WEBVTT Kind: captions Language: en-GB 00:00:02.429 --> 00:00:04.819 … for joining us for today’s ESC seminar. 00:00:04.819 --> 00:00:08.750 A couple of announcements first. On Friday at 11:00, we have 00:00:08.750 --> 00:00:11.800 an all-hands meeting. And our own Tamara Jeppson will 00:00:11.800 --> 00:00:16.840 give us next week’s ESC seminar. A few reminders. 00:00:16.840 --> 00:00:21.590 Please disconnect your VPN and make sure that your microphone is muted. 00:00:21.590 --> 00:00:26.140 If you’d like to enter full-screen mode to better see the speakers' slides 00:00:26.140 --> 00:00:30.860 by clicking on the three dots for more actions on the menu bar. 00:00:30.860 --> 00:00:35.660 And please do send us suggestions for more seminar speakers for the summer. 00:00:35.660 --> 00:00:40.120 And, with that, I’ll pass it on to Annemarie to introduce today’s speaker. 00:00:42.560 --> 00:00:45.900 [Silence] 00:00:45.900 --> 00:00:50.420 - Wait, wait, wait. - Annemarie, we can’t hear you. 00:00:50.420 --> 00:00:52.700 - I was muted. [laughs] Sorry. 00:00:52.700 --> 00:00:56.579 So, thanks, Noha, for – and Kathryn for organizing, 00:00:56.579 --> 00:01:00.580 especially during this pandemic. Today’s speaker is Lucia Gualtieri, 00:01:00.580 --> 00:01:03.110 who is a professor at Stanford University. 00:01:03.110 --> 00:01:06.280 And, as far as I can tell, my qualifications for introducing 00:01:06.280 --> 00:01:10.170 Lucia are that I spent some time at Stanford. 00:01:10.170 --> 00:01:14.720 So we’re really excited that you are here at Stanford in the Bay Area. 00:01:14.720 --> 00:01:20.450 And hopefully we can get together in person when this is all over. 00:01:20.450 --> 00:01:23.170 So Lucia has been at Stanford since October, and her group is called 00:01:23.170 --> 00:01:26.370 the Environmental and Computationl Seismology Group. 00:01:26.370 --> 00:01:30.200 So she received a bachelor’s and master’s in physics from the University 00:01:30.200 --> 00:01:34.810 of Bologna in Italy and then went on to get a dual Ph.D. both from the 00:01:34.810 --> 00:01:39.040 University of Bologna and the Institut de Physique du Globe de Paris in France. 00:01:39.040 --> 00:01:43.200 After that, she moved a little bit west across the pond – became a postdoctoral 00:01:43.200 --> 00:01:47.340 fellow at Lamont-Doherty and then also did a postdoc at Princeton 00:01:47.340 --> 00:01:53.470 before continuing the true west and out to Stanford this past fall. 00:01:53.470 --> 00:01:57.210 So she has been widely recognized for her work, including the Laura Bassi 00:01:57.210 --> 00:02:00.540 Young Scientist Award from the Italian Physical Society, 00:02:00.540 --> 00:02:04.300 the Aki Scientist Award from AGU, the Blavatnik Postdoctoral Award 00:02:04.300 --> 00:02:06.800 for Young Scientists from the New York Academy of Scientists. 00:02:06.800 --> 00:02:09.759 So that’s a very impressive list. 00:02:09.759 --> 00:02:12.609 Her research interests are in solving problems related to emerging fields 00:02:12.609 --> 00:02:16.549 in seismology, such as ambient seismic noise; signals due to mass wasting 00:02:16.549 --> 00:02:20.099 events; interaction between atmosphere, ocean, and solid Earth; signals from 00:02:20.099 --> 00:02:25.239 rock falls and landfalls – landslides; and also she’s also interested in 00:02:25.239 --> 00:02:28.099 tomography, Earth structure, and other inverse problems. 00:02:28.099 --> 00:02:32.540 So today she will talk to us about turning noise into signal and listening 00:02:32.540 --> 00:02:36.020 to the environment with seismic waves. Thanks. 00:02:36.660 --> 00:02:40.239 - Thank you. Thank you for this very nice introduction. 00:02:40.239 --> 00:02:46.059 And thanks for joining this virtual seminar today. 