WEBVTT Kind: captions Language: en 00:00:00.599 --> 00:00:02.959 Good morning, and welcome to the Earthquake Science Center 00:00:02.960 --> 00:00:06.420 seminar for August 23rd. 00:00:07.490 --> 00:00:09.920 Just wanted to let you know that next week, we’re going to be 00:00:09.930 --> 00:00:16.001 having a talk on the seismic safety and earthquake engineering of the 00:00:16.001 --> 00:00:20.870 Oakland-San Francisco Bay Bridge, which might be of interest to those of us 00:00:20.870 --> 00:00:27.560 who use the bridge, by Professor Hassan Astaneh-Asl of UC-Berkeley. 00:00:27.560 --> 00:00:33.780 And during the week of SCEC, which is – we’ll be meeting on September 13. 00:00:33.780 --> 00:00:36.560 We’re not planning to have seminar, 00:00:36.560 --> 00:00:42.110 but there will be a seminar for the weeks prior to the SCEC meeting. 00:00:42.110 --> 00:00:47.479 Our speaker today, Lise Retailleau, is a – has a Ph.D. from the Institut 00:00:47.479 --> 00:00:55.149 du Globe – Physique du Globe in Paris, and she did a postdoc at ISTerre 00:00:55.149 --> 00:01:00.749 at the Université Grenoble. And she’s been at Stanford 00:01:00.749 --> 00:01:03.059 for the past six months or so. - Yeah. 00:01:03.059 --> 00:01:08.469 - So she’s very kindly agreed to tell us about the core-mantle boundary. 00:01:08.469 --> 00:01:14.729 We will be having lunch with Lise, and we do have some time for people who 00:01:14.729 --> 00:01:19.620 want – might want to talk to Lise this afternoon after about 2:30. So, Lise? 00:01:20.120 --> 00:01:21.520 - Thank you. 00:01:22.240 --> 00:01:25.020 Okay, thank you for the introduction, and thank you for inviting me. 00:01:25.030 --> 00:01:27.880 Do you hear me well? Okay, great. 00:01:27.880 --> 00:01:32.740 So I’m going to talk to you about things that I did when I was in Grenoble. 00:01:32.740 --> 00:01:36.719 So at ISTerre, in the laboratory there. 00:01:36.720 --> 00:01:42.120 So I worked with Pierre Boué, who I think you’ve seen two years ago maybe, 00:01:42.120 --> 00:01:47.040 Lei Li, who is a Ph.D. student there, and Michel Campillo. 00:01:47.049 --> 00:01:51.399 And so I’m going to talk to you about how we can use correlation – 00:01:51.400 --> 00:01:56.340 noise correlations to try to image the core-mantle boundary. 00:01:57.369 --> 00:02:01.620 So first, very briefly, in seismology, one of the big fields 00:02:01.630 --> 00:02:05.520 is to try to image the Earth at different depths. 00:02:05.520 --> 00:02:07.980 And so this is usually done by earthquake data. 00:02:07.980 --> 00:02:11.670 So very, very simply, when you have an earthquake, you will look at 00:02:11.670 --> 00:02:19.580 the different arrivals of the wave and try to get ideas of [inaudible] 00:02:19.580 --> 00:02:24.000 reflections within the Earth to try to get its structure. 00:02:24.000 --> 00:02:27.970 One of the limitations of this method is that you need to have 00:02:27.970 --> 00:02:30.970 earthquakes everywhere, which is not the case. 00:02:30.970 --> 00:02:35.170 And so you cannot really sample everywhere in the Earth. 00:02:35.170 --> 00:02:39.240 So that’s one of the reasons why a method has been developed 00:02:39.240 --> 00:02:44.120 in the last decades, which is ambient noise correlations. 00:02:44.120 --> 00:02:49.540 So the basic idea of ambient noise correlations is that, 00:02:49.540 --> 00:02:54.190 if you have a random wave field, and you compute the correlations 00:02:54.190 --> 00:03:00.160 between two receivers, then you can get an approximation 00:03:00.160 --> 00:03:03.540 of the propagation between these stations. 00:03:03.550 --> 00:03:09.150 So the signal that we will get with this correlation will be as if you had – 00:03:09.150 --> 00:03:15.400 one of the receivers was a source and the other one was recording this source. 00:03:15.400 --> 00:03:18.050 And you do this – you can have this the other way. 00:03:18.050 --> 00:03:21.470 When you calculate your correlation, you will have a causal and yet an 00:03:21.470 --> 00:03:26.150 anticausal part, so a positive and a negative time, which will be, 00:03:26.150 --> 00:03:29.930 one, the propagation between receiver 1 to receiver 2, 00:03:29.930 --> 00:03:33.020 and the other way, receiver 2 to receiver 1. 00:03:33.020 --> 00:03:38.470 So this is very interesting because it can lead to have a lot – new data sets. 00:03:38.470 --> 00:03:42.830 Because you don’t need to have the location of the earthquake. 00:03:42.830 --> 00:03:46.070 You’re dependent only on the station. 00:03:46.070 --> 00:03:49.200 So you’re also dependent on the – on the location of the station, 00:03:49.200 --> 00:03:54.760 which are also not everywhere, but you are a bit less limited. 00:03:54.760 --> 00:03:58.900 So this is an example that shows what you get with correlation. 00:03:58.900 --> 00:04:06.140 For example, so this is from Lin et al. in 2009. Where is my … 00:04:06.140 --> 00:04:07.780 There. 00:04:07.780 --> 00:04:13.070 Okay, so if you look at the right figure, you have – you see a star. 00:04:13.070 --> 00:04:14.600 And this is in California. 00:04:14.600 --> 00:04:18.750 And this star is not an earthquake. It’s a station. 00:04:18.750 --> 00:04:22.120 And if we correlated this station – they correlated this station 00:04:22.120 --> 00:04:24.720 with all of the other stations, and you can reproduce 00:04:24.720 --> 00:04:30.670 a propagation of waves from this reference station. 00:04:30.670 --> 00:04:34.730 So you can see that, at t-1, you have the waves, and then they propagate 00:04:34.730 --> 00:04:40.170 with the time increasing. So you reproduce as if you had an earthquake. 00:04:40.170 --> 00:04:45.160 So this has been used with surface waves, mainly, 00:04:45.160 --> 00:04:50.250 because they are the strongest. And this has been used at different 00:04:50.250 --> 00:04:55.840 scales, so at very small scales, for example, with Mordret et al. in 2013. 00:04:55.840 --> 00:05:01.060 Then one that’s very new in the center of California from Shapiro. 00:05:01.060 --> 00:05:02.440 So this is regional. 00:05:02.440 --> 00:05:07.340 And then at very big scales – you can see on the right from Nishida. 00:05:07.340 --> 00:05:12.190 The question now is, can we do the same types of studies but with body waves. 00:05:12.190 --> 00:05:14.560 Because with surface waves, you can get just information 00:05:14.560 --> 00:05:18.060 about the surface and the shallower parts of the Earth. 00:05:18.060 --> 00:05:21.900 But now can we go deeper and get information about, 00:05:21.900 --> 00:05:24.190 for example, the core-mantle boundary? 