment we need to reinterpret the data and looking for maybe four four trucks that will be happened before others earthquake but with the lower magnitude looking using the learning matching techniques. And so far, I think that we are not study well at least for the Chilean earthquake, the intrapainter may have been events. These are more the most destructive in Chile. And we are not observe and clear a precursor phase for these events. The most clear like I meant to is always for the events that happened in the mega stress. And the other problem in Chile for the precursor is in Chile, we have many swarms but probably now we are in the identified also slow slip events. But many of these swarms after they don't continuous with a large earthquake. Okay, thank you. If you have any question, please. Thank you very much, Sergio. That was perfectly timed. My alarm was about to go off at 15 minutes. Thank you. Brilliant. Do you have any questions? We have five minutes. Okay, I'll ask a question. I mean, Sergio, this is really a remarkable set of observations that you've just summarized prior to multiple earthquakes in Chile. I mean, there's no denying the, you know, I don't know how to put it but the potential here in terms of seeing processes that are clearly going on before large magnitude earthquakes. I wonder what the reaction is in Chile. As you heard Emily talking about observational capabilities to try and understand these observations better. I wonder within Chile, what is the discussion like in terms of putting out instrumentation to be able to make more observations like this potentially in the future? Yes, I think that obviously I share the idea that deploy, that we need a stage on the below of the onshore in the ocean here but also at this moment in Chile we have a good network in many places. In many places, for example, the Javelin Airport was an event in 2015. We don't have a seismological station close the rupture zone. Then in this zone, the magnitude completeness, for example, I think it's a magnitude around 3.5 and with some more stations close to this rupture zone, I think that we can take a lower magnitude, the completeness magnitude and then maybe there is some fortune magnitude too or any sort of thing. We are at this moment in Chile, we have a good network but the priority is to do a more uniform in the space or land stations. Okay, thank you. We have time for maybe one quick more. One other question. Can I call out to Anne Sopé, who's actually here, did I coincide it? Yeah, did you want to add anything? Well, yeah, Sergio, thank you very much for your talking with you. I was, maybe we can find something that has been, maybe move a little closer here just so he can hear you really well. Thank you very much, Sergio, for your review. So we can probably point something that has been seen in Chile, it's that interaction between earthquakes. So you mentioned the interaction between Trapacas-La Pula earthquake and Iquique-Megatras earthquake but you've shown also that there are large-scale transient deformation after Maula earthquake that cannot be successfully modeled by post seismic viscoelastic relaxation. And so this was compatible with an increase in geodetic coupling before Iliapal earthquake. So this is something that is very intriguing and that is something that we do not really understand. And so other studies have shown that there are large-scale interaction between seismicity, so namely an increase in deep seismicity before Iliapal earthquake that started exactly at the time of Iquique earthquake. So this is something that I think we should probably tackle as a community. The fact that we have several earthquakes that are occurring on one given seduction zones and so I think that probably it's not only focusing on our short-term precursor but probably over a wider time scale try to understand what are the relationships and the measurements that we can do to look at evolution of coupling, long-term evolution of coupling and long-term evolution of seismicity. So what do you think about this, Sergio? Yes, no, thank you Anne for your comment. Yes, we also observed like Anne said after Maula earthquake was a mega-thrust event that's the trend velocity in the GPS change at very long distance of the rupture zone. For example, Iliapal earthquake that I mentioned or maybe I have not mapped. Yes, Maula happened here and we observed that the GPS change his velocity trend in a zone that happened here. These are 400 kilometers and the velocity trends change exactly in the zone that after occurred the Iliapal earthquake. Then also, I agree, we think that Maula earthquake produced probably and this relaxation at very far distance and probably Iliapal that was an event so they considered probably this distance is related with the Maula earthquake. We also after the proposed the same for other event that happened in the southern Chile was a Chilo event from 1927 to 1926 because also Maula earthquake changed the velocity or the coupling in the southern Chile. Yes, I think that it's also I think that it's a very good it's an interesting observation that now we need to understand and we model well to really understand that happened in the Sebastian zone where the mega-thrust earthquake occurs. All right, thank you so much Sergio and we're going to move on now. There's another question but it can I think wait until our more general panel discussion. So Paul Segal, thank you so much for being here also. Paul is going to talk about long-term transient deformation prior to the 2011 magnitude 9 to Hoku earthquake. Okay, so first off, I don't think this is a precursor when I'm about to talk about at least in the traditional sense. Presumably I'm