00:02:46.059 --> 00:02:51.830 So today I’m going to tell you about what we record other than earthquakes 00:02:51.830 --> 00:02:55.629 on seismic records – so everything that comes from the environment 00:02:55.629 --> 00:03:00.900 and how we can use those signals for studying the environment itself. 00:03:00.900 --> 00:03:05.249 So let’s start from the very beginning. Let’s start from seismograms. 00:03:05.249 --> 00:03:09.819 So what do we record on Earth? So here you see a month of – 00:03:09.819 --> 00:03:15.599 a seismogram in Massachusetts. So you have days here on the – 00:03:15.599 --> 00:03:21.829 on the Y axis and hours here. So each row is 24 hours of record. 00:03:21.829 --> 00:03:25.650 So you see there are two major earthquakes magnitude larger than 7. 00:03:25.650 --> 00:03:29.349 And you see also that, all the time, you record a very small, tiny 00:03:29.349 --> 00:03:33.610 vibration going on. When you get something like this – 00:03:33.610 --> 00:03:38.569 so a flat seismogram – you know that your station is not working anymore. 00:03:38.569 --> 00:03:42.979 So we expect to have what we call noise from the seismic records. 00:03:42.979 --> 00:03:50.200 So we historically define signal as everything that is desired in the data. 00:03:50.200 --> 00:03:53.180 And we call noise everything that we don’t want and 00:03:53.180 --> 00:03:55.860 we want to discard to look at the signal. 00:03:55.860 --> 00:03:58.930 But I hope to convince you by the end of this talk that the noise 00:03:58.930 --> 00:04:03.000 that we have in the records is actually a signal itself. 00:04:03.460 --> 00:04:06.260 So let’s look carefully at this seismogram. 00:04:06.269 --> 00:04:10.359 If you look at the beginning of the month here, you’ll see that the amplitude 00:04:10.359 --> 00:04:16.310 of the noise is actually much larger than during the rest of the month. 00:04:16.310 --> 00:04:20.880 And this is due to the fact that, at that moment, during those days, 00:04:20.880 --> 00:04:24.630 in the Atlantic Ocean, there was a large hurricane going on. 00:04:24.630 --> 00:04:28.389 And here you see all the days, one after the other. You see that the amplitude 00:04:28.389 --> 00:04:34.600 is increasing and then decreasing as the hurricane basically died. 00:04:34.600 --> 00:04:38.290 So, if you look carefully, and you zoom in the middle of the month, 00:04:38.290 --> 00:04:42.680 you’ll see there is another interesting signal coming from the environment. 00:04:42.680 --> 00:04:48.280 This is the signal of a large, massive landslide that occurred in Alaska. 00:04:48.280 --> 00:04:51.630 And you see there was an earthquake before and then the signal 00:04:51.630 --> 00:04:56.139 of the landslide here. So in the seismic record, we have a lot 00:04:56.139 --> 00:05:00.830 of signal coming from the environment. And the other examples are those. 00:05:00.830 --> 00:05:04.590 On the left, you see the signal associated to a tornado. 00:05:04.590 --> 00:05:11.880 So this is the Transportable Array in the U.S., and this is a tornado that actually 00:05:11.880 --> 00:05:16.940 was moving and forming 2 kilometers away from one station. 00:05:16.949 --> 00:05:20.530 When you look at the records, you’ll see that there is an increase in amplitude. 00:05:20.530 --> 00:05:24.441 You have a clear signal both in terms of up-and-down movement 00:05:24.441 --> 00:05:27.980 of the ground but also in terms of pressure. 00:05:28.480 --> 00:05:31.500 On the right, you have, instead, the signature of the 00:05:31.510 --> 00:05:35.509 sediment transport along rivers. So here we are in Nepal. 00:05:35.509 --> 00:05:38.400 We are close to one of the major rivers. 00:05:38.400 --> 00:05:42.710 There was one seismic station there. And here you have the spectrogram 00:05:42.710 --> 00:05:47.729 which means frequency on the Y axis as a function of time. 00:05:47.729 --> 00:05:50.370 And the color represents the power spectral density. 