WEBVTT Kind: captions Language: en 00:00:01.040 --> 00:00:03.870 Good morning, everyone. Welcome to seminar. 00:00:03.870 --> 00:00:09.320 Next week, our speaker will be Ralph Archuleta from UC-Santa Barbara. 00:00:10.020 --> 00:00:13.900 He will be speaking about earthquake moment-rate scaling. 00:00:13.900 --> 00:00:18.160 This week, it’s my pleasure to introduce Barbara Romanowicz, 00:00:18.160 --> 00:00:21.500 which I will give a brief introduction to. 00:00:22.240 --> 00:00:24.440 Barbara did her bachelor’s degree in math 00:00:24.450 --> 00:00:28.140 at the Ecole Normale Supérieure in Paris. 00:00:28.140 --> 00:00:31.890 She then did a master’s at Harvard in applied physics. 00:00:31.890 --> 00:00:35.180 And finally returned to Paris to get a doctorate in geophysics 00:00:35.180 --> 00:00:38.000 at the Université Paris 7. 00:00:38.000 --> 00:00:41.430 She then did a few postdocs before joining the faculty 00:00:41.430 --> 00:00:45.080 of the Institut de Physique du Globe in Paris, where she also 00:00:45.080 --> 00:00:48.690 became director of the Geoscope program. 00:00:48.690 --> 00:00:53.040 After a number of years there, she then came to Berkeley 00:00:53.040 --> 00:00:58.079 and served for 20 years as director of Berkeley Seismo Lab 00:00:58.079 --> 00:01:02.000 as well as being a professor of geophysics there. 00:01:02.000 --> 00:01:04.840 She also served a shorter term as department chair. 00:01:04.840 --> 00:01:08.960 More recently, she also has had a second faculty position in physics 00:01:08.960 --> 00:01:13.020 of the Earth’s interior at Collège de France. 00:01:13.020 --> 00:01:17.360 Throughout her career, she’s worked in a number of topics and physics 00:01:17.360 --> 00:01:27.799 of the deep Earth, earthquake modeling, and parameter estimation, as well as 00:01:27.799 --> 00:01:31.520 construction of seismic networks and laboratories. 00:01:31.520 --> 00:01:33.830 So today I guess we’re going to hear something about the first of 00:01:33.830 --> 00:01:38.660 those categories – mantle plumes and seismic tomography. 00:01:38.660 --> 00:01:40.240 Barbara? 00:01:42.560 --> 00:01:45.720 - Thank you very much, Belle, for this nice introduction. 00:01:45.720 --> 00:01:48.240 And good morning. I’m glad to be here. 00:01:48.240 --> 00:01:51.200 I’m seeing a lot of old friends, which is nice because I’ve lost 00:01:51.200 --> 00:01:55.790 a little bit touch with the earthquake community in the last few years. 00:01:55.790 --> 00:01:59.830 And so I’ll be talking to you about some of the recent work I’ve 00:01:59.830 --> 00:02:03.810 been involved with my students. And so, first of all, I’d like to 00:02:03.810 --> 00:02:06.620 acknowledge them because, without them, I wouldn’t be 00:02:06.620 --> 00:02:09.200 standing here talking to you about this. 00:02:09.200 --> 00:02:14.569 So my former students Ved Lekic, Sanne Cottaar, and Scott French, and current 00:02:14.569 --> 00:02:21.400 postdoc Kai Yuan, whose recent results I hope to be able to show you at the end. 00:02:21.400 --> 00:02:27.290 So just to situate the progress in global tomography, I’m going to be talking 00:02:27.290 --> 00:02:34.380 about global mantle structure as we determine it with earthquake waves. 00:02:34.380 --> 00:02:43.191 And this top one is the first complete upper mantle tomographic model 00:02:43.191 --> 00:02:47.450 of shear velocity that was developed with seismic waveforms 00:02:47.450 --> 00:02:50.870 in 1984 by Woodhouse and Dziewonski. 00:02:50.870 --> 00:02:56.019 And the bottom one is, well, almost 30 years later. 00:02:56.019 --> 00:03:01.889 The study – a recent study by Ritsema et al. 00:03:01.889 --> 00:03:08.519 This is to illustrate that, first of all, we do have higher resolution over the years. 00:03:08.519 --> 00:03:14.730 But also, that already, at that time, in 1984, most – the very important 00:03:14.730 --> 00:03:17.829 first-order features were present in these models. 00:03:17.829 --> 00:03:21.989 You can see the ridge system appear here as low velocities, 00:03:21.989 --> 00:03:26.769 hotter-than-average mantle at 100 kilometers’ depth. 00:03:26.769 --> 00:03:33.569 You can see the colder roots of the cratons of the old continents. 00:03:33.569 --> 00:03:37.669 You can see the back-arch system, and you can see also the age progression 00:03:37.669 --> 00:03:44.209 of the – this is going to be hard with the mouse – gosh – of the Pacific Ocean – 00:03:44.209 --> 00:03:48.209 the plate cooling with the colder and colder temperatures 00:03:48.209 --> 00:03:51.440 as you go away from the East Pacific rise. 00:03:51.440 --> 00:03:58.819 And so you – at 100 kilometers’ depth, the map really reflects plate tectonics. 00:03:58.820 --> 00:04:01.980 And now we see this in more details. 00:04:02.580 --> 00:04:07.639 And this is just to set also the stage for the whole mantle as we go deeper. 00:04:07.639 --> 00:04:12.339 This is also one of the recent global models – shear velocity models. 00:04:12.339 --> 00:04:17.880 As you go deeper, the signature of the tectonics progressively 00:04:17.880 --> 00:04:25.440 disappears at 600 kilometers’ depth. The map is dominated by the fast 00:04:25.440 --> 00:04:28.840 subduction zones in the Western Pacific and under South America, 00:04:28.840 --> 00:04:34.590 which you can better see here in cross- sections of tomographic models that 00:04:34.590 --> 00:04:41.880 were built with higher density, more regional data in the subduction zones. 00:04:41.880 --> 00:04:47.870 At 1,000 kilometers, that image disappears – that signature also 00:04:47.870 --> 00:04:51.130 disappears in most places except in several places where you can 00:04:51.130 --> 00:04:54.030 still see some of the subducting slabs, 00:04:54.030 --> 00:04:57.630 like in Fiji-Tonga and in northern South America. 00:04:57.630 --> 00:05:00.180 And by the time you get to the – near the core-mantle boundary 00:05:00.180 --> 00:05:04.410 here at 2,800 kilometers, the picture has completely changed. 00:05:04.410 --> 00:05:09.400 It has not much to do with what we saw related to 00:05:09.400 --> 00:05:13.120 tectonics at the top – in the top of the mantle. 00:05:13.120 --> 00:05:19.880 And it’s dominated by two very large regions of low shear velocity centered in 00:05:19.880 --> 00:05:23.350 the Pacific and – where is my mouse? 00:05:23.350 --> 00:05:26.350 [laughs] And in Africa. 00:05:26.350 --> 00:05:30.940 And surrounded by a ring of fast velocities, which is 00:05:30.940 --> 00:05:35.100 thought to be the – basically the graveyard of slabs, 00:05:35.100 --> 00:05:38.250 although we, of course, haven’t been there to check that. 00:05:38.250 --> 00:05:41.590 And the spectrum of heterogeneity on the right 00:05:41.590 --> 00:05:46.420 shows basically the wavelengths that we see as a function of depth. 00:05:46.420 --> 00:05:50.150 This is as a function of angular degree and historical harmonics expansion, 00:05:50.150 --> 00:05:54.170 but it means the – to the right, the higher the degree, the shorter the 00:05:54.170 --> 00:05:59.080 wavelength we’re looking at. And the strong energy is in the red. 00:05:59.080 --> 00:06:02.200 And you can see that, at the top and at the bottom of the mantle, 00:06:02.200 --> 00:06:07.150 the heterogeneity is organized in very long wavelengths – 00:06:07.150 --> 00:06:08.860 dominated by very long wavelengths. 00:06:08.860 --> 00:06:13.380 In fact, at the bottom of the mantle, it’s dominated by one single degree – degree 00:06:13.380 --> 00:06:17.910 two, which is plus/minus, plus/minus, which you can see in this – in this map. 00:06:17.910 --> 00:06:21.650 And then, in the middle of the mantle, it seems to be more like a 00:06:21.650 --> 00:06:28.870 white spectrum that is shorter wavelengths represented as well. 00:06:28.870 --> 00:06:34.860 So we’re – at the bottom of the mantle, these two large provinces have been 00:06:34.860 --> 00:06:39.090 named large low shear velocity provinces, which is a horrible name – 00:06:39.090 --> 00:06:44.170 so LLSVP in short. We’re still looking for a better name. 00:06:44.170 --> 00:06:49.260 But the map at the very bottom shows how well the different tomographic 00:06:49.260 --> 00:06:53.490 models that have been developed using different kinds of techniques, 00:06:53.490 --> 00:06:59.070 different sets of data by different groups. They do agree on the – on the structure. 00:06:59.070 --> 00:07:02.211 It’s showing where the – where all models agree that the 00:07:02.211 --> 00:07:08.360 velocity is lower than average in the red and higher than average in the blue. 00:07:08.360 --> 00:07:14.780 And you see the structures quite clearly, not only in the general long-wavelength 00:07:14.780 --> 00:07:18.180 shapes, but even the borders are quite well-described. 