here because someone thought so. I'll explain why, but I've spent, you know, now quite a few years trying to even convince myself that this is what I'm going to show you that it's actually real. It's the ground doing something, earth doing something, not an artifact. So actually this comes at good time because it's been on my mind that this was a good time to go back and revisit some of the assumptions. So this is work in the Hoku region that Emily already spoke about what this needs to be working. No pointer. We're going to go back. Anyway, that's okay. So this is in coming from the GeoNet data, the GPS continuous network in Japan, Northern Japan looking at data prior to the earthquake. And I want to say that that we didn't start out to study previous redeformation. We sort of dragged into this by the data just to try to follow the data since I discovered this. So I just want to walk you through the evidence and what we've done more recently to try to go back and check. So I was going to give a talk at the AGU meeting the year of the earthquake on something about inverse problems, null spaces, things like that. But I needed the time series and when I plotted the time series on my screen I just looked at this and went, wait a minute, this is not straight. There's curvature in these time series and they were holding up a piece of paper against my computer screen so what's going on here? So I immediately wrote to my colleagues in Japan and they said, well, we kind of know about this. And the first thing you want to do is to say, is this present in all of the stations and it's not. You go to the stable parts of Japan, the most stable parts of Japan in Southwest Japan, they show a pretty linear trend. The next question is, is this a processing artifact? And there's reasons for thinking that it could be because of the way the Geographical Survey Institute of Japan was doing their solutions. They were doing solutions in subnetworks and then combining those subnetworks. And you can imagine that if there were stations dropping out in the backbone network that they used to tie them together that you could get artifacts. I'm really worried about that quite a bit and I talked to colleagues at JPL who were doing independent processing using what's called precise point positioning which is, you don't need these subnetworks and we were able to see their results but not publish them and we found that we analyzed those data and not publish them and we found that yes, we can see the same kind of thing. So we knew it wasn't completely an artifact of the processing, although I'll show you that there's probably some component of that in Japan. And then there are transits, we divided it into six and a half and larger earthquakes in this area during this time period which starts in 1996 when we continued this network got more or less complete and you can see some of these transients here so you have to worry about these transits and take them out. So some of this is due to the known transient behavior but not all of it. And I will say I'm going to come back to this and talk about this area in general and that is that the post seismic transients for these modest earthquakes are very large. And whether that's something that's geographic that has to do, there's something about this area that's different or whether it's a temporal change that didn't look like this 200 years ago, I don't know. Anyway, so this is what got us started and this was the part of the thesis like was the thesis work of Andreas Mokomotis and so Andreas did is to take these data. We did a reference frame correction the best we could. It was a state-of-the-art way of correcting for the reference frame in the junior data like you know in the details and then he did two things. So normally we fit the time series, these are positional time series but the linear term he just added a quadratic term because you can sort of see there's some curvature in these and importantly that curvature starts before the first earthquake. So first one was in 2003 and we went through a whole procedure which is pretty complicated to try to remove these post seismic transients. You see them here and what you left with this is the residual. It actually fits a quadratic pretty well and the interesting thing is that these are statistically significant curvature. We don't believe that there's a constant acceleration that was going on for hundreds of years. That makes no sense but during the time period from 1906 on up to 2010 it really fits this kind of constant acceleration model and so we ended up just plotting acceleration vectors. Normally you see velocity vectors. This is the second time derivative of the position time series and they're spatially coherent. So it's not just one station or two stations but a lot of stations that are exhibiting this and in the second is Trenchwood which is consistent with an increase in slip time on the plate interface. Up in the San Rico area here they're the opposite sewing which is a decrease in slip rate and some of these, as I mentioned some of these explorations are statistically significant at high level. So we think we know what's going on in the San Rico area. There was an earthquake here in 1994 called the San Rico Okay Earthquake and that caused the earthquake to be transferred from that earthquake. In fact, it's a famous nature paper by Heckie. This is the first time we had continuous GPS data showing a post-sizes earthquake transient but what was remarkable about this is that the afterslip presumably this is mostly the afterslip in the year after the earthquake was comparable to the displacement during the post-sizes phase itself. So that's what I'm saying, there's a