00:05:24.190 --> 00:05:31.220 So body waves have been observed for still a few – also a few years now. 00:05:31.220 --> 00:05:35.200 And they have been observed at different scales – at very 00:05:35.210 --> 00:05:40.900 local scales and then at crustal scales. And this is on the second figure an 00:05:40.900 --> 00:05:48.050 example where you can observe SmS – so S wave reflected at the Moho. 00:05:48.800 --> 00:05:56.200 A few years after, Poli et al. in 2012 showed that you could – that you could 00:05:56.210 --> 00:06:01.010 observe deeper reflections so that you – so that you could go deeper in the Earth. 00:06:01.010 --> 00:06:04.680 So this is another [inaudible] they made on the LAPNET array, 00:06:04.680 --> 00:06:12.650 which is in Finland. And the two figures that you can see are vespagrams. 00:06:12.650 --> 00:06:19.400 So a vespagram, I will explain a bit more later, but is a representation 00:06:19.400 --> 00:06:24.860 of the signals as opposed to time and slowness. 00:06:24.860 --> 00:06:31.590 And so we observed – if you look at the first – at the data picture, this is 00:06:31.590 --> 00:06:35.960 the correlations that they computed, and you see two different arrivals. 00:06:35.960 --> 00:06:42.420 And if you look under this at the model, it also shows these two arrivals that 00:06:42.420 --> 00:06:50.420 correspond to the – to two interfaces – the 410 kilometers and then 660. 00:06:50.420 --> 00:06:52.930 And so this means that with correlation, 00:06:52.930 --> 00:06:58.440 we can go deep and start seeing these kind of interfaces. 00:06:58.440 --> 00:07:04.240 So from this, Boué et al. in 2013 tried to see if we could see all of the – 00:07:04.240 --> 00:07:07.230 all of the body wave phases that propagate through the Earth 00:07:07.230 --> 00:07:10.970 and if we could – and wondering if you could use them. 00:07:10.970 --> 00:07:17.610 So this is what they got – the time- distance representation that they got. 00:07:17.610 --> 00:07:21.620 And using a lot of stations – [inaudible]. 00:07:23.960 --> 00:07:27.540 So, on the left, you can see the correlations that they got – that they 00:07:27.550 --> 00:07:33.210 computed using a lot of stations that are represented on the top right. 00:07:33.210 --> 00:07:40.169 So they used, I think, about 400 stations and correlated all of the signals and 00:07:40.169 --> 00:07:46.300 plotted them as opposed to distance. So it is the X axis. 00:07:46.300 --> 00:07:49.820 And then you can see, in time, the different arrival of the different waves. 00:07:49.820 --> 00:07:54.120 So they were able to extract a lot of phases. 00:07:54.120 --> 00:07:57.310 And on the right, you can see the synthetics 00:07:57.310 --> 00:07:59.710 that they – that they made to compare. 00:07:59.710 --> 00:08:01.670 And you can see that it corresponds quite well 00:08:01.670 --> 00:08:06.160 and so that you can extract a lot of these waves. 00:08:06.160 --> 00:08:11.300 For what you can see also, on the bottom figure, which shows 00:08:11.300 --> 00:08:17.500 the number of pairs of stations that they used for each bin of distance. 00:08:17.500 --> 00:08:20.470 So you see that they used a lot of stations. 00:08:20.470 --> 00:08:25.850 And what you can see also is that the repetition is not – is not homogenous, 00:08:25.850 --> 00:08:31.100 and for example, at antipodal distance, there is not much data. 00:08:31.720 --> 00:08:40.360 And finally, one – another paper by Poli in 2015 showed that you could 00:08:40.370 --> 00:08:46.820 see even deeper phases quite precisely. Because before, we show – we can see 00:08:46.820 --> 00:08:50.700 that we reproduce the arrivals, but then, what do they mean? 00:08:50.700 --> 00:09:00.120 And so here is an observation of the PdP arrival at the core-mantle boundary. 00:09:00.129 --> 00:09:03.980 So that shows the D-double-prime reflection, 00:09:03.980 --> 00:09:07.129 so which is above the core-mantle boundary. 00:09:07.129 --> 00:09:10.249 So here they compared – so this is, again, a vespagram, 00:09:10.249 --> 00:09:15.720 so a representation of the signals as opposed to time and slowness. 00:09:15.720 --> 00:09:17.690 And you can see on the left an earthquake, 00:09:17.690 --> 00:09:20.670 and on the right, the correlations that they get. 00:09:20.670 --> 00:09:23.319 And they compare this, and in both cases, you can see 00:09:23.319 --> 00:09:28.440 the P arrival, which is quite clear, then a PcP and a PdP. 00:09:28.440 --> 00:09:31.740 Both PdP don’t correspond exactly to each other 00:09:31.740 --> 00:09:36.360 because the earthquake that they used to compare is at depth. 00:09:36.360 --> 00:09:39.100 So the arrival is going to be a bit different. 00:09:39.100 --> 00:09:42.320 But what they showed is that you can – you could really extract a PdP 00:09:42.320 --> 00:09:46.840 that could be used to image the core-mantle boundary. 00:09:47.660 --> 00:09:51.520 So this is what I’m going to talk about now. 00:09:51.529 --> 00:09:56.410 So what can we do something about it? Can we really observe – can we get 00:09:56.410 --> 00:10:01.189 enough information to be able to get a real image of the core-mantle boundary? 00:10:01.189 --> 00:10:03.100 So I will first talk a little bit about the 00:10:03.100 --> 00:10:06.660 sources of body waves and how they are reconstructed. 00:10:06.660 --> 00:10:09.949 And then I will show you what we did to image 00:10:09.949 --> 00:10:14.440 the northern Atlantic Ocean – well, around there. 00:10:14.440 --> 00:10:21.860 So now we’re starting to have an idea of what the sources of body waves are 00:10:21.860 --> 00:10:24.999 because usually people have been studying surface waves, 00:10:24.999 --> 00:10:27.100 and in surface waves, you have your station, 00:10:27.100 --> 00:10:32.360 and you have your sources in the – in [inaudible] behind the stations. 00:10:32.360 --> 00:10:35.269 And this will generate the signals. In body waves, it’s a little bit 00:10:35.269 --> 00:10:41.529 more complicated because you want wave arrivals at high take-off angles. 00:10:41.529 --> 00:10:46.110 And so what we suppose is that, for example, to create a P phase 00:10:46.110 --> 00:10:48.960 between two stations, you will have to have a source that is, 00:10:48.960 --> 00:10:53.259 for example – it’s the top right figure, you have – you have the star. 00:10:53.259 --> 00:10:56.619 You have to have a star that generates a P arrival 00:10:56.619 --> 00:11:01.379 at the first station and then a P arrival at the second station. 