00:05:50.370 --> 00:05:53.880 So what you see here is that, during the summer months, 00:05:53.880 --> 00:05:57.380 you have a large patch of energy around 1 hertz. 00:05:57.380 --> 00:06:00.129 You can see the same just looking at seismograms. 00:06:00.129 --> 00:06:05.999 So here you have the black seismograms, which is – was recorded 00:06:05.999 --> 00:06:10.830 in July of that year, and the gray seismogram recorded in January. 00:06:10.830 --> 00:06:14.600 You see a big difference in terms of amplitude. 00:06:14.600 --> 00:06:19.340 So, in this region, this is the region where the monsoon happens. 00:06:19.340 --> 00:06:22.219 And so that’s why you have this large patch. 00:06:22.219 --> 00:06:25.020 Because, during summer, you have a lot of rain, 00:06:25.020 --> 00:06:30.960 and so the water in the – in the river moves so much that basically 00:06:30.960 --> 00:06:34.139 it can transport the sediments along the river bed. 00:06:34.140 --> 00:06:40.420 And all those sediments impacting the ground can generate this loud signal. 00:06:40.420 --> 00:06:44.420 Other environmental signals, and peculiar signals, 00:06:44.420 --> 00:06:49.180 occur in the polar regions are called glacial earthquakes. 00:06:49.180 --> 00:06:55.460 So those are – signal like those, you see on the bottom, are featured 00:06:55.460 --> 00:07:01.530 in two different frequency bands and are due to the calving of glaciers. 00:07:01.530 --> 00:07:07.529 So basically, the ice breaks apart and generate forces 00:07:07.529 --> 00:07:12.620 that can actually lead to the generation of seismic waves. 00:07:12.620 --> 00:07:15.259 And those are pretty different respect to tectonic earthquakes. 00:07:15.259 --> 00:07:18.999 So you can see that, for example, looking at the histograms for the number 00:07:18.999 --> 00:07:25.419 of those events as a function of month here – so the green histogram is the 00:07:25.420 --> 00:07:28.920 one associated with glacial earthquakes. And you see that those are pretty 00:07:28.920 --> 00:07:33.000 seasonal. There is a peak in summer due to the fact that the temperature 00:07:33.009 --> 00:07:37.400 is much higher in summer. And so the ice melts and breaks 00:07:37.400 --> 00:07:42.630 much easier than – easier than during the wintertime. 00:07:42.630 --> 00:07:48.380 On the other hand, you can also see that, over the years, this green histogram 00:07:48.380 --> 00:07:52.440 is basically increasing. And this is the sign, probably – although 00:07:52.440 --> 00:07:59.560 the time series here is pretty short, it could be the sign of global warming. 00:07:59.570 --> 00:08:03.009 So we have many signal coming from the environment, but why it is 00:08:03.009 --> 00:08:06.669 interesting to study those events? Well, the first motivation I can 00:08:06.669 --> 00:08:12.210 give you is that this is really the major part of the data we have – 00:08:12.210 --> 00:08:15.100 and we record with seismometers on Earth. 00:08:15.100 --> 00:08:21.860 As I show you with this seismogram, it’s really the dominant amount of data. 00:08:21.860 --> 00:08:27.639 And, on the other hand, people showed that it is possible to monitor the quality 00:08:27.639 --> 00:08:31.419 and the performance of a seismic station during the signal. 00:08:31.419 --> 00:08:36.060 And, in particular, I’m referring to – and will go into this – to the 00:08:36.060 --> 00:08:42.510 signal generated by the ocean. If you take long time series of this 00:08:42.510 --> 00:08:45.910 quasi-random vibration, and you compute the power spectral density 00:08:45.910 --> 00:08:49.930 as you see here on the right, you’ll see basically this peculiar shape. 