00:07:18.180 --> 00:07:25.300 And models do agree on the – some of the details in the – in those borders. 00:07:25.300 --> 00:07:32.240 So the large-scale structure of the whole mantle is now quite well-constrained, 00:07:32.240 --> 00:07:36.230 and there’s a lot of agreement between different groups. 00:07:36.230 --> 00:07:43.800 There are, however, some topics that are still debated and still worthy of trying to 00:07:43.800 --> 00:07:50.780 get a handle on. And one of them is the quest for the origin of hot spots. 00:07:50.780 --> 00:07:53.470 So this is Hawaii, which, of course, you know, 00:07:53.470 --> 00:07:59.030 with its age-progressive line of volcanoes. 00:07:59.030 --> 00:08:03.040 And right in the middle of the Pacific Plate, so why does it exist? 00:08:03.040 --> 00:08:07.460 It’s not really explained by the first-order plate tectonics. 00:08:07.460 --> 00:08:15.650 And Tuzo Wilson first suggested that these hot spots could be the 00:08:15.650 --> 00:08:21.500 surface manifestations of hot spot – of plumes of hot material grounded 00:08:21.500 --> 00:08:24.700 in some boundary layer in the Earth. 00:08:24.700 --> 00:08:29.500 And then Morgan, in 1971, suggested that this boundary layer 00:08:29.500 --> 00:08:31.970 might be actually the core-mantle boundary. 00:08:31.970 --> 00:08:40.389 Okay, now these hot spots definitely sample different reservoirs in the 00:08:40.389 --> 00:08:47.410 geochemistry sense, as is, for example, illustrated here in this – 00:08:47.410 --> 00:08:52.100 in this geochemical plot showing the difference in isotopic composition 00:08:52.100 --> 00:08:58.000 of some trace elements – in this case, neodymium and samarium – where you 00:08:58.000 --> 00:09:00.810 can distinguish the lavas of mid-ocean ridge basalts 00:09:00.810 --> 00:09:03.959 very clearly from the hot spot volcanoes. 00:09:03.959 --> 00:09:10.000 And which has led geochemists to propose that the hot spots 00:09:10.000 --> 00:09:15.000 sampled a deeper different reservoir than the mid-ocean ridges. 00:09:15.649 --> 00:09:22.720 The idea of hot spot plumes is also inspired by laboratory and 00:09:22.730 --> 00:09:28.089 numerical experiments which show that, if you heat a fluid from below 00:09:28.089 --> 00:09:33.529 and cool it from above, as a – as time progresses, which here, on the bottom, 00:09:33.529 --> 00:09:37.080 to the right, showing the temperature field in such a – 00:09:37.080 --> 00:09:42.399 this is a numerical experiment – you will – these plumes will appear 00:09:42.399 --> 00:09:46.839 and progressively rise to the top of the box. 00:09:46.839 --> 00:09:49.850 And characteristically, in these experiments, 00:09:49.850 --> 00:09:57.769 which are simple experiments – convective experiments, 00:09:57.769 --> 00:10:04.309 which usually constant viscosity and such, you see the morphology 00:10:04.309 --> 00:10:09.129 of these plumes is they have these narrow stems and big heads. 00:10:09.129 --> 00:10:15.869 And translated – or, scaled to the Earth mantle conditions, these stems 00:10:15.869 --> 00:10:19.920 would be rather narrow. They would be 100 to 200 kilometers wide. 00:10:19.920 --> 00:10:24.619 So they present a real challenge for global tomography for 00:10:24.619 --> 00:10:30.199 the question of resolution. Can we resolve these kind of conduits? 00:10:30.199 --> 00:10:35.339 So there have been a lot of attempts. And here is a slide that I borrowed 00:10:35.339 --> 00:10:44.899 from work by Guust Nolet showing some cross-sections across 00:10:44.899 --> 00:10:49.529 places that have hot spot volcanoes. This is from a model by Montelli et al. 00:10:49.529 --> 00:10:54.720 that was – attracted a lot of attention at the time. 00:10:54.720 --> 00:10:58.500 Because they proposed that they were actually seeing those plumes. 00:10:58.500 --> 00:11:07.019 And, indeed, you see, under these – under these different hot spots, 00:11:07.020 --> 00:11:15.020 you see some, you know, hot – red, hot anomalies that, if you are – 00:11:15.020 --> 00:11:22.040 you know, if you are kind of inclined to accept this idea of deep plumes, 00:11:22.050 --> 00:11:25.840 you might see them. But if you are a critic, you would say, 00:11:25.840 --> 00:11:31.059 yeah, but, you know, you have to – it’s a leap of faith to connect 00:11:31.059 --> 00:11:36.089 all these conduits. So the debate continues. 00:11:36.089 --> 00:11:40.519 And here is another one showing – so there was this plume experiment 00:11:40.519 --> 00:11:46.290 in Hawaii which installed stations on the ocean floor to get larger aperture 00:11:46.290 --> 00:11:49.619 than what you could get from just putting seismic stations 00:11:49.619 --> 00:11:53.740 on the islands and looking at travel time tomography. 00:11:53.740 --> 00:11:59.129 And this is the model of Wolfe et al., 2011, and these are two cross-sections. 00:11:59.129 --> 00:12:01.009 The two cross-sections shown on the right. 00:12:01.009 --> 00:12:06.769 The one northwest-southeast, the other one southeast – southwest-northeast. 00:12:06.769 --> 00:12:10.309 And you can see the – you know, the red anomaly here. 00:12:10.309 --> 00:12:14.230 The lower-than-average velocities, which we translate to first order 00:12:14.230 --> 00:12:16.639 as meaning hotter than average. 00:12:16.639 --> 00:12:23.000 And the there is a clear signature in the top 200 kilometers, but below that, 00:12:23.000 --> 00:12:29.959 you can see this trail of red going down to very deep. 00:12:29.959 --> 00:12:35.120 However, again, skeptics will say, well, in this kind of tomography 00:12:35.120 --> 00:12:39.740 program, what’s really important is to have illumination from different sides. 00:12:39.740 --> 00:12:44.309 If you want to resolve an anomaly, you have to be crossing the region 00:12:44.309 --> 00:12:46.720 that you want to resolve in many different directions 00:12:46.720 --> 00:12:48.800 in order to be able to resolve it. 00:12:48.800 --> 00:12:53.740 Otherwise, you might be smearing your anomaly along the ray paths. 00:12:53.740 --> 00:12:58.970 And so this geometry could be the effect of smearing along the paths. 00:12:58.970 --> 00:13:02.600 The box would indicate, you know, the region – tomographic region 00:13:02.600 --> 00:13:07.389 under Hawaii, but the rays originate outside of the box, and there are not – 00:13:07.389 --> 00:13:11.679 there’s not much constraint on what happens along the rest of the ray. 00:13:11.679 --> 00:13:15.809 So again, this is an issue – an issue which is – gets the 00:13:15.809 --> 00:13:20.730 right illumination from many sides in order to be able to 00:13:20.730 --> 00:13:24.249 constrain both the location and size of the anomalies. 00:13:24.249 --> 00:13:33.009 There is another – and of course, there is a lot of information in seismograms, 00:13:33.009 --> 00:13:36.550 which could be used to improve the situation. 00:13:36.550 --> 00:13:42.040 I’m just showing here a typical seismogram – teleseismic seismogram, 00:13:42.040 --> 00:13:46.629 which shows the arrival of the different main phases and the surface waves. 00:13:46.629 --> 00:13:52.089 And so far, most people have used only part of this seismogram 00:13:52.089 --> 00:13:55.819 to basically construct tomographic models. 00:13:55.819 --> 00:13:59.739 The parts that are well-separated in the seismogram – so the P wave arrival, 00:13:59.739 --> 00:14:03.050 the S wave arrival, sometimes SS, and then the surface wave train. 00:14:03.050 --> 00:14:08.400 But there’s many others that could be used to achieve the, 00:14:08.400 --> 00:14:12.509 you know, better illumination. This is just a series of diagrams 00:14:12.509 --> 00:14:16.369 borrowed from Ed Garnero’s web page showing how you could 00:14:16.369 --> 00:14:21.170 improve illumination. And this is just examples of some phases. 00:14:21.170 --> 00:14:25.149 But buried in these seismograms, there are many more that 00:14:25.149 --> 00:14:27.369 carry information on the structure. 00:14:27.369 --> 00:14:31.949 There is another issue, which is, if you just use travel times, 00:14:31.949 --> 00:14:36.720 you actually cannot really resolve narrow, low-velocity bodies. 00:14:36.720 --> 00:14:42.059 Because, when – if you think about it, if your source here is on the left, 00:14:42.059 --> 00:14:45.709 and you’re observing on the other side of the low-velocity body, 00:14:45.709 --> 00:14:49.959 the first arrival is actually not going to go through the anomaly. 00:14:49.959 --> 00:14:56.940 It’s going around it because it’s – the structure makes it go faster around it. 00:14:56.940 --> 00:15:00.110 And so, in order to see the low-velocity body, you have to 00:15:00.110 --> 00:15:05.980 look deeper into the waveforms, not only at the first arrival, 00:15:05.980 --> 00:15:11.589 in order to really, fully kind of be able to resolve it. 00:15:11.589 --> 00:15:17.080 And so this brings – basically this brings us to the 00:15:17.080 --> 00:15:22.139 necessity of really exploiting the waveforms – the seismic waveforms 00:15:22.139 --> 00:15:25.180 more completely than has been done before. 