00:11:01.379 --> 00:11:03.850 And if you correlate these signals, 00:11:03.850 --> 00:11:06.600 then you will get a P between the two stations. 00:11:06.600 --> 00:11:11.980 So this shows that it’s not very easy to have these kind of body waves, 00:11:11.980 --> 00:11:16.149 and so that’s why we will have to stack a lot, lot of stations – 00:11:16.149 --> 00:11:22.180 a lot of signals and a lot of time to have a nice signal-to-noise ratio. 00:11:22.180 --> 00:11:28.259 And so this is the same for – to get a PcP, which is an example here. 00:11:28.259 --> 00:11:32.449 You will have to have a source that generates a PKP at one of the – 00:11:32.449 --> 00:11:36.269 at one of the stations and a PKPPcP at the other one, 00:11:36.269 --> 00:11:39.160 so that when you correlate, you get a PcP. 00:11:39.160 --> 00:11:43.220 So then, what are exactly the sources for body waves? 00:11:43.220 --> 00:11:47.749 So it will be – we suppose that it should be, and it has been shown that 00:11:47.749 --> 00:11:52.970 it was linked to the microseismic sources, so – that are linked to the interaction 00:11:52.970 --> 00:11:59.910 between the ocean and the solid earth. So oceanic waves and all of this. 00:11:59.910 --> 00:12:05.560 But Boué et al. in 2014 have also observed – so this is, again, 00:12:05.560 --> 00:12:13.200 a section of distance as opposed to time of the correlations that they have. 00:12:13.209 --> 00:12:15.730 And on the right, you also have the synthetics. 00:12:15.730 --> 00:12:20.480 And if you look at the left panel, which is the correlations, 00:12:20.480 --> 00:12:23.720 if you look at it, you can see that some arrivals are 00:12:23.720 --> 00:12:26.569 a bit surprising compared to [inaudible] synthetics. 00:12:26.569 --> 00:12:31.269 For example, if you look at the ScS, where they are over here, you can see 00:12:31.269 --> 00:12:36.350 that the arrival of the ScS is quite strong, which we do not really expect. 00:12:36.350 --> 00:12:39.429 And this is a vertical component. 00:12:39.429 --> 00:12:42.369 Yeah, so we don’t – do not really expect this, 00:12:42.369 --> 00:12:46.600 so they were wondering what it could be linked to. 00:12:46.600 --> 00:12:53.989 And so they observed – they compared the seismic moment during the year. 00:12:53.989 --> 00:12:58.750 So this is 2008 – yes. So they computed the 00:12:58.750 --> 00:13:02.800 seismic moment during the year. So there was an earthquake, for example. 00:13:02.800 --> 00:13:07.019 And they plotted it – this is the first panel, so it’s the black line. 00:13:07.019 --> 00:13:09.920 And so you can see that the line goes up when you have an earthquake. 00:13:09.920 --> 00:13:16.959 And they compared this to how well is each body wave reconstructed. 00:13:16.959 --> 00:13:23.350 So what they do is they take each day – each correction each day, 00:13:23.350 --> 00:13:28.749 and they compare it to the stack of one year of correlations. 00:13:28.749 --> 00:13:34.480 And so they see – this shows, if one day can reconstruct – 00:13:34.480 --> 00:13:40.769 they can reconstruct the signals as well as one-year stack, basically. 00:13:40.769 --> 00:13:46.060 So what you can see is that what it mainly shows is that, here if you 00:13:46.060 --> 00:13:50.629 look at big earthquakes, you can also see picks for the different phases. 00:13:50.629 --> 00:13:53.910 So here, you have different examples of phases. 00:13:53.910 --> 00:13:55.809 And these are at long-period. 00:13:55.809 --> 00:13:58.639 So you can see that it corresponds quite well. 00:13:58.639 --> 00:14:03.049 So it’s – earthquakes here seem to generate some signals. 00:14:03.049 --> 00:14:08.990 And this is – it can be not so great because, since earthquakes 00:14:08.990 --> 00:14:13.680 are quite deep, so the reconstruction may be a bit – it might – it might generate 00:14:13.680 --> 00:14:19.540 some problems in amplitude and of phase. Because if you have a source that is 00:14:19.540 --> 00:14:26.740 very precise, it might – yeah, generate some problems and things like that. 00:14:27.660 --> 00:14:31.720 And also some other mechanisms that might happen that – 00:14:31.720 --> 00:14:35.000 because you have one source that is very strong, it might generate 00:14:35.000 --> 00:14:41.209 something that I will not talk about here, but that we’re trying to work on. 00:14:41.209 --> 00:14:45.290 And so now, here is what you get if you have – so the 00:14:45.290 --> 00:14:51.080 two right panels are still the same. And on the left one, you have the 00:14:51.080 --> 00:14:55.160 correlations without the earthquakes – well, without the main earthquakes. 00:14:55.170 --> 00:14:58.689 And here you can see that what you get seems a bit more 00:14:58.689 --> 00:15:03.189 what we expect and what we want. But you don’t have such a strong 00:15:03.189 --> 00:15:07.360 ScS anymore. You don’t have such a strong PcP anymore. 00:15:07.360 --> 00:15:13.240 And it looks a bit more – well, more like the synthetic section. 00:15:14.300 --> 00:15:22.459 However, if you – if you look at short periods – so shorter periods, then – 00:15:22.459 --> 00:15:26.569 and you do the same process, so you compare the earthquakes 00:15:26.569 --> 00:15:30.839 with the reconstruction of the body waves, then you will 00:15:30.839 --> 00:15:41.910 not see such a big correlation between earthquake and the correlation. 00:15:41.910 --> 00:15:45.699 And so – which is what I’m working to – 00:15:45.699 --> 00:15:50.160 what I’m going to talk about later. I’m going to show short periods. 00:15:50.160 --> 00:15:53.489 But this shows that, in short periods, we’re quite confident that we are 00:15:53.489 --> 00:15:58.680 not just selecting one source that might have influence over all of it, 00:15:58.680 --> 00:16:01.149 which might create some problems. 00:16:01.149 --> 00:16:05.470 This also shows that, if you look at the correlation coefficient 00:16:05.470 --> 00:16:09.660 for the short periods, it’s quite low, which means that we need to 00:16:09.660 --> 00:16:13.740 stack a lot of days to really reconstruct something. 00:16:14.840 --> 00:16:19.809 So this is going to be one of the very important points is that, in correlation, 00:16:19.809 --> 00:16:23.929 and especially in body waves, but also in surface waves, 00:16:23.929 --> 00:16:30.739 you need to stack a lot of stations to – so to sum and to combine a lot of days 00:16:30.740 --> 00:16:35.380 to have a nice signal-to-noise ratio to do something good. 00:16:36.460 --> 00:16:44.400 So the body wave sources are expected to be linked to the interaction 00:16:44.410 --> 00:16:48.189 between ocean and the solid earth. So this has been observed in different 00:16:48.189 --> 00:16:54.320 studies – Gerstoft in 2006 and 2008, [inaudible] associated the 00:16:54.320 --> 00:17:00.299 observation of body waves to storms – to storm-like events. 