00:08:49.930 --> 00:08:59.040 We’ll go into this later on in my talk. And this peculiar shape is basically – 00:08:59.040 --> 00:09:02.060 there are some characteristics that remain the same and monitoring 00:09:02.060 --> 00:09:05.990 over time those characteristics so you can be sure that your station 00:09:05.990 --> 00:09:09.840 is actually healthy and working as usual. 00:09:09.840 --> 00:09:14.050 The other motivation I can give you is that this is really a unique data set 00:09:14.050 --> 00:09:19.190 in the sense that it [inaudible] information of different systems – 00:09:19.190 --> 00:09:23.780 different Earth systems. So basically, let’s make the case, 00:09:23.780 --> 00:09:28.529 for example, of landslide or signal generated by a hurricane. 00:09:28.529 --> 00:09:35.279 In this case, you have, like, the atmosphere generating the big hurricane. 00:09:35.279 --> 00:09:41.250 Then the hurricane affects ocean waves. And then, for some mechanism I will 00:09:41.250 --> 00:09:44.870 explain, you’ll get seismic waves. So you have, really, the interaction 00:09:44.870 --> 00:09:50.579 between atmosphere, ocean, solid Earth, that is basically in those data. 00:09:50.579 --> 00:09:55.260 It’s somewhere hidden in those data. And so extracting those features 00:09:55.260 --> 00:09:59.460 from this data set can be very useful to monitor the environment. 00:09:59.460 --> 00:10:07.400 And, in fact, in the past, has been recognized that studying the process that 00:10:07.400 --> 00:10:13.480 govern the interaction between different Earth systems is really one of the great 00:10:13.480 --> 00:10:20.839 challenges that, as a seismologist, we should care about for the future. 00:10:20.839 --> 00:10:24.990 The third motivation I want to give you is mostly solid Earth-driven motivation, 00:10:24.990 --> 00:10:32.070 and is the fact that, years ago – a few decades ago – a couple of decades ago, 00:10:32.070 --> 00:10:37.480 people showed that using this continuum quasi-random vibration, it is possible 00:10:37.480 --> 00:10:42.690 to monitor the solid Earth to realize tomographies of the solid Earth. 00:10:42.690 --> 00:10:46.680 So we have been doing with earthquakes for a long time. 00:10:46.680 --> 00:10:51.150 And the advantage of also using this is that you have this signal 00:10:51.150 --> 00:10:53.570 going on all the time. You have seen that basically 00:10:53.570 --> 00:10:57.650 the major part of the seismogram is composed by this signal. 00:10:57.650 --> 00:11:01.040 And so what you can do – and you see this here on the bottom – 00:11:01.040 --> 00:11:07.399 you can fill the gaps of the classical earthquake-based tomography. 00:11:07.400 --> 00:11:12.780 So we can improve our knowledge of the solid Earth itself using those signals. 00:11:12.780 --> 00:11:18.820 So, in my talk today, I will talk about two major fields in what is today 00:11:18.829 --> 00:11:22.090 called environmental seismology. So the first one is about mass 00:11:22.090 --> 00:11:27.399 wasting events, so mostly landslides. And the second one is about ocean and 00:11:27.400 --> 00:11:33.140 atmospheric activity. And I will make the case of tropical cyclones. 00:11:35.880 --> 00:11:41.400 So let’s start with mass wasting events. So mass wasting events are rockfall, 00:11:41.400 --> 00:11:46.840 landslide, and other phenomena – complex phenomena that involve a 00:11:46.850 --> 00:11:56.050 variety of materials – rocks, but also other – can be mud, can be a lot of dust, 00:11:56.050 --> 00:12:01.310 small particles, debris, and so on. So different sizes in – 00:12:01.310 --> 00:12:03.720 that slides down in a very complex way. 00:12:03.720 --> 00:12:07.450 So it is very difficult to describe with a simple physical model 00:12:07.450 --> 00:12:11.970 those phenomena. And what complicates things here 00:12:11.970 --> 00:12:15.770 is that, most of the times, those event happens in very remote 00:12:15.770 --> 00:12:20.150 places like mountains. You can see here some examples. 