00:15:25.180 --> 00:15:29.100 This is just an example of showing that just the effect – 00:15:29.100 --> 00:15:34.540 maybe I should skip this, but that the effect of the – of the – 00:15:34.540 --> 00:15:38.990 I’m going to try to use my mouse here because it might be easier. 00:15:38.990 --> 00:15:43.389 No. Anyway, the effect of a synthetic plume – of a hypothetical 00:15:43.389 --> 00:15:47.110 plume here under Hawaii, on the waveforms, 00:15:47.110 --> 00:15:51.249 on the left we’re showing the seismograms without the 00:15:51.249 --> 00:15:53.720 plume and with the plume. With the plume is in red. 00:15:53.720 --> 00:15:58.809 And on the right, the blue traces are the different seismograms. 00:15:58.809 --> 00:16:02.439 And what I want to show here is that most of the effect is in the 00:16:02.439 --> 00:16:08.089 coda of the main – of the phases. It’s not in the first arrival. 00:16:08.089 --> 00:16:13.610 So to do that, we need to – we need to basically move to 00:16:13.610 --> 00:16:16.480 what’s called waveform tomography – full waveform tomography – 00:16:16.480 --> 00:16:21.000 the use of full waveforms to do tomography. 00:16:21.000 --> 00:16:24.519 And so, in that case, what you need to do is, of course, have a way 00:16:24.519 --> 00:16:29.269 to compute synthetics – to compute the wave field in a particular 00:16:29.269 --> 00:16:34.790 complicated Earth model and then – and then compare these synthetics 00:16:34.790 --> 00:16:41.899 to the data. And then, through an inverse process, explain the difference 00:16:41.899 --> 00:16:44.820 between the data and the synthetics to improve your model. 00:16:44.820 --> 00:16:48.860 And until about 10 years ago, we didn’t have that many ways 00:16:48.860 --> 00:16:55.119 of computing synthetics in a 3D mantle model in an accurate way. 00:16:55.119 --> 00:16:59.170 We had normal mode perturbation theory, which worked pretty well for – 00:16:59.170 --> 00:17:02.089 when we were interested in long wavelengths models. 00:17:02.089 --> 00:17:10.200 But in fact, as you go to stronger, more localized anomalies, 00:17:10.200 --> 00:17:16.780 this theory is no longer accurate. However, we are lucky that, 00:17:16.780 --> 00:17:23.570 in the meantime, computers have become much more – much faster 00:17:23.570 --> 00:17:27.320 and also methodologies have been developed to compute accurate wave 00:17:27.320 --> 00:17:32.830 fields in arbitrary 3D models. And the method of choice for this 00:17:32.830 --> 00:17:37.330 right now, for global seismology, is the so-called spectral element method. 00:17:37.330 --> 00:17:43.370 And so, I will not enter into the details. Just mention the challenges of this. 00:17:43.370 --> 00:17:47.550 The challenges are heavy computations because the computational cost increases 00:17:47.550 --> 00:17:52.200 as the cube of the frequency. Then also there are conditions – 00:17:52.200 --> 00:17:58.160 so Courant condition, which basically you have to decrease 00:17:58.160 --> 00:18:03.570 your mesh size and your time step when you – when your velocity decreases. 00:18:03.570 --> 00:18:08.320 So the crust becomes a problem with a lot of thin layers. 00:18:08.320 --> 00:18:12.220 And of course, for – to solve for a model, you need several hundred 00:18:12.220 --> 00:18:17.090 of events and a number of iterations because the problem is nonlinear. 00:18:17.090 --> 00:18:23.160 So it is very heavy computationally, however at least now the theory – 00:18:23.160 --> 00:18:27.800 the computed synthetics, we are pretty confident are very accurate. 00:18:27.800 --> 00:18:34.220 And so what I’m going to show is the results of our own attempts 00:18:34.220 --> 00:18:37.340 at implementing and taking advantage of these methodologies. 00:18:37.340 --> 00:18:41.510 And this is essentially the work of two of my former graduate students – 00:18:41.510 --> 00:18:46.280 Ved Lekic and Scott French. Progressively, we built – 00:18:46.280 --> 00:18:50.270 first, we built an upper mantle model, and then more recently, 00:18:50.270 --> 00:18:53.720 a whole mantle model, increasing the frequency 00:18:53.720 --> 00:19:00.670 to which we considered the waveforms in this – in this manner. 00:19:00.670 --> 00:19:05.860 And so I’m going to switch now – this was the background – to just the results. 00:19:05.860 --> 00:19:09.720 And actually, I’m going to start in the lower mantle and move my way up, 00:19:09.720 --> 00:19:14.860 even though, in reality, the models were constructed the other way around. 00:19:14.860 --> 00:19:18.590 So I’m going to show you first some cross-sections in the lower mantle 00:19:18.590 --> 00:19:21.870 through some hot spot plumes. 00:19:21.870 --> 00:19:24.690 Here you see two of them – MacDonald and Pitcairn. 00:19:24.690 --> 00:19:26.920 And this model is the one that I showed you at the beginning. 00:19:26.920 --> 00:19:32.800 It’s S40RTS, so Ritsema et al., built in 2011 using conventional 00:19:32.800 --> 00:19:37.180 tomographic models. So basically ray theory. 00:19:37.180 --> 00:19:43.540 And this is the model that we obtained in – well, just two years ago 00:19:43.540 --> 00:19:49.120 now, showing basically that the large-scale structure is really 00:19:49.120 --> 00:19:56.020 very compatible with S40RTS. But now, if you just looked at S40RTS, 00:19:56.020 --> 00:20:01.080 you could perhaps – if you’re a believer, you could perhaps connect the dots 00:20:01.080 --> 00:20:04.560 from the core-mantle boundary – this blob on the core-mantle boundary 00:20:04.560 --> 00:20:11.820 to the – to the – there we go – to the – to the hot spot. 00:20:11.820 --> 00:20:14.950 But again, you might say, well, there’s some yellow here, 00:20:14.950 --> 00:20:18.950 and there’s other blotches of yellow equally well elsewhere. 00:20:18.950 --> 00:20:22.740 So how – you know, why is this more significant? 00:20:22.740 --> 00:20:26.320 But now it’s become clear that you do have a connection between 00:20:26.320 --> 00:20:31.520 the lower-most mantle and the hot spot location. 00:20:31.520 --> 00:20:36.220 In fact, you can see that it’s slightly displaced. 00:20:36.220 --> 00:20:38.760 This happens at 1,000 kilometers – around 1,000 kilometers. 00:20:38.760 --> 00:20:40.640 I’m going to talk about it a little more. 00:20:40.640 --> 00:20:45.560 It’s actually already here in Ritsema’s model, but not as clear. 00:20:45.560 --> 00:20:49.170 And this is the other example of Pitcairn. 00:20:49.170 --> 00:20:55.780 And so we see this under many of the major hot spots. 00:20:55.780 --> 00:20:58.730 This is – on the top are the two that I already showed you, 00:20:58.730 --> 00:21:03.480 but on the bottom are two other cross-sections going through – 00:21:03.480 --> 00:21:06.860 well, on the right are the ones I showed you, and on the left, 00:21:06.860 --> 00:21:10.500 I’m looking at cross-section through Samoa here. 00:21:10.500 --> 00:21:13.730 And then one cross-section which shows three in a row – Samoa, Tahiti, 00:21:13.730 --> 00:21:19.360 and Marquesas, which you can see we could – we would hopefully 00:21:19.360 --> 00:21:22.330 can still improve the resolution in the future, but you can see 00:21:22.330 --> 00:21:28.830 quite well the – that they are – they look like they are separated. 00:21:28.830 --> 00:21:30.500 And this is in the Pacific Ocean. 00:21:30.500 --> 00:21:35.700 In the Atlantic Ocean, we also see these – such features. 00:21:35.700 --> 00:21:42.740 Here, the Cape Verde, which we kind of lose here, likely because of resolution. 00:21:42.750 --> 00:21:49.160 Canary Islands. Iceland. Iceland is a little bit less resolved, 00:21:49.160 --> 00:21:54.850 but you can see that, if you compare with whatever structure there is around 00:21:54.850 --> 00:22:01.370 it, it does stand out as the only feature of lower-than-average velocity here. 00:22:01.370 --> 00:22:06.720 And this is the well-known southern part of the African so-called 00:22:06.720 --> 00:22:12.790 super plume, or LLSVP, which is fatter. And actually, I think it’s probably 00:22:12.790 --> 00:22:17.430 made up of a large number of these conduits that are kind of 00:22:17.430 --> 00:22:19.530 glued together because of lack of resolution. 00:22:19.530 --> 00:22:25.090 But that is – that is just a, you know, suggestion for now. 00:22:25.090 --> 00:22:28.590 Yes, this was just to show that our resolution of Iceland 00:22:28.590 --> 00:22:33.160 is quite compatible with another study of the upper mantle using similar 00:22:33.160 --> 00:22:38.690 approaches, but just the regional study of the first 1,200 kilometers 00:22:38.690 --> 00:22:43.140 of the mantle by Rickers et al. that shows very clearly the same 00:22:43.140 --> 00:22:49.240 behavior with an anomaly that goes from Iceland down. 00:22:49.240 --> 00:22:51.640 And then around 1,000 kilometers, 00:22:51.640 --> 00:22:58.140 there’s some ponding, as it seems, around 1,000 kilometers. 