00:17:00.299 --> 00:17:08.380 And Landès et al. shows – these are the figures – that you could observe also 00:17:08.380 --> 00:17:16.990 original variations of the P wave reconstruct – P wave in the microseisms. 00:17:16.990 --> 00:17:23.160 And then other people, like for example, Ardhuin and Farra and [inaudible] 00:17:23.160 --> 00:17:28.639 and all these people are working on how exactly are 00:17:28.640 --> 00:17:33.680 the body waves generated by the microseisms. 00:17:34.720 --> 00:17:38.100 What we can still observe with this is that – 00:17:38.100 --> 00:17:41.389 I said that the problem was that – with earthquakes was that 00:17:41.389 --> 00:17:44.790 they were very localized and so could generate bias. 00:17:44.790 --> 00:17:51.409 And if we look at this, we can also see that the microseismic noise is not – 00:17:51.409 --> 00:17:57.220 is not constant all about the Earth. Because first, it’s not inland. 00:17:57.220 --> 00:18:01.240 It’s always in the ocean. And also it moves with seasons. 00:18:01.250 --> 00:18:05.220 So what is going to be very important too is to always try to 00:18:05.220 --> 00:18:10.919 understand how the sources work to not be just imaging the sources, 00:18:10.919 --> 00:18:14.780 but being – imaging areas of structure that you want. 00:18:14.780 --> 00:18:18.679 And so for this, what we’ll always be looking at is the comparison 00:18:18.679 --> 00:18:23.080 between the causal and anticausal part. Because, as I said, if you look at 00:18:23.080 --> 00:18:27.130 the causal part in the correlation, you will be looking at the propagation 00:18:27.130 --> 00:18:30.990 from one station to the other, and the anticausal part is the other way. 00:18:30.990 --> 00:18:36.090 So you will be using different sources, and so if you can get a similar result 00:18:36.090 --> 00:18:38.860 in both cases, you’re more confident that at least you’re 00:18:38.860 --> 00:18:41.580 not just looking at the sources. 00:18:42.900 --> 00:18:48.299 And finally, I wanted to talk to you about some other – another thing 00:18:48.299 --> 00:18:52.890 that we observe on correlation of body waves is that, for example, here – 00:18:52.890 --> 00:18:59.330 and this is work by Li – Lei Li, who is doing a Ph.D. in Grenoble. 00:18:59.330 --> 00:19:02.220 So we correlated the signals between the LAPNET in Finland 00:19:02.220 --> 00:19:09.809 and the FNET in Japan and tried to – and tried to reconstruct the P – 00:19:09.809 --> 00:19:16.060 the PcP waves and also potential PdP. 00:19:16.060 --> 00:19:18.360 And if you look first at the vespagram on the right, 00:19:18.360 --> 00:19:22.100 you can see that you get the P, the PcP, but you also see another 00:19:22.100 --> 00:19:26.080 arrival before, which is quite surprising because we don’t – 00:19:26.080 --> 00:19:28.470 well, we don’t expect it, and it should not be there. 00:19:28.470 --> 00:19:32.840 And if you look at the signals – on the left, you see the signal. 00:19:32.840 --> 00:19:36.760 You still the P, the PcP, and you really see a clear – a clear other phase 00:19:36.760 --> 00:19:41.740 that is not – it doesn’t correspond to the propagation between the stations. 00:19:41.740 --> 00:19:48.110 So this is what you – what we call a spurious phase, which is a 00:19:48.110 --> 00:19:52.580 precursory phase, and that we also see – we also see some surface waves. 00:19:52.580 --> 00:19:59.789 But here – and the way he interpreted it is that it is linked to 00:19:59.789 --> 00:20:04.080 the correlation between – if you have a source, and then it arrives at 00:20:04.080 --> 00:20:09.600 the P on one of the stations, and at PKP on the other station, 00:20:09.600 --> 00:20:12.750 when you correlate the signals, it will get you an arrival 00:20:12.750 --> 00:20:18.660 that seems to be quite – I mean, normal arrival. 00:20:18.660 --> 00:20:23.289 And if you – and he compared, and he observed, that it will lead to an arrival 00:20:23.289 --> 00:20:28.519 with this slowness and this time arrival. Which is a bit worrying because it means 00:20:28.520 --> 00:20:34.330 that we can have arrivals that do not correspond to the path that we want. 00:20:36.040 --> 00:20:42.170 But in this case, it’s before the P, so it’s easy to say that it’s not what we want. 00:20:42.170 --> 00:20:47.580 But still, we always have to check the time arrival and the slowness of the 00:20:47.580 --> 00:20:52.120 phase that we want to see to see they don’t look a bit weird and if they’re 00:20:52.120 --> 00:20:57.040 really what we want to observe and not some weird combination of phases. 00:20:57.049 --> 00:21:04.050 And Lei continued in this study because he observed this spurious arrival, 00:21:04.050 --> 00:21:11.169 but actually, you can have a lot of them. Because different body waves can 00:21:11.169 --> 00:21:17.370 interact weirdly and generate some arrivals in the correlations. 00:21:17.370 --> 00:21:22.150 And here is just, on the left, a figure that he got combining a lot of 00:21:22.150 --> 00:21:28.649 different body waves to see how it will – which arrival would be linked to it. 00:21:28.649 --> 00:21:32.230 And if you look, for example, at the first arrivals, which are 00:21:32.230 --> 00:21:39.039 way before the P arrival, you can see that it quite corresponds to the right panel, 00:21:39.039 --> 00:21:42.000 which was actually a figure that I showed you before, 00:21:42.000 --> 00:21:45.570 which was the figure with – where, in the correlation, 00:21:45.570 --> 00:21:48.669 we used earthquake – they used earthquakes. 00:21:48.669 --> 00:21:51.840 And you can see that you have – this first arrival, you can see, 00:21:51.840 --> 00:21:54.740 with [inaudible], you have this first arrival, which we 00:21:54.750 --> 00:22:00.950 do not correspond to the real propagation between the stations. 00:22:00.950 --> 00:22:05.610 And these are linked to these combinations of different waves. 00:22:05.610 --> 00:22:09.980 So this is really something that we need to check each time to not, 00:22:09.980 --> 00:22:15.799 yeah, be just imaging something that – thinking that we are imaging something 00:22:15.799 --> 00:22:19.250 where we are actually imaging something else. 00:22:19.250 --> 00:22:24.340 So now for the real – well, full imaging. 