lot bit this to a logarithmic function, we could just take his fit and extend it out another 15 years. Actually, I think the reason would be, well, it's kind of amazing. So we're pretty confident that that's a post-sizernitransient. We actually asked the question, can we explain this data as post-sizernitransient? So we went through all the large earthquakes that occurred in Japan in the last 100 years, used viscoelastic models to see if we could come up with some way of making this signal. And there's only one earthquake that even has the right sign, and that's the Niigata earthquake over here. That's 1966. And so the predicted amplitude is much too small. So we eventually concluded that this residual acceleration here is probably the fact that we didn't fully correct out for Sunriko, but this part here is unexplained. Now I'm going to show you an inversion that I'll, first of all, want to show you. So this is a traditional velocity vector, so in a North American fixed reference frame you can see we've got convergence towards, away from the trend. So this is compression, and this is opposite and sign. So whether you believe the inversions are not, I think this is relevant. If you just take the vertical gradient in velocity in 96, at the start of the time series, whether 2011, just before the earthquake, there's a 30% difference in the gradient velocity. That's a 30% difference in spring rate. So that's significant. If you ascribe the acceleration to accelerating slip on the plate interface, you can do an inversion. And the way Andreas did this was to do a so-called minimum norm inversion. So the idea was we try to put the least acceleration possible on the plate interface. This is slip acceleration to explain the data. Now I want to say that there's nothing necessarily that causes the signal to be on the plate interface. There is a weak correlation, which if I go back, there's a weak correlation between velocity and acceleration. So that sort of suggests they're coming from the same effect. But if you assert that it's due to acceleration on the plate interface, you're left with this sort of pattern here, deceleration in the north, we think we understand that this is a period of acceleration here, which has been in units of millimeters here squared. I don't believe this is a precursor in the traditional sense, and it doesn't look anything like a liquidation phase. Spread over this huge area, there's no evidence that it localized towards the type of center, and it's going on at this constant rate. It's not accelerating as you go. It's a constant acceleration, but it's not like models we have for how things behave on frictional faults where things just ramp up as you go into the earthquake. So there's something going on, we think, but I don't think it's a precursor in the traditional sense. It is true that if this is the right interpretation, the faster slip here would increase the stress where the horizontal line was. So it may have incidentally accelerated the occurrence of the event, but I don't think it's a nucleation phase. So we're sort of stuck with that, and then the question we find some independent evidence was actually, this is the home of repeating earthquakes here at Berkeley, and Roland had been working with Achila-San, who has had a network, sort of a catalog of repeating earthquakes. Roland suggested maybe we would look at that. And so the idea is that these repeating earthquakes are supposed to be failure of a sparity over time, and so if the slip rate is accelerating, then the recurrence time between these earthquakes should get shorter. If it's decelerating, then the recurrence time should get longer. So Andreas took the catalog of Achila, he went through and had a lot of selection criteria. We known it down by more than a factor of 10 to get a certain number of events that had criteria that stable magnitudes, things like that, above the threshold, well above the threshold. And then he applied something called the Mankendle test, which looks for monotonic changes. In this case, we're looking for monotonic changes in the recurrence interval. So here's a case where the recurrence intervals are getting longer, here's a case where the recurrence intervals are getting shorter. So he just took all these events that met the threshold and then plotted them in space, and this is a very remarkable result. All the ones that get the criterion that had decreasing recurrence intervals were in this area where GPS slip is accelerating, and all the ones that had increasing recurrence intervals were in the area where GPS 7 is slowing down, and that's I think a pretty compelling result because they're completely independent data sets. Now you can also take that data and try to estimate slip and then slip velocity and slip acceleration of those repeating events. I will show you that because it's model dependent, but he did that, and then you could do joint inversions of the repeating earthquakes and the GPS statement that was shown here on the right. But that is model dependent. So in the meantime, we've been sort of thinking about, well, what could possibly explain this data? And due to time limitations, I can't really go into the details, but we have this idea that it's possible that so-called seismic disparities are actually shrinking mechanically over time. That is, if dynamic ruptures expand into nominally stable parts of the fall, then the inner seismically, you would expect them to creep, reducing the area of the locked asperity. Kai Johnson's been working on this, and if you assert that that's the correct interpretation, this is showing the velocities and the locked zones in dark here in 1998 at the velocity field. This is a different