00:22:58.150 --> 00:23:06.270 So, I mean, in summary, in our model, we see these structures under hot spots 00:23:06.270 --> 00:23:11.060 plumes for hot spot plume – for hot spots that are located 00:23:11.060 --> 00:23:14.570 above these large low shear velocity provinces. 00:23:14.570 --> 00:23:19.190 This map is showing the map at 2,800 kilometers of our model. 00:23:19.190 --> 00:23:25.310 And, on top of it, the different symbols – the different circles indicate where we 00:23:25.310 --> 00:23:31.170 clearly see a – you know, a conduit from the core mantle boundary 00:23:31.170 --> 00:23:37.100 to the vicinity of the hot spot, just like those that I showed you before. 00:23:37.100 --> 00:23:39.000 And then we have some gray ones, 00:23:39.000 --> 00:23:44.090 which are well-resolved but a little less strong. 00:23:44.090 --> 00:23:50.070 And some white ones that we think are there, but maybe we’re not as sure. 00:23:50.070 --> 00:23:55.480 What’s really striking is that any hot spot that is not above 00:23:55.480 --> 00:23:58.910 the LLSVPs, which is here indicated by a green circle – 00:23:58.910 --> 00:24:08.170 small green circle does not have a clear, broad conduit associated with it at the 00:24:08.170 --> 00:24:12.300 base of the mantle. This is, in particular, the case for Yellowstone. 00:24:12.300 --> 00:24:17.320 I should have pointed it out on the previous map. Maybe I can go back. 00:24:20.820 --> 00:24:22.980 This one, yes. On this one, you can see here 00:24:22.980 --> 00:24:26.110 on the lower left where Yellowstone is. 00:24:26.110 --> 00:24:30.040 And there is nothing – there looks like there is something in the upper mantle, 00:24:30.040 --> 00:24:35.150 but not necessarily – you cannot resolve anything in the lower-most mantle. 00:24:35.150 --> 00:24:41.430 And Yellowstone is sitting above the fast region – fast ring of – 00:24:41.430 --> 00:24:45.490 ring of fast velocities around the LLSVPs. 00:24:45.490 --> 00:24:52.190 So there seems to be a real association of these big plumes with – 00:24:52.190 --> 00:24:55.890 I will call them plumes. They’re really conduits, or chimneys, 00:24:55.890 --> 00:25:00.430 as someone suggested I should call them, 00:25:00.430 --> 00:25:04.310 that are over the large low shear velocity provinces. 00:25:04.310 --> 00:25:11.040 So now this is just to show that, if I filter this model to just the long wavelengths, 00:25:11.040 --> 00:25:17.010 this model is very compatible with what other models have seen. 00:25:17.010 --> 00:25:21.570 So again, the long wavelengths features are there. 00:25:21.570 --> 00:25:24.260 What we see are these large plumes, 00:25:24.260 --> 00:25:27.580 but – which are separated from one another. 00:25:27.580 --> 00:25:34.030 And we cannot – we have enough resolution that, if they were big – 00:25:34.030 --> 00:25:37.270 as some people have suggested that the LLSVPs go all the way up 00:25:37.270 --> 00:25:41.870 to throughout the upper mantle, in fact, we could resolve this. 00:25:41.870 --> 00:25:47.060 The top figure is putting an input synthetic model into our last step 00:25:47.060 --> 00:25:50.550 of our inversion, and the bottom is what we get from the inversion. 00:25:50.550 --> 00:25:54.870 And you can see that we wouldn’t break it up into several pieces. 00:25:54.870 --> 00:26:01.950 And same if – even if only the first 1,000 kilometers of the lower mantle 00:26:01.950 --> 00:26:06.840 were all a kind of uniform structure, we would also be able to recover that, 00:26:06.840 --> 00:26:10.180 so we’re pretty confident that they are separated. 00:26:10.180 --> 00:26:13.010 And now I’m going to walk you through some of the characteristics. 00:26:13.010 --> 00:26:16.330 So first, as you have noticed, these plumes are broad. 00:26:16.330 --> 00:26:21.870 They’re very broad. They’re probably more than 500 kilometers wide. 00:26:21.870 --> 00:26:26.310 They appear broader than that. They’re not as broad in reality, 00:26:26.310 --> 00:26:30.000 probably, because we do have limited resolution. 00:26:30.000 --> 00:26:35.370 And here is a test to show this. If we have a 400-kilometer input plume 00:26:35.370 --> 00:26:41.920 at the – at the – in the lower mantle, what we recover in this synthetic test, 00:26:41.920 --> 00:26:47.540 we recover it, but with a very much reduced amplitude 00:26:47.540 --> 00:26:52.910 and somewhat broadening. So in order to see 2% anomaly – 00:26:52.910 --> 00:26:56.620 lower anomaly in our models, we really need these plumes 00:26:56.620 --> 00:27:02.250 to be much stronger or wider. And which also indicates that they’re 00:27:02.250 --> 00:27:06.210 probably not just thermal plumes because this would indicate that they 00:27:06.210 --> 00:27:12.980 would have to be 1,500 degrees hotter than average – the average mantle, 00:27:12.980 --> 00:27:16.480 which seems quite excessive from our measurements – 00:27:16.480 --> 00:27:21.870 estimates that have been made of – so they’re probably thermochemical. 00:27:21.870 --> 00:27:25.130 And so this is just to illustrate what I mean by this. 00:27:25.130 --> 00:27:29.870 On the top, you have – it’s a geodynamic simulation with – 00:27:29.870 --> 00:27:33.630 starting with a dense layer at the base of the mantle, 00:27:33.630 --> 00:27:38.240 which has a higher density and different composition. 00:27:38.240 --> 00:27:42.970 And the right sides of these plumes show the temperature. 00:27:42.970 --> 00:27:46.560 The left sides – the white shows the composition. 00:27:46.560 --> 00:27:52.370 And if the layer is very thin, and the density contrast is very small, 00:27:52.370 --> 00:27:55.050 you see kind of traditional plumes. 00:27:55.050 --> 00:27:59.830 In order – the bottom figure shows the case where the dense layer 00:27:59.830 --> 00:28:06.020 is thicker and stronger density contrast with the ambient mantle. 00:28:06.020 --> 00:28:09.650 And then you start seeing plumes that have morphologies 00:28:09.650 --> 00:28:15.590 that look more like what we see. So that’s one sort of suggestion is the 00:28:15.590 --> 00:28:22.480 fact that we see them so broad indicates that there’s something in our view of the 00:28:22.480 --> 00:28:27.990 rheology and of the composition of the mantle that needs to be revised. 00:28:27.990 --> 00:28:33.530 The other really remarkable thing is that all of these plumes are 00:28:33.530 --> 00:28:37.900 rooted in these patches of really reduced low velocities. 00:28:37.900 --> 00:28:42.210 They’re kind of very strong velocity reduction 00:28:42.210 --> 00:28:47.300 in excess of 2% - 2, 3, 4% at the base of the mantle. 00:28:47.300 --> 00:28:55.970 And so this connects to work that we have been doing independently, 00:28:55.970 --> 00:29:00.360 and others have been too, of trying to model structure at the base of the 00:29:00.360 --> 00:29:07.420 mantle by doing forward modeling of waves that are sensitive to the 00:29:07.420 --> 00:29:11.420 so-called D-double prime region – the last 200 kilometers of the mantle. 00:29:11.420 --> 00:29:13.380 In our case, we’ve been using diffracted waves – 00:29:13.380 --> 00:29:15.960 waves that diffract on the core mantle boundary. 00:29:15.960 --> 00:29:20.280 And the green part of the path indicates which part of the 00:29:20.280 --> 00:29:25.340 lower-most mantle is sampled by this particular – as diffracted waves. 00:29:25.340 --> 00:29:29.480 And background is one of the tomographic models showing that 00:29:29.480 --> 00:29:35.410 we are sampling the region going into the Pacific LLSVP 00:29:35.410 --> 00:29:39.780 from the north to the east. And on the right is just – 00:29:39.780 --> 00:29:44.100 you can see very clearly these are the observed waveforms. 00:29:44.100 --> 00:29:47.810 They’re filtered between 10 and 30 seconds, I think, 00:29:47.810 --> 00:29:50.000 and – of the – as diffracted as a function of azimuth 00:29:50.000 --> 00:29:52.300 as you go from north to south. 00:29:52.300 --> 00:29:57.630 And you can see that the predictions from the tomographic model do not – 00:29:57.630 --> 00:30:02.190 you know, cannot predict this. Just the tomographic model itself, 00:30:02.190 --> 00:30:07.540 no matter which one, even our latest tomographic model doesn’t help. 00:30:07.540 --> 00:30:14.000 To reproduce the characteristics of these waveforms which are here indicated – 00:30:14.000 --> 00:30:18.860 the circle – red circle – the red ellipse shows the post-cursor. 