00:22:24.340 --> 00:22:28.919 So what we did is that we tried to – so to get an image of the 00:22:28.919 --> 00:22:33.289 northern Atlantic Ocean. Because this is a place where 00:22:33.289 --> 00:22:36.480 there is not so many earthquakes. And so this is a place 00:22:36.480 --> 00:22:42.340 where information – seismological information could be helpful. 00:22:42.340 --> 00:22:48.130 And so we used 900 stations in the year 2014. 00:22:48.130 --> 00:22:52.809 And so we used all the 2014 to be able to stack quite a lot. 00:22:52.809 --> 00:22:55.980 We used the vertical component. 00:22:55.980 --> 00:23:03.389 And so very briefly, the process of the – to work on the data. 00:23:03.389 --> 00:23:08.200 So we correct the instrument response of all the daily records. 00:23:08.200 --> 00:23:13.039 And then the records are split in four-hour segments. 00:23:13.039 --> 00:23:18.450 And when there’s an event, so where there is a peak in the signal, 00:23:18.450 --> 00:23:22.840 we suppress that segment to suppress a potential earthquake 00:23:22.840 --> 00:23:27.590 or other signal that we don’t want to use. 00:23:27.590 --> 00:23:33.720 Then the signals are normalized in the frequency domain. 00:23:33.720 --> 00:23:36.860 And then the corrections are computed on all the four-hour segments 00:23:36.860 --> 00:23:42.580 and then stacked to get only one correlation for the whole year. 00:23:42.580 --> 00:23:45.809 And then we filter in the frequency that we’re going to work on. 00:23:45.809 --> 00:23:51.150 So this is the signal that we get for – between all of these stations. 00:23:51.150 --> 00:23:54.690 So we compute the correlation between all of the different coupled stations – 00:23:54.690 --> 00:23:59.190 all of the couple that go through the Atlantic. 00:23:59.190 --> 00:24:00.919 And then we stack them in different – 00:24:00.919 --> 00:24:05.149 as opposed to the – we put them as opposed to their distance. 00:24:05.149 --> 00:24:08.730 And so you can see that we reconstruct quite well 00:24:08.730 --> 00:24:14.220 the P and the PcP, which was already quite a good news. 00:24:15.460 --> 00:24:20.240 The signal-to-noise ratio, it’s quite good, because if you look at the right, 00:24:20.250 --> 00:24:26.820 the figure on the right is the number of pairs of stations at each bin. 00:24:26.820 --> 00:24:31.850 So at – for example – and I don’t know – 55 degrees of distance, you have 00:24:31.850 --> 00:24:37.220 about 300 pairs of stations that have sent in just this distance bin. 00:24:37.220 --> 00:24:42.470 So we sum a lot – the signals to get something that looks quite nice. 00:24:42.470 --> 00:24:48.070 That’s really the point, that we have to really send things and, like, 00:24:48.070 --> 00:24:53.679 that we’re able to extract a very nice – very nice phases. 00:24:53.679 --> 00:24:56.900 We also see that we also have a spurious arrival here, 00:24:56.900 --> 00:25:00.730 which is a bit less clear than the others. But we still see these arrivals, 00:25:00.730 --> 00:25:04.190 and we suppose that it’s linked to what I explained to you before – 00:25:04.190 --> 00:25:08.700 the combination of body waves that do not reconstruct 00:25:08.700 --> 00:25:13.320 and the propagation between the stations that we’re looking at. 00:25:13.320 --> 00:25:16.700 So now I also told you that what we were always doing was comparing 00:25:16.700 --> 00:25:22.309 the causal and anticausal parts. So this is – this is – this is it. 00:25:22.309 --> 00:25:27.860 So here, we are between 3 and 8 seconds of period – so short-period. 00:25:27.860 --> 00:25:30.860 And you can see – and you have the anticausal part 00:25:30.860 --> 00:25:34.110 on the left and the causal part on the right. 00:25:34.110 --> 00:25:37.519 And so the causal part – the anticausal part shows the propagation 00:25:37.519 --> 00:25:41.860 between the U.S. and Europe. And the causal part, the other way. 00:25:41.860 --> 00:25:46.100 And what we see is that, in both cases, we reconstruct quite well the different – 00:25:46.100 --> 00:25:49.870 the two phases that we want – that we study. 00:25:49.870 --> 00:25:54.620 So this is also quite a good news. 00:25:54.620 --> 00:26:00.029 What we also see is that – we don’t really see any variations. 00:26:00.029 --> 00:26:05.179 Because if you want to get information of what is happening above the core- 00:26:05.179 --> 00:26:12.260 mantle boundary, what we are expecting to find is arrival before the PcP phase. 00:26:12.260 --> 00:26:16.169 And so here we stack over the whole region, which is quite big. 00:26:16.169 --> 00:26:19.830 And so we average, over the [inaudible] variations, it varies. 00:26:19.830 --> 00:26:22.380 So here we are only – we don’t see much. 00:26:22.380 --> 00:26:25.639 And it corresponds very well to the 1D model which are 00:26:25.639 --> 00:26:29.720 the different – the points that are plotting here. 00:26:29.720 --> 00:26:37.840 So how do we try to image? What we do is that we did 00:26:37.840 --> 00:26:42.760 an array analysis of a lot of different couples of subarrays. 00:26:42.769 --> 00:26:46.190 We don’t just take one station on each side and correlate it. 00:26:46.190 --> 00:26:51.179 We take subarrays, take the – and each couple of subarray 00:26:51.180 --> 00:26:56.700 will correspond to a reflection point that I show here and – approximately. 00:26:56.700 --> 00:27:01.919 And we – so we take all of the potential couples and do analysis 00:27:01.920 --> 00:27:05.420 on this that will correspond to one reflection point. 00:27:06.540 --> 00:27:12.360 So for this, we do vespagram analysis. So what is a vespagram analysis? 00:27:12.370 --> 00:27:18.279 So if we have – the basic idea of this analysis is that you suppose that 00:27:18.279 --> 00:27:21.519 you have a plane wave that arrives to your array. 00:27:21.519 --> 00:27:26.909 And that, in that case, the arrival – the time arrivals 00:27:26.909 --> 00:27:32.830 of the phase will be linked only to the geometry of the array 00:27:32.830 --> 00:27:36.809 and the slowness of the – of the wave that arrives. 00:27:36.809 --> 00:27:42.970 And in that case, you can just try different slownesses. 00:27:42.970 --> 00:27:48.980 And if you are at the correct slowness of the wave, then the – then the signal 00:27:48.980 --> 00:27:54.649 will be in phase. And when you stack them, you will have a peak of energy. 