00:12:20.150 --> 00:12:23.160 This is the Hudson River in New York. 00:12:23.160 --> 00:12:28.930 You can see a very big rockfall along the Palisades. 00:12:28.930 --> 00:12:32.019 Here you have another example in the Yosemite National Park and 00:12:32.019 --> 00:12:37.269 here in the Zion National Park. So one rockfall basically blocking 00:12:37.269 --> 00:12:42.860 completely the road. And so those phenomena are also very dangerous. 00:12:42.860 --> 00:12:46.230 And if you look around the world, the number of fatalities associated 00:12:46.230 --> 00:12:51.200 with those events, you’ll see there is – there are places on Earth where those 00:12:51.200 --> 00:12:55.550 events are very, very dangerous. And one example I want to make 00:12:55.550 --> 00:12:59.250 is this big blob here in Taiwan. 00:12:59.250 --> 00:13:05.920 So basically, in 2009, in summer, one big landslide happened in the 00:13:05.920 --> 00:13:12.579 mountains where there was a village. And you can see here the picture of the 00:13:12.579 --> 00:13:17.350 village before and after the landslide. You see that basically the village 00:13:17.350 --> 00:13:22.720 was completely destroyed. And this – the worst thing of this event 00:13:22.720 --> 00:13:27.459 was that basically it took some time before the government and people 00:13:27.459 --> 00:13:33.850 outside the village realized what happened in that place. 00:13:33.850 --> 00:13:37.790 If you look in seismic records, you’ll see that we have an equivalent 00:13:37.790 --> 00:13:42.399 of magnitude 5 earthquake. This is an example of a seismogram 00:13:42.399 --> 00:13:49.339 200 kilometers away, and you see that this signal is really big compared to 00:13:49.339 --> 00:13:53.889 the ground noise signal. So this signal really stands up and, 00:13:53.889 --> 00:13:58.690 as you can imagine, this signal is basically recorded in near real time. 00:13:58.690 --> 00:14:03.959 So having a way to recognize that that one was a landslide and 00:14:03.959 --> 00:14:07.700 where this landslide happened could actually – 00:14:07.700 --> 00:14:11.889 could have made the difference in that case. 00:14:11.889 --> 00:14:17.700 So another example is pretty recent in – basically a year and a half ago, 00:14:17.700 --> 00:14:22.870 there was a big tsunami in Indonesia. And here you see an image. 00:14:22.870 --> 00:14:30.839 And the – basically, the screen shot of the GFZ web page in Potsdam 00:14:30.839 --> 00:14:34.790 where they released the information that this was 00:14:34.790 --> 00:14:38.570 initially thought as a tectonic earthquake. 00:14:38.570 --> 00:14:42.209 But the signal was pretty strange in the sense that there was 00:14:42.209 --> 00:14:45.890 no high-frequency signal. Was only at low frequency. 00:14:45.890 --> 00:14:50.399 And it’s pretty strange in the sense that, for example, you don’t get any P waves. 00:14:50.399 --> 00:14:55.170 You have this beautiful surface waves coming up without 00:14:55.170 --> 00:15:00.600 any visible body waves. And actually, after days, 00:15:00.600 --> 00:15:03.460 people realized that this was not a tectonic earthquake. 00:15:03.460 --> 00:15:08.860 It was a submarine landslide. And here you can see that the flank 00:15:08.860 --> 00:15:13.640 of the Krakatau volcano before the event and after the event. 00:15:13.640 --> 00:15:16.920 So, after the event, was basically missing totally. 00:15:16.920 --> 00:15:23.320 So the flank went down and caused the tsunami tha struck Indonesia. 00:15:23.320 --> 00:15:26.959 So this is another example. I actually took this picture 00:15:26.959 --> 00:15:30.649 during my summer holidays. This is a rockfall. 00:15:30.649 --> 00:15:35.930 And when I saw all those debris and rocks, I said, okay, this should 00:15:35.930 --> 00:15:40.339 be the sign of a rockfall. I went back to the office and look at 00:15:40.339 --> 00:15:47.600 the data, and in fact, you can see two beautiful signal there, really apart. 