00:30:18.860 --> 00:30:22.610 So the appearance of energy much later – 40, 50 seconds 00:30:22.610 --> 00:30:26.030 after the expected arrival time that is diffracted, 00:30:26.030 --> 00:30:30.360 while the amplitude of the main phase has been reduced. 00:30:30.360 --> 00:30:35.270 And in order to model – in order to reproduce this kind of structure, 00:30:35.270 --> 00:30:40.820 we have – we have had to introduce a very extreme structure. 00:30:40.820 --> 00:30:46.520 We modeled it simply to first order by a cylinder. 00:30:46.520 --> 00:30:51.350 And it’s basically what’s called an ultra-low velocity zone, 00:30:51.350 --> 00:30:55.650 which is very thin – 25 kilometers above the core-mantle boundary. 00:30:55.650 --> 00:30:59.850 Its diameter is about 800 kilometers, so it’s quite wide. 00:30:59.850 --> 00:31:02.830 And the shear velocity reduction is about 20%. 00:31:02.830 --> 00:31:05.110 And there are some tradeoffs between the height, the diameter, 00:31:05.110 --> 00:31:08.280 and the velocity reduction. Little bit on the position. 00:31:08.280 --> 00:31:14.840 But it’s really a lens effect that you have, shown here in the diagram. 00:31:14.840 --> 00:31:18.020 And that reproduces these observations quite well. 00:31:18.020 --> 00:31:23.620 And so this is – and this is sitting right at the bottom at the base 00:31:23.620 --> 00:31:29.490 of the Hawaii plume. A little bit – you can see it here. 00:31:29.490 --> 00:31:32.960 And then there is – you can see also the Samoa one, 00:31:32.960 --> 00:31:37.090 which actually was studied by another group – by Michael Thorne 00:31:37.090 --> 00:31:40.970 and his collaborators, and also found a large ultra-low velocity zone 00:31:40.970 --> 00:31:46.930 in that – under this – under this particular hot spot. 00:31:46.930 --> 00:31:52.200 And so what’s – I’m going to show you now, very quickly, 00:31:52.200 --> 00:31:59.200 results of recent work by my postdoc Kai Yuan on the Iceland plume, 00:31:59.200 --> 00:32:03.200 which we’re kind of excited about. This is not published yet. 00:32:03.200 --> 00:32:06.920 In this particular case, the illumination was available 00:32:06.920 --> 00:32:09.990 from many different sides, as seen on the right. 00:32:09.990 --> 00:32:13.210 These are still diffracted waves that are illuminating from many sides. 00:32:13.210 --> 00:32:18.050 In Hawaii, we only had one corridor going from Fiji-Tonga 00:32:18.050 --> 00:32:23.110 to North America, to the U.S. array network. Here we have many different 00:32:23.110 --> 00:32:26.720 directions, so we can look at this in many different directions. 00:32:26.720 --> 00:32:33.920 And again, if you look at the data, you see these features – similar features, 00:32:33.920 --> 00:32:38.850 kind of, where you see a post-cursor in a azimuth range. 00:32:38.850 --> 00:32:45.210 The pink dot is where the Hawaii – is around Hawaii. 00:32:45.210 --> 00:32:48.170 And disappearing – amplitude is decreasing 00:32:48.170 --> 00:32:53.660 the main phase, and post-cursor. And then there is also some 00:32:53.660 --> 00:32:59.080 features behind the – behind the main phase at larger azimuths. 00:32:59.080 --> 00:33:02.380 This is the predictions from our tomographic model. 00:33:02.380 --> 00:33:07.540 You can – you cannot see much. And in order to – and you can’t 00:33:07.540 --> 00:33:13.660 explain it by simply strengthening the amplitude of anomalies in the plume, 00:33:13.660 --> 00:33:16.840 say, because we lose amplitude by tomography, so – 00:33:16.840 --> 00:33:20.740 but this doesn’t explain these observations, either. 00:33:20.740 --> 00:33:24.260 It explains only part of them. However, if you introduce 00:33:24.260 --> 00:33:28.380 an ultra-low velocity zone, and in this particular case, also a cylinder of 00:33:28.380 --> 00:33:33.900 about 800 kilometers’ width, it’s a bit – it’s also about 20 kilometers thick. 00:33:33.900 --> 00:33:36.330 It’s very similar to the one in Hawaii. 00:33:36.330 --> 00:33:43.140 You can reproduce quite well these observations that we see in the data. 00:33:43.140 --> 00:33:47.020 What’s really exciting about it is that we put a very simple structure 00:33:47.020 --> 00:33:49.030 with a circular base. 00:33:49.030 --> 00:33:52.950 And in fact, this same structure, which was determined from 00:33:52.950 --> 00:33:59.481 forward modeling in this model, works also very well for 00:33:59.481 --> 00:34:02.980 other events, for other illuminations. So I’m going to go quickly, 00:34:02.980 --> 00:34:07.150 but anyway, the first column of the data shows the data. 00:34:07.150 --> 00:34:11.040 The 3D structure with the ultra-low velocity zones is the second one. 00:34:11.040 --> 00:34:13.669 And then the predictions from a tomographic model, 00:34:13.669 --> 00:34:16.830 which doesn’t show these features, is at the end. 00:34:16.830 --> 00:34:23.210 And so what’s really – is exciting is that this shows that in fact we have a 00:34:23.210 --> 00:34:28.330 circular base of these particular type of ultra-low velocity zone. 00:34:28.330 --> 00:34:33.360 And this is a very poor cartoon, but, you know, to kind of explain 00:34:33.360 --> 00:34:38.530 with exaggerated – vertically exaggerated, showing that the – 00:34:38.530 --> 00:34:42.629 that the fact that it’s circular indicates that it’s really under the plume – 00:34:42.629 --> 00:34:47.010 under the – you know, inside the plume indicates that this kind of 00:34:47.010 --> 00:34:49.200 ultra-low velocity zones are really related to the 00:34:49.200 --> 00:34:56.540 dynamics of the plume itself. And we’ve proposed that it’s 00:34:56.540 --> 00:35:04.820 basically a manifestation of an up-warping of some layer that has – 00:35:04.830 --> 00:35:09.600 maybe of a different composition, may have partial melting, or in any case, it 00:35:09.600 --> 00:35:14.680 would have low viscosity so that it could be shaped by the dynamics of the plume. 00:35:14.680 --> 00:35:20.740 Okay, how much time do I have left? I still have a little bit of time? Yes. 00:35:20.740 --> 00:35:27.780 So another characteristic of these – of these plumes, then, moving upward 00:35:27.790 --> 00:35:33.230 in the mantle is that they all – a lot of them seem to kind of either 00:35:33.230 --> 00:35:37.870 disappear or be deflected when you get to about 1,000 kilometers’ depth. 00:35:37.870 --> 00:35:43.600 So the broken lines are at depths of 400, 660, and 1,000 kilometers. 00:35:43.600 --> 00:35:48.720 So one expects things to happen at about 660 kilometers, but at 1,000 kilometers, 00:35:48.720 --> 00:35:55.540 there is, as yet, no, you know, indication of a global phase change, 00:35:55.540 --> 00:36:00.700 of a global transition in minerology that we know of. 00:36:00.700 --> 00:36:04.780 But these plumes deflect and maybe become narrower in the upper mantle, 00:36:04.780 --> 00:36:10.640 which is why we lose some of them. And just because we can’t resolve them. 00:36:10.640 --> 00:36:15.119 Also, this note here is to indicate that they are very straight 00:36:15.119 --> 00:36:17.840 in the lower mantle. They really rise quite vertically 00:36:17.840 --> 00:36:22.990 in the lower mantle, and they start deflecting only above 1,000 kilometers. 00:36:22.990 --> 00:36:26.240 So this is reminiscent of some simulations here – 00:36:26.240 --> 00:36:31.050 numerical simulations of convection where there is a viscosity change. 00:36:31.050 --> 00:36:35.570 In this particular case, the viscosity change was put at 660 because it’s a 00:36:35.570 --> 00:36:38.990 natural place to put it, with lower viscosity in the upper mantle, 00:36:38.990 --> 00:36:44.960 higher viscosity in the lower mantle. And you see these plumes rise and 00:36:44.960 --> 00:36:49.100 kind of pond at this – at this discontinuity in viscosity, 00:36:49.100 --> 00:36:55.530 and then become thinner as they rise through the lower viscosity regions. 00:36:55.530 --> 00:36:59.440 So this is really something we’re thinking about. 00:36:59.440 --> 00:37:06.150 And just to contrast this, or to remind you that there is also regions 00:37:06.150 --> 00:37:10.810 of fast velocities in subduction zones. And many of them pond. 00:37:10.810 --> 00:37:13.850 Many of these subduction zones are now – very clearly seem to pond 00:37:13.850 --> 00:37:17.450 around 660 kilometers, as in this example from Fukao 00:37:17.450 --> 00:37:21.680 and Obayashi’s model in northern Bonin. 00:37:21.680 --> 00:37:27.310 But there are other places where – like, for example, here in Java, 00:37:27.310 --> 00:37:32.330 where you can clearly see that the subduction zone actually goes through 00:37:32.330 --> 00:37:38.120 the 660 and ponds at deeper depths around 1,000 kilometers. 00:37:38.120 --> 00:37:42.820 Again, the bottom line here is 1,000 kilometers. 