00:27:54.649 --> 00:27:59.590 And so what you do is just – you try different slownesses, and you stack 00:27:59.590 --> 00:28:03.669 what you have shifted your signal with the different slownesses. 00:28:03.669 --> 00:28:05.860 And when you are at the right slowness – for example, 00:28:05.860 --> 00:28:08.179 you see on the right – when you are at the right slowness, 00:28:08.180 --> 00:28:10.140 then you will have more energy. 00:28:10.140 --> 00:28:15.499 That’s basically the idea. You just [inaudible] stack, and that’s it. 00:28:15.499 --> 00:28:21.380 So we do this for all of the couples of subarrays around the Atlantic. 00:28:21.380 --> 00:28:26.730 And this is an example – a nice one – of what we get. 00:28:26.730 --> 00:28:32.320 So here – the vespagrams are plotted as opposed to the P arrival. 00:28:34.009 --> 00:28:41.399 And so what we – what we see here, that we see quite well the P and the 00:28:41.399 --> 00:28:47.659 PcP arrival and that they correspond well with the 1D – with the 1D model. 00:28:47.659 --> 00:28:52.779 And what we also see that, before the PcP, we see kind of an arrival. 00:28:52.779 --> 00:28:55.330 And what we want to do is that we want to automatically 00:28:55.330 --> 00:28:59.909 try to find if there is something before the PcP. 00:28:59.909 --> 00:29:05.259 So what we do in a – in a first simple analysis, what we did is that we were – 00:29:05.259 --> 00:29:09.720 we tried to look, like, if there is an arrival between – if there is 00:29:09.720 --> 00:29:17.950 an interface above the core-mantle boundary, what time and – 00:29:17.950 --> 00:29:22.770 at what time and at which slowness would this imply an arrival. 00:29:22.770 --> 00:29:27.880 So if you have the PcP – if you have – if you have a reflection at the 00:29:27.880 --> 00:29:32.029 core-mantle boundary, then you will have an arrival, which is the PcP. 00:29:32.029 --> 00:29:37.200 And then we do this for all of – for all potential depths between the – 00:29:37.200 --> 00:29:39.830 between the P and the PcP. 00:29:39.830 --> 00:29:46.750 And this is the green points that show, if there was a reflection at a 00:29:46.750 --> 00:29:50.340 certain point, then you would have an arrival there. 00:29:50.340 --> 00:29:54.120 And so we do this, and we take the amplitude 00:29:54.120 --> 00:29:58.700 and the vespagram at all of these different points, 00:29:58.700 --> 00:30:03.100 and we get the energy of that to see if we really have something. 00:30:04.080 --> 00:30:08.620 And so this is what it leads to. So you still have the vespagram here. 00:30:08.630 --> 00:30:11.950 And what we do is just take the energy – so for example, for the P, 00:30:11.950 --> 00:30:16.809 we take the energy around the P, and we do this for all the 00:30:16.809 --> 00:30:20.789 different depths between the P and PcP and also above. 00:30:20.789 --> 00:30:24.029 So we don’t – we don’t expect to have a reflection near the P. 00:30:24.029 --> 00:30:27.390 But it was just to not – to not make a choice and say, we don’t take – 00:30:27.390 --> 00:30:30.720 we only want to look at things between this depth. 00:30:30.720 --> 00:30:33.919 We look – we tried to look at all of it. 00:30:33.919 --> 00:30:36.230 And so we get – we kind of – this kind of figure. 00:30:36.230 --> 00:30:39.940 So you have the PcP, the P, and then another – 00:30:39.940 --> 00:30:45.309 what appears to maybe be an arrival before the PcP. 00:30:45.309 --> 00:30:49.620 So we do this for all the different couple of subarrays. 00:30:49.620 --> 00:30:53.990 So we have subarrays on each side. Each time we compute a vespagram. 00:30:53.990 --> 00:30:58.889 From the vespagram, we get this figure. So it’s kind of – we first have 00:30:58.889 --> 00:31:01.299 a vespagram as opposed to time and slowness, 00:31:01.299 --> 00:31:06.490 and we transform it to get kind of a vespagram as opposed to depth. 00:31:06.490 --> 00:31:08.710 And so we do this for all of our different couples, 00:31:08.710 --> 00:31:13.799 so for all of our reflection points. And this is what we get on the right. 00:31:13.799 --> 00:31:18.500 So what you have on the right is the figure on the bottom left, 00:31:18.500 --> 00:31:22.899 but as opposed to its distance. And so you have the 00:31:22.899 --> 00:31:27.789 black very small lines. And then I just put a threshold 00:31:27.789 --> 00:31:32.659 at a certain amplitude and put color above that. 00:31:32.660 --> 00:31:36.990 But it is the same curve as you have on the bottom left. 00:31:37.760 --> 00:31:42.280 But I just plot this signal as opposed to distance 00:31:42.289 --> 00:31:45.369 and make it nice with the colors. 00:31:45.369 --> 00:31:49.899 And so what you see, basically, is, on the first arrival 00:31:49.899 --> 00:31:54.279 on the left is the P – is the P arrival. 00:31:54.279 --> 00:31:58.960 What you can see is that it gets quite broad with distance. 00:31:58.960 --> 00:32:02.450 This is linked to the way we do it, and this is something that has to 00:32:02.450 --> 00:32:07.210 be maybe worked on a bit. Because when – at the beginning, 00:32:07.210 --> 00:32:14.490 after the P, you can see that the – our curve moves – changes phase 00:32:14.490 --> 00:32:20.009 in the same time, and for a while, you will get a lot of energy of the P. 00:32:20.009 --> 00:32:24.850 And when you get at far distances, then this spot is going to be big and 00:32:24.850 --> 00:32:30.700 has a lot of influence. And so that’s why it’s a bit wide and not that nice. 00:32:30.700 --> 00:32:34.620 But what you also can see is that we see another arrival, which corresponds to 00:32:34.620 --> 00:32:37.909 the PcP, and which corresponds to the core-mantle boundary. 00:32:37.909 --> 00:32:43.029 So this is – this is nice. Because it means that, at least with 00:32:43.029 --> 00:32:46.940 the body waves, we are able to recover the core-mantle boundary. 00:32:46.940 --> 00:32:50.139 And we are able to, again, see that, under the core-mantle boundary, 00:32:50.139 --> 00:32:53.980 we don’t have any signal. So, yeah, so we should look at the – 00:32:53.980 --> 00:32:57.759 yeah, sorry – the signal is as opposed to depth. 00:32:57.759 --> 00:33:01.070 And so you see the PcP and the P arrival. 00:33:01.070 --> 00:33:08.039 And so this figure shows that we have a PcP and that sometimes we have the 00:33:08.039 --> 00:33:13.509 impression that there’s something a bit wider that may be an arrival before. 00:33:13.509 --> 00:33:17.900 And what we want to do is plot it as opposed to their location. 00:33:17.900 --> 00:33:21.619 So in longitude and latitude. 