00:15:47.600 --> 00:15:55.820 So, like, you had two different collapse of this flank here. 00:15:55.820 --> 00:16:02.879 So can we recognize if this signal is really a rockfall or it is a landslide 00:16:02.879 --> 00:16:07.250 or it is a tectonic earthquake? So the signal looks pretty different. 00:16:07.250 --> 00:16:12.100 So here is an example of displacement as a function of time for a rockfall, 00:16:12.100 --> 00:16:16.459 a landslide, and an earthquake. All signal filtered in the same 00:16:16.459 --> 00:16:21.379 frequency band. So you can see that the main difference between a rockfall, 00:16:21.380 --> 00:16:25.000 landslide, and earthquake is about the body waves. 00:16:25.000 --> 00:16:28.459 So body waves emerge pretty clearly for an earthquake. 00:16:28.459 --> 00:16:33.430 They don’t for a landslide and a rockfall. And the rockfall and landslide have this 00:16:33.430 --> 00:16:37.959 very gentle onset at the beginning, while, for an earthquake, you have this 00:16:37.959 --> 00:16:45.100 standard arrival of the surface waves. So now we can ask ourselves how 00:16:45.100 --> 00:16:49.950 those rockfalls, landslide, those event can generate seismic waves. 00:16:49.950 --> 00:16:57.129 So the best way, I think, for visualizing and have an idea of what happened 00:16:57.129 --> 00:17:02.750 during those event is to make the comparison with a roller coaster. 00:17:02.750 --> 00:17:07.550 So a landslide is like a roller coaster. And when you are sitting over 00:17:07.550 --> 00:17:10.830 a roller coaster is like you are sitting over a landslide. 00:17:10.830 --> 00:17:16.740 So what happen is that, basically, the Earth exerts on you – on the 00:17:16.740 --> 00:17:22.720 roller coaster – a set of forces. Those forces are gravity, centripetal 00:17:22.730 --> 00:17:28.030 forces, can be friction, and so on. But each one of those forces for 00:17:28.030 --> 00:17:31.700 Newton’s third law, basically they have a reactive 00:17:31.700 --> 00:17:36.070 counterpart – so a force going in this opposite direction – that one. 00:17:36.070 --> 00:17:40.320 So this red arrow is what the Earth exerts on you. 00:17:40.320 --> 00:17:46.900 This blue arrow is what you can exert on the Earth as a reaction of the forces 00:17:46.900 --> 00:17:53.000 that the Earth are exerting. So this is all referred to the center of mass. 00:17:53.000 --> 00:17:56.800 So we are talking about [inaudible] refer to the center of mass here. 00:17:56.800 --> 00:18:01.930 And so this blue arrow here is basically a force, or in the real world, 00:18:01.930 --> 00:18:05.780 a set of forces, that are applied at the surface of the Earth. 00:18:05.780 --> 00:18:10.330 And so every force applied at the surface of the Earth can generate seismic waves. 00:18:10.330 --> 00:18:14.960 So that’s why you record seismic waves due to those events. 00:18:15.540 --> 00:18:20.580 So, to show you what we can do with this, I make the case of a large, massive 00:18:20.580 --> 00:18:26.860 landslide that occurred in 2015 in Alaska in this very remote place. 00:18:26.860 --> 00:18:33.470 This occurred in October, and people were able to go there only months later, 00:18:33.470 --> 00:18:39.540 in August the year after. Because this is a very remote place, and 00:18:39.540 --> 00:18:44.510 there was snow everywhere at that time. So what happened in this case was 00:18:44.510 --> 00:18:49.870 that the flank of this mountain went down into the fjord. 00:18:49.870 --> 00:18:52.960 And so we are at the terminus of a glacier. 00:18:52.960 --> 00:18:57.810 You see the glacier here. And here it start the fjords. 