00:37:42.820 --> 00:37:48.410 And you can see that also in Fiji-Tonga quite clearly and 00:37:48.410 --> 00:37:53.020 in some places in North America. So there seems to be, both in fast 00:37:53.020 --> 00:37:55.460 velocity regions and in slow velocity regions, 00:37:55.460 --> 00:37:57.740 something going on at 1,000 kilometers. 00:37:57.740 --> 00:38:04.280 And in fact, if you go back and look at even – well, this is our model, 00:38:04.290 --> 00:38:08.800 but even if you look at – back at models from Harvard from 20 years ago, 00:38:08.800 --> 00:38:11.880 you actually can see this. You see a decorrelation 00:38:11.880 --> 00:38:15.830 of the structure between the lower mantle and the upper mantle. 00:38:15.830 --> 00:38:22.070 I don’t think I have time to go into the details. Just look at the left plots here. 00:38:22.070 --> 00:38:25.910 What it is is radial correlation functions – so correlating the 00:38:25.910 --> 00:38:31.000 structure layer by layer in depth and plotting the resulting correlation. 00:38:31.000 --> 00:38:33.470 The diagonal should be red because you’re correlating the 00:38:33.470 --> 00:38:36.900 same depth with the same depth. And you can see the lower mantle, 00:38:36.900 --> 00:38:45.810 which is this big square here – this big red square showing up in the left plots. 00:38:45.810 --> 00:38:49.960 That’s the lower mantle correlated with other lower mantle depths. 00:38:49.960 --> 00:38:52.330 So the structures seem to be well-correlated. 00:38:52.330 --> 00:38:54.670 This is also what we see with these chimneys going 00:38:54.670 --> 00:38:59.690 straight up in the lower mantle. And the blue that you see corresponds 00:38:59.690 --> 00:39:04.750 to depth ranges between – sort of the extended transition zone 00:39:04.750 --> 00:39:09.970 from about 400 to 1,000 kilometers. And you can see that this zone 00:39:09.970 --> 00:39:14.240 is decorrelated with the rest – with the rest of the lower mantle. 00:39:14.240 --> 00:39:19.760 So even at long periods, these are degrees – the long wavelengths – 00:39:19.760 --> 00:39:23.700 actually, it’s seen mostly at long wavelengths, you see this feature. 00:39:23.700 --> 00:39:27.820 And we only now kind of have realized it. 00:39:27.820 --> 00:39:32.190 Because of these higher resolution models, we can go back to 00:39:32.190 --> 00:39:35.390 the lower resolution models and see these new features. 00:39:35.390 --> 00:39:38.270 Okay. I’ll skip this. 00:39:38.270 --> 00:39:45.180 So this is – yes, so now I’m going to spend a little time – a few minutes 00:39:45.180 --> 00:39:48.390 on the upper mantle. And I’m going to focus on one region 00:39:48.390 --> 00:39:52.590 of the Pacific Ocean, which is the South Pacific Superswell, 00:39:52.590 --> 00:39:57.300 which has these, you know, hot spots – a concentration of hot spots – 00:39:57.300 --> 00:40:04.220 Samoa, Tahiti, Marquesas, and a few others, and then Hawaii to the north. 00:40:04.220 --> 00:40:06.100 And this – I showed you this picture already. 00:40:06.100 --> 00:40:07.970 We looked at the lower mantle, and now we’re going to look 00:40:07.970 --> 00:40:10.359 in the box in the upper mantle. 00:40:10.359 --> 00:40:16.590 And at the bottom here is a 3D rendering of this region seen from the south. 00:40:16.590 --> 00:40:23.500 And so we’re seeing – I’m actually going to show it – yeah, okay, sorry. 00:40:24.440 --> 00:40:27.600 We’re seeing it from the south, and we’re seeing the MacDonald- 00:40:27.610 --> 00:40:30.960 Society plume coming up. This is now in the upper 00:40:30.960 --> 00:40:34.550 1,000 kilometers, well-separated from the other conduits. 00:40:34.550 --> 00:40:36.920 Of course, there is vertical exaggeration. 00:40:36.920 --> 00:40:40.130 And you can see – at the top, you can see the low-velocity zone 00:40:40.130 --> 00:40:44.580 coming from the East Pacific Rise. So in the top 200 kilometers 00:40:44.580 --> 00:40:49.900 and kind of petering out as you go to older ocean ages. 00:40:49.900 --> 00:40:53.260 So what’s interesting, though, is that, below the – this is now 00:40:53.260 --> 00:40:59.550 a view from the east. You see these conduits going through. 00:40:59.550 --> 00:41:02.820 And you see the low-velocity zone up about, say – look at 100 kilometers, 00:41:02.820 --> 00:41:07.030 very – marked by very dark red, so very low velocities. 00:41:07.030 --> 00:41:12.930 And underneath it, you can see these sorts of corrugations that seem to – 00:41:12.930 --> 00:41:18.800 or drips that seem to go deeper – that seem to go to depths in extent of – 00:41:18.800 --> 00:41:22.210 excess of 300 kilometers. And they seem to be regular. 00:41:22.210 --> 00:41:26.080 In fact, they have a periodicity of about 1,000 kilometers. 00:41:26.080 --> 00:41:29.030 And they are aligned in the direction of plate motion, 00:41:29.030 --> 00:41:34.380 which is here, more or less perpendicular to the white board. 00:41:34.380 --> 00:41:39.640 And if we do – if we make cross-sections perpendicular 00:41:39.640 --> 00:41:43.460 to the plate motion – on the left is a map of the model at 00:41:43.460 --> 00:41:47.530 250 kilometers’ depth showing these elongated structures. 00:41:47.530 --> 00:41:50.550 And on the – and these are in relative velocities compared to 00:41:50.550 --> 00:41:55.640 the mean in that depth range. On the right, you see absolute velocities 00:41:55.640 --> 00:42:00.990 now in those three cross-sections – one, two, three – all of them perpendicular 00:42:00.990 --> 00:42:06.700 to the direction of plate motion of the Pacific Plate – absolute plate motion. 00:42:06.700 --> 00:42:10.430 And you can see these periodic features. 00:42:10.430 --> 00:42:15.270 And important thing is here that they do reach to greater depths. 00:42:15.270 --> 00:42:18.540 And the temperature – if you translate this into temperature, 00:42:18.540 --> 00:42:27.520 say 4.4 kilometers per second at 300 kilometers, is as low a velocity as in 00:42:27.520 --> 00:42:33.160 the low-velocity zone at 100 kilometers in between two of these features. 00:42:33.160 --> 00:42:37.590 And so this is an indication of really hotter-than-average material, 00:42:37.590 --> 00:42:42.650 which is therefore in a – you know, kind of organized periodically. 00:42:42.650 --> 00:42:48.740 And, in fact, if you look globally at the model at 250 kilometers’ depth, 00:42:48.740 --> 00:42:52.670 and you superimpose absolute plate motions, you see these 00:42:52.670 --> 00:42:58.030 elongated structures, not only in the Pacific, but essentially above all plates. 00:42:58.030 --> 00:43:04.609 And, well, I’ll walk you through what you see in the Indian Ocean here. 00:43:04.609 --> 00:43:11.830 The broken lines is absolute velocity of the – of the plate. 00:43:11.830 --> 00:43:16.080 So independent of the – what the model was developed. 00:43:16.080 --> 00:43:22.260 It’s just superimposed, you know, at the location of these fingers. 00:43:22.270 --> 00:43:26.920 And if you compare with the hot spot tracks, you can see that, of course, 00:43:26.920 --> 00:43:30.550 there’s the Reunion hot spot here. If you look at the hot spot track, 00:43:30.550 --> 00:43:38.830 it actually goes across the ridge and then turns towards the west part of India. 00:43:38.830 --> 00:43:45.859 Whereas, in fact, the finger that you see here in the tomographic model 00:43:45.859 --> 00:43:54.380 continues in – on the east side of India in the direction of the current plate motion. 00:43:54.380 --> 00:43:57.730 And so this is an indication that these features are really dynamic 00:43:57.730 --> 00:44:03.070 and represent the current dynamics in the upper mantle. 00:44:03.070 --> 00:44:07.150 And the same can be done for the Ninety East Ridge. 00:44:07.150 --> 00:44:10.150 And same thing in the Atlantic. 00:44:10.150 --> 00:44:18.420 So adding those absolute plate motion parallel broken lines and 00:44:18.420 --> 00:44:24.380 then associating them with some known features in Atlantic. 00:44:24.380 --> 00:44:30.630 So this is reminiscent of different things, but one of them is fingering. 00:44:30.630 --> 00:44:36.020 When you inject low-viscosity fluid – so coming from the plume, say, 00:44:36.020 --> 00:44:41.619 from those conduits, into a higher-viscosity channel – 00:44:41.619 --> 00:44:47.040 so the plumes would be very low viscosity, the channel is 00:44:47.040 --> 00:44:51.820 the channel in the upper mantle under the lithosphere. 00:44:51.820 --> 00:44:57.780 You see these flower patterns with irregular shape – with irregular spacing, 00:44:57.780 --> 00:45:02.440 which depends on the viscosity and on the thickness of the – 00:45:02.440 --> 00:45:04.330 of the layer into which you inject. 00:45:04.330 --> 00:45:13.340 And if you impose motion to the upper boundary, you see these petals, or these 00:45:13.340 --> 00:45:21.060 fingers, align in the direction and in the opposite direction to plate motion. 