velocity field because of acceleration. You could see these asperities would have to shrink a lot. So much so that it alarms me, and I think, you know, we make this fit the data, but do we really believe that that's true? I was just in Japan in March and ended up visiting my colleague, Takeshi Sukiya at Nagoya University. Turns out, he had a master's student who went back and looked at this data from scratch all the way from the beginning completely independently, and he did these PPP solutions, precise point positioning solutions with modern orbits, modern clocks, earth orientation parameters, and so forth, and he did find a discrepancy with the GSI F3 solution. It's not huge, but let's see. The F3, these are the ones that we were using are shown in red, is one modern solution shown in blue, and so they're largely similar, but you can see there are differences. If you turn that into accelerations, you can see that same pattern is here. It's the acceleration of this line, we're here, it's pinch, we're there, so it's kind of reassuring it gets the same pattern, the amplitude is reduced, and that's nice. Also, his solutions are accurate enough that he can get acceleration in the vertical component. Ours were too noisy to get acceleration in the vertical. So I think this is comforting in that it looks like completely independent analysis. And by the way, he had to go through a separate way of removing the post-sides of expecting all the six and a half, some sevens, although it's largely the same as what we've done. So a completely independent analysis is showing something that's very similar, although the amplitude is reduced. And that's good, because maybe then the shrinking disparity ID won't be so extreme that it causes me discomfort. I've been working with Camilla or Captainia, and we have a better understanding. I think I'm a better mechanical understanding of how these repeating earthquakes work. And I asked Camilla if she would take a re-look at the repeating earthquakes in the China's catalog, and she's looking at more events, and she realized that you don't have to just restrict to be as restrictive as Andreas was. This is all based on the fact that recurrence interval scales with the moment of the one-sixth power instead of the standard length of power, as well as the plane velocity. But if you want to look at acceleration, you can just take pairs of earthquakes and get the relative acceleration scaled by whatever the background velocity is, creep rate. So in the way she did it, she took the sort of plane velocity of 80 millimeters per year. But you could see her solutions largely looked like the Andreas. So if you look at the repeating earthquakes, it gets us more or less to the same spot. And I think this is the end. I just want to say these other relevant observations. Emily talked about the pressure gauge data. I just emphasize again that there's just a lot of after-slip for these earthquakes. It looks more like the creeping zone of the sitting grass than it does like most other faults that I'm aware of. And the other thing is this magnitude 7.34 sharp, this is something, jump of wire, and the shades of merozaki and I look at. There's just an enormous amount of full-size deformation in the time between the four sharp and the main sharp. So there are just things that are unusual about this area. And again, whether this is something special geographically about the materials and the faults on there or whether this is some characteristic behavior that changed, I don't believe it. Thank you. Thank you very much, Paul. So we have a few minutes for questions for Paul. And then we'll go to 10 minutes of panel questions. Paul, I'm curious that that we analysis would introduce such a diminishing of a long term feature. Can you then expect to see that also in areas where you didn't see those long term rate changes? Why would it be new 2DS analysis produced with angle patterns in the future? Something I was going to say, it's the same signal that we were saying, but it's diminished, right? So if you look at his, he did a pretty careful comparison of the F3 solutions with his solutions. And there are systematic differences. And some of that shows up looking like the curvature. And I think what probably was going on was that some networks are actually moving in respect to you. They weren't just stable as there's genocide going on. But he did do something where he took an independent network in Europe and did the same analysis and showed that this difference didn't show up. So there's something about the way the Japanese were analyzed in Japan. Actually, we've been also ready to put this data on. So we have been doing like a concrete re-analysis of the GPS data in Japan using a double difference approach. So it's another independent analysis. And we also find this exploration, which is also diminished, I think. That it's not straight or want to match yourself into a good reference frame in Japan. Because everything in Japan is moving. So when you come to the genetic point of view, you need to establish a good reference frame. And this is not completely straight forward. So I think it would be worth actually to compare the different types of issues from the digital analyzers. And also, depending on the way you correct for the co-and-post-sites information on individual canceres, then you will find an increased or decreased amount of lung cancer. And this is a really modern definition. So if you use a lung that is not too complicated, then you can find it. If you overfit the data and everything in the post-sites material, then you don't find it anymore. Yeah, two good points. I'm sorry, I should have mentioned Tan and her student lung. We're