00:33:21.619 --> 00:33:26.519 And so what we do is, very simply, we choose – we choose a depth, 00:33:26.519 --> 00:33:32.080 take the amplitude that we get, and plot this amplitude as opposed to 00:33:32.080 --> 00:33:35.850 the reflection point location, and then interpolate. 00:33:35.850 --> 00:33:43.460 And this will give us original kind of representation of our analysis. 00:33:43.460 --> 00:33:48.899 So this is – so this is an example of what we get at the depth 00:33:48.900 --> 00:33:51.300 of the core-mantle boundary. 00:33:53.620 --> 00:33:55.380 Yeah. 00:33:55.390 --> 00:34:00.549 And then we can – you can look at the other depths. And this is what you get. 00:34:00.549 --> 00:34:04.630 So we’re looking – we were observing the northern Atlantic Ocean. 00:34:04.630 --> 00:34:10.210 The top – the top color panel is just – is just the bathymetry of this location. 00:34:10.210 --> 00:34:14.000 It’s just to know where we are. And it’s just – the yellow part 00:34:14.000 --> 00:34:19.980 shows that it’s a ridge. But it’s just to see where are, 00:34:19.980 --> 00:34:26.730 but it’s not – we’re not really using it for interpretation. 00:34:26.730 --> 00:34:31.530 So we have – we have plotted our results as opposed to depth. 00:34:31.530 --> 00:34:34.730 And so what we can see first is that, at the core-mantle boundary – 00:34:34.730 --> 00:34:44.480 so 2,890 kilometers of depth is the spot where we have the most – 00:34:44.480 --> 00:34:47.660 where we have the most energy. So it means that it’s where we 00:34:47.670 --> 00:34:52.160 have the most reflection, which is what we expect. 00:34:52.160 --> 00:34:58.760 What we see also is that, under it, so in the core, we don’t really – 00:34:58.760 --> 00:35:02.350 we don’t see much. The amplitudes are very low. 00:35:02.350 --> 00:35:06.820 Which is also what we expect because we did not really expect to have – 00:35:06.820 --> 00:35:14.650 to see some amplitude or some reflection inside the core. 00:35:14.650 --> 00:35:19.520 But what we can also see is that there seems to be a sort of structure 00:35:19.520 --> 00:35:24.610 in the amplitude. And if we look at the slices 00:35:24.610 --> 00:35:28.630 above the core-mantle boundary, we can see that there seems to be 00:35:28.630 --> 00:35:31.110 a structure between the west and the east. 00:35:31.110 --> 00:35:35.640 So this would mean a precursory arrival to the PcP that could be 00:35:35.640 --> 00:35:42.650 linked to the D-double-prime or to some – to a topography 00:35:42.650 --> 00:35:45.200 above the core-mantle boundary. 00:35:45.200 --> 00:35:49.000 And so – and finally, I just wanted to show you, again, the comparison 00:35:49.010 --> 00:35:53.150 between the causal part of the signal and the anticausal part. 00:35:53.150 --> 00:35:56.110 So the results are mainly the same, 00:35:56.110 --> 00:36:01.110 which is what we expect, even if the causal part is a bit noisier. 00:36:01.110 --> 00:36:08.520 But we still get a strong core-mantle boundary – a lot less under it. 00:36:08.520 --> 00:36:16.900 And then some – there seems to be a west-east kind of structure that, yeah, 00:36:16.900 --> 00:36:20.380 might show a structure above the core-mantle boundary, 00:36:20.380 --> 00:36:24.100 which is what we were looking for. 00:36:24.100 --> 00:36:32.180 Okay, so as a conclusion, so we are – with noise correlation, 00:36:32.190 --> 00:36:37.250 we are able to now start to get quite a lot of information 00:36:37.250 --> 00:36:41.280 of the Earth at different depths. And we’re starting to be able to 00:36:41.280 --> 00:36:48.160 look at very deep – very deep places at the core-mantle boundary. 00:36:48.160 --> 00:36:53.290 It has to be noted that it is very important to use a lot of signals 00:36:53.290 --> 00:36:59.620 and to stack a lot, to always check the causal and anticausal parts 00:36:59.620 --> 00:37:02.990 to really compare – to compare them to see if 00:37:02.990 --> 00:37:07.670 we are not just looking at some – at the sources of the signals. 00:37:07.670 --> 00:37:13.060 And then to – it’s also very important to look at the time arrivals 00:37:13.060 --> 00:37:17.900 and the slowness to be sure that we’re looking at the right phases. 00:37:19.120 --> 00:37:21.980 And then, as perspectives, what we want to do now 00:37:21.980 --> 00:37:26.880 is to try to compare to models and to see exactly what we are 00:37:26.880 --> 00:37:32.500 observing and compare – we also want to compare what we get 00:37:32.500 --> 00:37:39.110 with earthquake data to try to find an earthquake and have some – 00:37:39.110 --> 00:37:43.140 and have some signals around the place where the earthquake occurred 00:37:43.140 --> 00:37:46.180 and then at signals on the other side and really be – 00:37:46.180 --> 00:37:50.540 and compare if we get the same observations or not. 00:37:50.540 --> 00:37:53.540 And hopefully we’ll get the same observations. 00:37:53.540 --> 00:37:55.920 And then we are also interested in 00:37:55.920 --> 00:37:59.940 understanding, still, how the sources work. 00:37:59.940 --> 00:38:01.880 Thank you. 00:38:01.880 --> 00:38:06.040 [ Applause ] 00:38:06.040 --> 00:38:09.240 - Thank you very much. Are there some questions? 00:38:10.660 --> 00:38:15.940 Could you – could you elaborate on the interpretation of a structure 00:38:15.940 --> 00:38:20.900 above the core-mantle boundary that could explain these observations? 00:38:20.900 --> 00:38:26.960 - So here – with noise correlation, there’s always a problem of amplitude. 00:38:26.960 --> 00:38:30.530 Like, why – how do you interpret the amplitude of the signals? 00:38:30.530 --> 00:38:33.920 It’s quite a big and difficult subject. 00:38:33.920 --> 00:38:39.680 Here what we represent is the amplitude as opposed to the P arrival. 00:38:39.680 --> 00:38:43.830 So it’s more like – yeah, comparison between the P – with the P arrival. 00:38:43.830 --> 00:38:46.340 And so with this [inaudible], we expect that we might be 00:38:46.340 --> 00:38:52.510 observing the D-double-prime, which is kind of the easy interpretation. 00:38:52.510 --> 00:38:57.980 But we still have to compare with synthetic signals to see if it’s really – 00:38:57.980 --> 00:39:00.640 would – if it would really lead to something like that. 00:39:00.640 --> 00:39:04.270 So that’s something that I’m working on now. 00:39:04.270 --> 00:39:07.940 If you have – if you have this kind of structure as a way we 00:39:07.940 --> 00:39:11.250 observe it, does it work the same way with the synthetics 00:39:11.250 --> 00:39:15.520 to really be able to interpret to really what we exactly see? 00:39:15.520 --> 00:39:19.000 So this is something that still has to be worked on. 00:39:21.800 --> 00:39:24.260 - I have – I have kind of the same question. 