00:45:21.070 --> 00:45:25.870 So this is, you know, one thing that these features remind us of. 00:45:25.870 --> 00:45:30.560 They also remind us of secondary scale convection, such as proposed 00:45:30.560 --> 00:45:35.359 by Richter and Parsons many years ago – the so-called Richter rolls 00:45:35.359 --> 00:45:39.119 that people have been looking for for a long time. 00:45:39.119 --> 00:45:41.710 And so what we think is that what we’re seeing is sort of 00:45:41.710 --> 00:45:43.700 a combination of these things. 00:45:43.700 --> 00:45:54.580 A secondary scale convection governed by the plate motions and fed by the – by 00:45:54.580 --> 00:45:58.940 the conduits coming from the chimneys or plumes from the lower mantle. 00:45:58.940 --> 00:46:06.570 So, in summary, I think – well, first of all, mantle dynamics framework 00:46:06.570 --> 00:46:11.870 inspired by these tomographic results. We have a layering in the mantle, 00:46:11.870 --> 00:46:16.180 which is slightly – we’re going back to the idea of an extended 00:46:16.180 --> 00:46:20.430 transition zone that extends down to 1,000 kilometers. 00:46:20.430 --> 00:46:26.650 And the – of course, the top, which is the heterosphere, this is a term coined by 00:46:26.650 --> 00:46:31.840 Adam Dziewonski who passed away last year. So wanted to give him credit. 00:46:31.840 --> 00:46:37.700 And so you have these kind of layering – the heterosphere dominated by the 00:46:37.700 --> 00:46:43.320 tectonics, then the extended transition zone, which would have basically time 00:46:43.320 --> 00:46:48.490 scales – relatively vigorous convection time scales of, say, 30 million years. 00:46:48.490 --> 00:46:52.710 And then, from 1,000 kilometers to the base of the mantle is a zone 00:46:52.710 --> 00:46:58.720 which probably has very sluggish motions and basically 00:46:58.720 --> 00:47:03.560 much longer time scales in excess of 200 million years. 00:47:03.560 --> 00:47:09.650 And it’s an anchor to bundles of plumes which are clustered in the – 00:47:09.650 --> 00:47:14.400 within the large low shear wave velocity provinces. 00:47:14.400 --> 00:47:19.140 And they are – you know, inside them, we posit – we now have at least 00:47:19.140 --> 00:47:23.800 two examples, maybe three, that inside them, there are these 00:47:23.810 --> 00:47:29.900 ultra-low velocity zones of circle shape that we hope will provide further insight 00:47:29.900 --> 00:47:35.849 onto, you know, the dynamics and what is going on at the base of the mantle. 00:47:35.849 --> 00:47:39.950 And most likely, the communication between the extended transition zone 00:47:39.950 --> 00:47:44.390 and the lower mantle is episodic, as we see the slabs ponding, 00:47:44.390 --> 00:47:48.300 and some of the plume material also extending horizontally at 00:47:48.300 --> 00:47:53.580 that 1,000-kilometer level. And so we’re reviving the idea 00:47:53.590 --> 00:47:57.210 of flushing downwelling events, which was very popular 20 years ago, 00:47:57.210 --> 00:48:00.020 but somehow disappeared from the scene. 00:48:00.020 --> 00:48:03.960 So that’s it. Thank you. And I will take questions. 00:48:04.520 --> 00:48:11.600 [ Applause ] 00:48:15.040 --> 00:48:24.000 [ Silence ] 00:48:24.860 --> 00:48:29.900 - It’s really neat to be able to portray what’s going on 00:48:29.900 --> 00:48:35.110 in correlation with hot spots. 00:48:35.110 --> 00:48:38.130 But then it brings the basic question. 00:48:38.130 --> 00:48:46.099 You show what you might call sources at the core-mantle boundary. 00:48:46.099 --> 00:48:53.170 The basic physics question is, what propels, or gives them the momentum, 00:48:53.170 --> 00:48:59.380 to head toward a resistive surface? 00:48:59.380 --> 00:49:04.100 You did touch down on thermal 00:49:04.100 --> 00:49:10.750 chemical plumes and so forth, but also I happen to be in the field of gravity, 00:49:10.750 --> 00:49:15.880 so [chuckles] immediately come up with, is there a buoyancy involved. 00:49:15.880 --> 00:49:21.420 But then, even if – the buoyancy, you have to still have that narrow conduit. 00:49:21.420 --> 00:49:29.100 And what is the physical property that makes it head for the surface? 00:49:29.100 --> 00:49:32.430 - Well … - It has this thermal conductivity is – 00:49:32.430 --> 00:49:37.760 enters into it, and possibly density. 00:49:37.760 --> 00:49:39.650 And you might have a gravity map, possibly. 00:49:39.650 --> 00:49:44.140 That might correlate with some of the major things too. 00:49:44.140 --> 00:49:49.420 - So, of course, I mean, I don’t have a full answer to any of this. 00:49:49.420 --> 00:49:54.170 My thinking right now is that a lot of it is – 00:49:54.170 --> 00:49:57.980 that you have to invoke, you know, localization – 00:49:57.980 --> 00:50:01.060 strain localization of some kind. 00:50:01.060 --> 00:50:05.660 - Pardon. I don’t – I can’t hear that. - I’m saying that you have to 00:50:05.660 --> 00:50:09.530 invoke strain localization. The way I’m seeing the lower mantle 00:50:09.530 --> 00:50:17.990 is something very – you know, very high viscosity with only some regions that, 00:50:17.990 --> 00:50:24.170 for some reason, you know, allow – have been broken up and allow material 00:50:24.170 --> 00:50:30.690 to move upward with some strain localization kind of rheology, 00:50:30.690 --> 00:50:35.440 which we need to think about. As for the source, well, I mean, 00:50:35.440 --> 00:50:40.090 there’s the heat from the core, right? And there’s the debate – the very 00:50:40.090 --> 00:50:43.330 long-term debate about how much heat is coming from the core and how much 00:50:43.330 --> 00:50:48.590 is generated from radioactivity and … - Well, ultimately, of course, 00:50:48.590 --> 00:50:54.070 the heat comes from the extremely hot core, of course, and … 00:50:54.070 --> 00:50:55.070 - Right. 00:50:55.070 --> 00:51:01.220 - … the core-mantle boundary is important in things like changes 00:51:01.220 --> 00:51:07.152 in other symptoms like magnetic – changes in the magnetic field 00:51:07.152 --> 00:51:12.180 through many, many years, though. - Yeah, right. But, I mean, it is – 00:51:12.180 --> 00:51:19.660 you know, it’s hard to think that the main driver of these sources is not – 00:51:19.660 --> 00:51:23.480 is something else than the heat coming from the core. 00:51:24.920 --> 00:51:30.980 [ Silence ] 00:51:32.000 --> 00:51:36.720 - That was a really – a very clear summary of the state of the art. 00:51:36.730 --> 00:51:40.980 I’m really grateful that you came here to speak to us. 00:51:40.980 --> 00:51:46.579 Is velocity anisotropy important in this full-wave tomography? 00:51:46.579 --> 00:51:50.770 - Well, velocity anisotropy is important for the Earth because we know 00:51:50.770 --> 00:51:57.580 it’s present. It’s certainly present in the uppermost, at least, 200 kilometers. 00:51:57.580 --> 00:52:01.690 And so, in fact, this model is a – is a radially anisotropic model. 00:52:01.690 --> 00:52:06.920 I didn’t talk about it, but we have to include radial anisotropy to be able to 00:52:06.920 --> 00:52:10.359 match surface waves – you know, the Love waves and the Rayleigh waves. 00:52:10.359 --> 00:52:13.790 You can’t do it otherwise. So that part is taken into 00:52:13.790 --> 00:52:17.490 account throughout the mantle. But, you know, beyond that, 00:52:17.490 --> 00:52:20.630 you would want to also include azimuthal anisotropies, 00:52:20.630 --> 00:52:24.800 which also we know is present in the upper mantle. 00:52:24.800 --> 00:52:27.970 We haven’t included it here, and that is something – 00:52:27.970 --> 00:52:32.940 you know, it’s a work-in-progress. We – you know, we think we can 00:52:32.940 --> 00:52:37.700 get away with it because we have relatively good azimuthal coverage, 00:52:37.700 --> 00:52:41.930 so we average out this effect. But, of course, until we actually 00:52:41.930 --> 00:52:47.790 take it into account properly, we – you know, we could be a little biased. 00:52:47.790 --> 00:52:53.800 But, I mean, not more than anybody else so far, so … [laughter] 00:52:53.800 --> 00:52:55.380 - Thank you. 00:52:57.400 --> 00:53:02.580 - So how do you take into account attenuation in the forward model model? 00:53:02.580 --> 00:53:06.880 - Well, it’s taken into account in the same simulations. 00:53:06.880 --> 00:53:15.119 It’s one of those attenuation models that has a spring and dashpot, 00:53:15.120 --> 00:53:16.880 you know, kind of thing. 00:53:16.880 --> 00:53:21.710 - I mean, I could see it affecting the amplitude of some of those arrivals. 00:53:21.710 --> 00:53:25.270 - Well, so – I see. Oh, and that’s – in the inversion. 00:53:25.270 --> 00:53:31.290 So we have a 1D attenuation model. It’s a – it’s the kind of standard 1D 00:53:31.290 --> 00:53:38.820 average for the Earth, which fits, on average, the waveforms quite well. 00:53:38.820 --> 00:53:43.700 There is, of course, 3D variations in attenuation that actually we are 00:53:43.700 --> 00:53:48.950 working on right now to try and resolve. It’s a small effect, though. 