just down at Stanford who realized that they're actually doing a re-analysis as well. So now we have three independent analyses of this. And the goal is to kind of very carefully look at all of them. I think with the global orbits that are determined using stations outside of Japan, and I can't verify for sure that that's what we use, but I think that's true, then the reference frame should be okay. And you're right. You have to be careful how you remove the post-sites with transients. But you can see the curvature before the first earthquake. That's the one thing that gives me confidence that this does not do just somehow how we're taking out the transients from the 90s to the 60s. If we didn't see that before the first event, I would worry much more about it. But we were also very careful, Ann is correct, that if you throw in enough parameters for the post-sites material, then you could take out a lot more. We used a physics-based model to tell us which stations should be affected by the post-sites with transients, but a little bit too much detail here. Yeah. Assuming this is correct, could you go one step further and break up your time series and see whether there's any spatial trend? We looked for it. You know, we looked to see did it propagate, was there any evidence it's starting in the south and propagating in the north or start deep and propagate? We just can't convince ourselves we could see any propagation. One thing that Takeshi and I are proposing to do, so, you know, the assumption is that the end of the time series is monopoly. But since it's a constant rate of curvature, you don't know that. So one thing we're planning to do is to go back and look at all the terrestrial classical there's a century of data and see does that data look more like the early part of the GPS time series or more like the later part of the GPS time series or neither. So we'd like to establish, you know, which is the, which is what's normal? There's no, it's accelerating into the earthquake, but we can say that. All right. Thank you very much. We have now maybe five minutes for questions for either speaker and general discussion, but I want to start Matt had a question that I didn't have time for after Sergio's talk. So Matt, do you want to go first with your question? Is he there? Sorry, I just got unmuted, I think. Can you guys hear me? Yep. All right, so my question was for Sergio, so I don't know if he needs to be unmuted too, but my question is really just thinking about one of the motivations for having this meeting, which is sort of thinking about the public response and emergency management. And I guess I'm curious to know from Sergio, because I remember from the 2014 event reading media reports about what the public's reaction was to the series of foreshocks seismic gap and being very concerned. I guess I don't know what happened before the Valparaiso earthquake, but I think in general I'm just curious with these examples under your belt, what do you think about in terms of issuing public warnings or statements about when we start to see a swarm, do we issue a statement or not, or how do we go about making that decision? Yes, I am here. Yes, for 2014 was, I don't know, it's expected that a large magnitude earthquake occur, but even we was to deploy a more seismological station before of the main shock, but for the people, for the management, we never did that the earthquake could be occur after the swarms, because in Chile we have many swarms, we have very regular swarms and when this happened the people is very worried because a large magnitude earthquake occur, then in general the Chilean seismologist that when happened the swarms, we try to explain to the persons that it's not necessary an earthquake can occur because there is swarms in Chile. There is not a public policy in Chile about this that happened when the swarms occur, only we try to to do the depopulation in a wide situation. Are there other questions for either speaker? I wonder what your thoughts are on the ongoing sea floor deployments that are happening right now off the coast of Chile and how they might feature in future efforts. There's a couple of groups working on sea floor instrumentation in Chile right now, and I wonder sort of moving forward which parts of the problem you think those will help with? I mean they're not continuous from what I understand, so you can sort of guess what some of the answers are, but I wonder are they going to help? Are they going to help with completeness, even if they're not continuous, are they going to help with the monitoring? Yes, at this moment there is some onshore deployment in the northern Chile, but the German people and also the group the Antejo and San Diego University did an deployment in on other zones of Chile, but at this moment we don't have many results about these deployments. I think that yes of course with this deployment we can know better which the marine structure of Chilean subduction that we have. At this moment we don't know very well and also what has really happened with the structure that controls the seismicity. Because for example I don't explain, but before the TKR with the largest Portugal was in magnitude 6.7 and this Portugal didn't happen in the interplay zone. This happened is a cruised 11 that happened in the South America plate, and then this structure we don't understand very well at this moment because we all our states are in land and obviously with onshore deployment we can improve all these kind of the structures that are controlled the seismicity. Thank you. Thank you. Can I ask you a clarification? Is this also relevant to this? Sergio, a couple of years ago there was a discussion about a cabled