00:39:24.270 --> 00:39:30.060 Let’s look at the image. You see the third panel down, it’s – 00:39:30.060 --> 00:39:32.860 or, third or fourth. It’s … - Oh, yeah, sorry. 00:39:32.860 --> 00:39:38.360 - It’s blue to the left, and then it becomes yellow. 00:39:38.360 --> 00:39:44.950 So first – my first question is, is the blue a positive – 00:39:44.950 --> 00:39:49.040 is it a velocity increase or decrease to get that amplitude? 00:39:49.040 --> 00:39:52.420 - So here – we’re not really representing the velocity. 00:39:52.420 --> 00:39:58.120 What we’re representing is – okay, so what we’re representing 00:39:58.120 --> 00:40:00.860 is really an observation of the vespagram. 00:40:00.860 --> 00:40:08.800 So it really shows the arrival – the arrival in the vespagram. 00:40:08.800 --> 00:40:12.310 So what’s – the energy exactly – 00:40:12.310 --> 00:40:18.020 the amplitude exactly says, I’m not completely sure. 00:40:18.300 --> 00:40:23.160 But, yeah, what we – what we show is not a – 00:40:23.160 --> 00:40:28.400 is more of the – of the fact that – do we observe precursory arrivals 00:40:28.400 --> 00:40:32.010 on that, or do we just observe a PcP? 00:40:32.010 --> 00:40:37.221 So it’s more about that. Like, this shows – doesn’t really show 00:40:37.221 --> 00:40:41.560 a variation of slowness. It shows that this should be 00:40:41.560 --> 00:40:47.920 something maybe that arrives above – I don’t know – above – is it above? 00:40:47.920 --> 00:40:51.080 [laughs] Over the core-mantle boundary. - Well, if you go back – 00:40:51.080 --> 00:40:55.500 if you go back several slides, you go – your actual data … 00:40:58.440 --> 00:41:00.500 Okay, there you are. 00:41:00.510 --> 00:41:05.440 So P – the blue is the direct arrival. That’s the direct P wave. 00:41:05.440 --> 00:41:10.270 And the red is the reflected PcP phase. - Yeah. 00:41:10.270 --> 00:41:14.980 - So that’s definitely a velocity increase. - Yeah. 00:41:15.560 --> 00:41:21.920 - So it’s a positive velocity step. It has to be. 00:41:21.920 --> 00:41:24.340 You’re going from the mantle to the core. 00:41:24.340 --> 00:41:27.320 - Oh. - So, but you show a variation – 00:41:27.320 --> 00:41:33.640 a very strong variation from blue to yellow, and I’m wondering if you 00:41:33.640 --> 00:41:36.980 thought about how to explain that variation in terms of 00:41:36.980 --> 00:41:39.720 composition or temperature. Or maybe you’re still processing 00:41:39.720 --> 00:41:42.670 your data and not thinking about composition or temperature changes. 00:41:42.670 --> 00:41:45.520 - So to get – to get information about this, 00:41:45.520 --> 00:41:48.730 what we’re also looking at is at the polarity. 00:41:48.730 --> 00:41:51.190 Because to really get information about this – about the variation 00:41:51.190 --> 00:41:54.260 of velocity, it will be with the polarity that we will get this. 00:41:54.260 --> 00:41:58.620 This is mostly really showing that we have something that’s happening, 00:41:58.620 --> 00:42:01.650 and that – and it’s more we are observing something in time. 00:42:01.650 --> 00:42:07.050 But to really know what this is, we really need to see polarity 00:42:07.050 --> 00:42:10.360 to see if we have a change of sign on it. 00:42:10.360 --> 00:42:14.040 - But, I mean, there’s your data. You don’t need to go to a vespagram. 00:42:14.040 --> 00:42:16.010 You can just look at this diagram, and it tells you the polarity. 00:42:16.010 --> 00:42:18.830 It’s right – it’s right there in front of you. 00:42:18.830 --> 00:42:21.650 - Yes, but this is a stack of all of them – 00:42:21.650 --> 00:42:26.000 of all of the – we get all of the signal, and it’s not original observation. 00:42:26.000 --> 00:42:28.020 And if we look at the original observation, then we will 00:42:28.020 --> 00:42:32.510 have a lot of less data. So it’s not as easy to see. 00:42:32.510 --> 00:42:34.810 And we have – and also, we haven’t seen – 00:42:34.810 --> 00:42:39.120 we haven’t looked at all of the different subarray couple signals. That’s true. 00:42:39.120 --> 00:42:42.880 - Okay. I like this better than the vespagram. 00:42:42.880 --> 00:42:44.740 [laughter] 00:42:47.480 --> 00:42:50.780 - You point out there are some issues with spurious phases? 00:42:50.780 --> 00:42:54.390 - Mm-hmm. - And I guess, in some cases, 00:42:54.390 --> 00:42:57.819 the spurious phases are very easy to see, like this case. 00:42:57.819 --> 00:42:59.530 - Yeah. - But in other cases, 00:42:59.530 --> 00:43:03.560 they may not be so clear. So … - Yeah, this is … 00:43:03.560 --> 00:43:08.490 - How do we deal with that? - So the way we do it is really try to 00:43:08.490 --> 00:43:16.250 observe the slowness and time arrival and what Pierre Boué has also 00:43:16.250 --> 00:43:21.270 showed that mostly it will be – you can – it’s – they are 00:43:21.270 --> 00:43:24.450 more observable if you have long periods. 00:43:24.450 --> 00:43:27.660 But it’s true that it might generate some things weird. 00:43:27.660 --> 00:43:33.920 Like, for example, for the – for the ScS, Piero Poli and Michel Campillo 00:43:33.920 --> 00:43:39.040 and Maarten de Hoop are trying to work on explaining why this ScS 00:43:39.040 --> 00:43:42.880 is very strong when you have earthquakes. 00:43:42.880 --> 00:43:47.340 Because it means that this ScS is not really what we expect to reconstruct, 00:43:47.340 --> 00:43:50.800 but it’s another mechanism. And it’s a kind of spurious phase. 00:43:50.800 --> 00:43:55.290 I don’t know if they would call it that way, but it’s kind of the same idea. 00:43:55.290 --> 00:43:58.250 What we reconstruct is not what we are looking for. 00:43:58.250 --> 00:44:02.320 And so we really have to look for what it is to understand it. 00:44:02.320 --> 00:44:05.530 And then maybe later use it if we really understand it, but for now, 00:44:05.530 --> 00:44:08.830 we’re trying to understand the one that we know a bit more. 00:44:08.830 --> 00:44:13.000 But, yeah, they’re a bit – it’s a big subject being sure 00:44:13.000 --> 00:44:15.520 what we are looking at. 00:44:19.160 --> 00:44:23.640 - Well, if there aren’t any more questions, let’s thank our speaker again, and we’ll 00:44:23.640 --> 00:44:27.800 be going out to lunch at, let’s say, 11:30. - Thank you. 00:44:27.800 --> 00:44:33.060 [ Applause ]