00:53:48.950 --> 00:53:50.869 And it affects the amplitudes. 00:53:50.869 --> 00:53:54.420 And most of the fit here is still in the phase. 00:53:54.420 --> 00:53:57.680 We have to fit the phase before we fit the amplitude. 00:53:59.100 --> 00:54:03.780 - So – [clears throat] excuse me – so the LLSVPs are … 00:54:03.780 --> 00:54:06.109 - Good. [chuckles] - … seem to be predominantly 00:54:06.109 --> 00:54:09.490 in the southern hemisphere? - No. They are actually – if you filter 00:54:09.490 --> 00:54:12.750 them to the longest wavelength, they’re kind of centered on the equator. 00:54:12.750 --> 00:54:14.790 - Okay. - They’re antipodal. 00:54:14.790 --> 00:54:20.030 And so they – you know, one idea is that they have – they may have 00:54:20.030 --> 00:54:25.160 been there for a long time, since maybe the magma ocean, 00:54:25.160 --> 00:54:30.910 because rotation of the Earth may have played a role in positioning them. 00:54:30.910 --> 00:54:36.700 And with present-day, you know, rheology for the mantle as we know it, 00:54:36.700 --> 00:54:40.070 that’s unlikely that they would move due to rotation. 00:54:40.070 --> 00:54:47.330 But you can conceive it that, you know, in the past, if the – if the viscosity 00:54:47.330 --> 00:54:52.460 was much lower, they could have been positioned like this antipodally. 00:54:52.460 --> 00:54:56.369 Because that’s a position which corresponds to the kind of 00:54:56.369 --> 00:55:01.900 stable energy for the moments of inertia of the Earth. 00:55:06.440 --> 00:55:13.520 - What’s the volume balance between the LLSVPs and the – your big plumes? 00:55:13.520 --> 00:55:18.839 - Oh. So I haven’t done that calculation. The LLSVPs themselves – you know, 00:55:18.839 --> 00:55:25.020 if you just take them as the surface area and maybe continue them a little bit into 00:55:25.020 --> 00:55:29.880 the – into the mantle, there are different estimates, but it’s no more than about 00:55:29.880 --> 00:55:35.359 7% in volume of the whole volume. So it’s a small – so it would be – 00:55:35.359 --> 00:55:41.200 I think, in fact, the LLSVPs is just the base – the very base. 00:55:41.200 --> 00:55:46.609 And the rest, you know, coming up into the mantle, it’s a bundle of plumes. 00:55:46.609 --> 00:55:50.200 So basically it’s the – all of it is in the plumes. 00:55:50.200 --> 00:55:54.840 - What I’m getting as is, how quickly do you use it up if you have big plumes? 00:55:54.840 --> 00:55:59.420 - Ah. [chuckles] I don’t – well, yeah, that’s the big question. 00:56:00.340 --> 00:56:04.160 You know, we only have a present snapshot, right? 00:56:04.160 --> 00:56:07.650 And then I think you have to look at how – you know, 00:56:07.650 --> 00:56:09.520 how active volcanoes have been. 00:56:09.520 --> 00:56:15.950 So, you know, if you look at Hawaii, it’s still active, right? 00:56:15.950 --> 00:56:21.830 So it’s probably at least 100 million years, right? [chuckles] Yeah. 00:56:21.830 --> 00:56:26.130 So you can do a lot of – I mean, you can follow up and see whether 00:56:26.130 --> 00:56:29.040 that gives you more constraints on what’s going on, but … 00:56:29.040 --> 00:56:35.080 - I have a less speculative question. And I’m sure you’ve thought about this. 00:56:35.080 --> 00:56:41.620 The volumes of ocean island magmas are quite small on Earth. 00:56:41.620 --> 00:56:47.960 And the isotopic characteristics of mid- ocean ridge basalts, although they have – 00:56:47.960 --> 00:56:54.080 they have influences of plume end members, they’re generally quite low. 00:56:54.080 --> 00:56:59.630 So if you have big plumes, they’re either not getting into the upper mantle, 00:56:59.630 --> 00:57:04.430 which would be the source of mid-ocean ridge basalts, or the 00:57:04.430 --> 00:57:09.110 material that’s getting into the mantle is – the upper mantle is so refractory 00:57:09.110 --> 00:57:13.600 that it’s not melting and it’s not contributing to mid-ocean ridge basalts. 00:57:13.600 --> 00:57:18.240 Do you have a sense as to which of those it is? 00:57:18.250 --> 00:57:21.420 - Not really. You know, that’s outside of my breadth. 00:57:21.420 --> 00:57:25.380 But you bring up a point, you know, from the discussion we had before. 00:57:25.380 --> 00:57:30.130 What is the source of these – you know, of the heat or of the buoyancy. 00:57:30.130 --> 00:57:35.170 You could have – I mean, you could conceivably be tapping this reservoir 00:57:35.170 --> 00:57:41.020 at the base of the mantle that would be, you know, high concentrations of these 00:57:41.020 --> 00:57:46.770 trace radioactive elements, you know. But answering your question further, 00:57:46.770 --> 00:57:51.120 I don’t – I mean, I don’t – I wouldn’t feel comfortable [chuckles] going there. 00:57:52.060 --> 00:57:54.869 Hi, Mark. - Hi. [chuckles] 00:57:54.869 --> 00:57:58.840 Most of the plumes you imaged are under the oceans. 00:57:58.840 --> 00:58:01.580 And you had difficulty imaging something under Yellowstone, 00:58:01.580 --> 00:58:03.340 and you wanted to comment … 00:58:03.340 --> 00:58:06.109 - Well, no. So they’re not all under the ocean. 00:58:06.109 --> 00:58:09.780 They’re under the ocean in the Pacific, but under Africa, they’re under Africa. 00:58:09.780 --> 00:58:11.500 - Right. - Right? 00:58:11.500 --> 00:58:13.800 Or they’re also around, right, in the Atlantic. 00:58:13.800 --> 00:58:16.580 But the – some of them are under Africa. 00:58:16.580 --> 00:58:21.790 Yellowstone is – happens to be outside of the LLSVPs. 00:58:21.790 --> 00:58:25.410 And it should be very well, you know, resolved because 00:58:25.410 --> 00:58:29.690 we have a lot of stations and a lot of data there. 00:58:29.690 --> 00:58:32.330 And I’m not saying that there are no – that this is 00:58:32.330 --> 00:58:36.390 the only type of plumes we have. There may be others that are much 00:58:36.390 --> 00:58:41.010 thinner, you know, different kind, you know, or have been active 00:58:41.010 --> 00:58:44.720 in the past but are – you know, are dead now or something, 00:58:44.720 --> 00:58:49.520 which we don’t resolve yet. And that could be the case of 00:58:49.520 --> 00:58:53.380 Yellowstone, but it would be very – you know, if you look very carefully 00:58:53.380 --> 00:58:57.500 at the model, you can see some yellow kind of little thing inside it. 00:58:57.500 --> 00:59:01.619 But it’s nothing compared to the big ones that we see under – 00:59:01.619 --> 00:59:05.840 over the LLSVPs. And we have – you know, 00:59:05.840 --> 00:59:11.180 we have at least as good resolution, if not better, under Yellowstone. 00:59:11.180 --> 00:59:14.710 We’re looking into that right now. Because actually, this segues – 00:59:14.710 --> 00:59:20.610 if I have – if I have – oh, no, I don’t have – no, sorry. 00:59:20.610 --> 00:59:24.421 This is just much – but what we’re doing now is trying to – you know, 00:59:24.421 --> 00:59:28.580 to increase resolution, it’s going to be very difficult at the global scale. 00:59:28.580 --> 00:59:32.520 Because it’s just very heavy computationally, and we don’t have the 00:59:32.520 --> 00:59:36.840 illumination everywhere to be able to get the same resolution everywhere. 00:59:36.840 --> 00:59:42.550 So we’re designing a method where we can put a box around some region, 00:59:42.550 --> 00:59:48.089 and we’re actually trying it out in Yellowstone, exactly, to basically do 00:59:48.089 --> 00:59:53.660 only one full wave field computation at high frequency teleseismically, 00:59:53.660 --> 00:59:58.720 and then constrain the following iterations of the inversion inside the box. 00:59:58.720 --> 01:00:03.480 It represents some technical issues, which we think we have resolved. 01:00:03.480 --> 01:00:07.560 Because if you have stations and sources outside of the box, 01:00:07.560 --> 01:00:12.920 you have to match the wave field very precisely at the box boundaries. 01:00:12.920 --> 01:00:16.880 So you have to use concepts of time reversal and mirrors 01:00:16.880 --> 01:00:20.610 and that sort of thing. But it’s feasible, and it will 01:00:20.610 --> 01:00:23.869 make it possible to go to much higher resolution in regions 01:00:23.869 --> 01:00:30.360 where we think we can get it by going to higher frequencies. 01:00:32.420 --> 01:00:41.180 [ Silence ] 01:00:42.360 --> 01:00:44.940 - Any other questions for Barbara? 01:00:47.280 --> 01:00:50.060 Okay, we’re going to take Barbara to lunch, so let’s meet 01:00:50.070 --> 01:00:54.270 at the flagpole at 11:45. And meanwhile, let’s thank 01:00:54.270 --> 01:00:57.650 our speaker and say happy birthday. - Oh, thank you. [laughs] 01:00:57.650 --> 01:01:02.180 [ Applause ] 01:01:02.180 --> 01:01:08.120 [ Silence ]