observatory. Is that discussion still going on or is that is that not happening? I think that this discussion is open now. Actually it's Sergio Varian, the director of the National Seismological Center that is moving this idea, but I'm not sure that we can this project prosper because it's for the Chilean politics is a little expensive. But actually I don't know in what a step is this project, but obviously it's not clear that in the in the next few years we have an onshore instrumentation. Thank you. Cindy, last question. I'd like to go back and comment and a couple or just try to ask Paul and Emily to elaborate a little bit more and somewhat implicit in your presentations were the need for additional physics-based modeling of behavior not necessarily sorry a new potential for additional components of the fault slip behavior with the fluid driving forces that Emily was talking about the experimental work by the French group I remember the authors and then Paul you were also talking about some additional physics-based modeling and that hasn't we have a lot of instrumentation but could you comment about the way that we're approaching and maybe the potential need for additional theory studies as well? Both Emily and Paul. Well from my point of view it was if we believe this is real then what the heck is going on and so we were racking our brains trying to see if there was a plausible physical model around the time developments in modeling had and sort of understanding that was really originally motivated by laboratory experiments showed that if you have strong dynamic weakening during earthquakes you can rupture into areas that are nominally sort of steady state velocity strengthening in the Great State Comfort. So you have to have weakening friction where earthquakes nucleate but if you have strong dynamic weakening you can have earthquakes where that that behavior at snow slip speeds is very different and there was a talk by Nadia's group, Nadia La Pusta's group at AGU that Andreas and I were sitting in and they talked about having earthquakes rupture down into a nominally stable zone in the San Andreas and we both looked at each other and thought hmm I wonder if that could explain what we see so we've been sort of pushing that and trying to see if that was viable and we can make models as I showed you we can make models that take the data but they require rather at least if we believe the old accelerations rather extreme rates of shrinking of disparities to the point that I just strange my credibility so now that we have two new independent sets of analyses we go back and want to relook at that and sort of see what get a better estimate of the range of uncertainties but it's really driven by the data this isn't this isn't something that we predicted and went out to try to find it was the data that just caused us to have to go down this path I think I'd like to add a couple things to that you can listen to this sure um so yeah this is an observational problem obviously precursors is an observational problem the theoretical framework we have to work with as Paul sort of implicitly said is this rated state friction framework and it's kind of problematic that we have just this one framework that we're working in um there are you know rated state in the form that it is commonly applied actually does not match all node laboratory data um and it um moreover a lot of what we end up appealing to to explain these uh phenomenon is for instance Paul was appealing to heterogeneity and so heterogeneity is just such a garbage pail right and so what do you do to get your hands around that and I do think there are observational approaches there and it's geology um is you go out and look at rocks and you look at rocks and you measure distribution of roughness you measure geometry you measure distribution of lithology and you ask what's realistic characterization of heterogeneity because we are really missing the mesoscale physics so that's what I want to get at is mesoscale physics which is rated state is a microscopic description and this is a mesoscale phenomenon and so it's a macroscopic phenomenon but there's a mezzo in between and so there is there is some missing physical process there are strategies observational strategies I think they're somewhat different than most of what we've talked about are precursors I do think you can make some progress through an active experiment where you these things like google yamia is doing who by the way is now here at LPL um but uh there's also a lot of geology to do okay so we're really getting behind we really have to is it really quick Paul um yeah I just want to one point this contradict anything Emily said but the pair of dimes that existed in Japan before 2001 is that there were seismic disparities that were fixed and then there was creeping fault around it that was to explain all this post seismic behavior and what we're saying is that maybe we can't assume that those asperities are fixed it may be varying in time what is that say about the apparent match between this slip and where the locking was inferred sorry what is that say about the apparent match of the actual coso seismic slip of topography and where the locking was inferred oh but you have the you don't have 90 different locking models and they're all different so I think so you think there's there's nothing to be learned so we're gonna have to stop it's your session Cindy I'm sorry we we because also we have people that are remote that are kind of trying to fit it into various schedules but I just I just want to thank both of our speakers for generating such interesting and for generating such interesting discussion and so thank you very much