00:18:57.810 --> 00:19:04.810 And so the mass went down, and so what we can – what happened 00:19:04.810 --> 00:19:10.120 was actually this massive landslide. But also, this landslide generated 00:19:10.120 --> 00:19:14.130 a tsunami, and the buoy farther away in the icy bay recorded 00:19:14.130 --> 00:19:21.480 a signature of this tsunami. This event was going unnoticed, 00:19:21.480 --> 00:19:24.790 if it wasn’t for seismology, in the sense that, in the framework 00:19:24.790 --> 00:19:31.530 of the CMT catalog, people recorded a signal that was not associated 00:19:31.530 --> 00:19:36.550 with tectonic earthquake, was a weird signal, and so looking carefully and 00:19:36.550 --> 00:19:40.750 waiting for satellite images in the sense that it was cloudy during 00:19:40.750 --> 00:19:45.660 the day of the – of the event. And actually, for the two weeks later – 00:19:45.660 --> 00:19:49.920 so, after two weeks, people were able to associate that 00:19:49.920 --> 00:19:53.850 those seismic waves were this landslide. 00:19:53.850 --> 00:19:58.120 So the seismic waves of the seismogram looks like this. 00:19:58.120 --> 00:20:04.430 So this is the displacement as a function of time at 17 kilometers away. 00:20:04.430 --> 00:20:07.200 This is the vertical component, and this is the spectrogram. 00:20:07.200 --> 00:20:11.000 Frequency as a function of time, a power spectral density in color. 00:20:11.000 --> 00:20:15.780 So you see that, as I showed before, you have no body waves – 00:20:15.780 --> 00:20:20.800 clear body waves. You start to have surface waves immediately. 00:20:20.800 --> 00:20:25.060 You can see, actually, the imprint of a small – actually, not a small – 00:20:25.060 --> 00:20:31.100 it was a magnitude 4 earthquake, or 4.5, I think, was in central Alaska, 00:20:31.100 --> 00:20:34.420 so not that far away from the landslide. 00:20:34.970 --> 00:20:40.600 So this landslide could be – could have been triggered by this earthquake. 00:20:40.610 --> 00:20:44.090 But also we have to say that, during the 48 hours before the event, 00:20:44.090 --> 00:20:48.120 there was a massive amount of rain in that region. 00:20:48.120 --> 00:20:52.560 And that region is actually famous for having landslides 00:20:52.560 --> 00:20:58.310 from time to time also due to the – to the amount of rain. 00:20:58.310 --> 00:21:02.430 So what we can do with this – where, in seismology, we are – 00:21:02.430 --> 00:21:07.460 usually we describe a seismogram as composed by two main terms. 00:21:07.460 --> 00:21:10.440 One is the source – in this case, a force history. 00:21:10.440 --> 00:21:15.330 So here – for an earthquake, you would have the moment tensor here. 00:21:15.330 --> 00:21:20.930 We are talking about forces, as I show you in the roller coaster slide. 00:21:20.930 --> 00:21:24.960 This force history, convolved with the Green’s function, 00:21:24.960 --> 00:21:26.420 makes the seismogram. 00:21:26.420 --> 00:21:29.710 What’s the Green’s function? Green’s function basically describes the 00:21:29.710 --> 00:21:34.330 propagation of seismic waves into the Earth, so is the response to an impulse. 00:21:34.330 --> 00:21:39.950 And so you have here the – basically the component due to 00:21:39.950 --> 00:21:43.560 the propagation of seismic waves between the source and the receiver. 00:21:43.560 --> 00:21:46.790 So we have the seismograms. We can compute Green’s functions 00:21:46.790 --> 00:21:51.360 in seismology because we have tools and computers. 00:21:51.360 --> 00:21:54.410 And so we can – and we have also models of the Earth – 00:21:54.410 --> 00:21:59.880 reliable models of the Earth, so we can compute the response to an impulse. 00:21:59.880 --> 00:22:02.790 And actually, what we don’t know is the force history. 00:22:02.790 --> 00:22:06.000 So we can actually retrieve this – try to understand what’s the 00:22:06.000 --> 00:22:07.700 source of a landslide. 00:22:07.700 --> 00:22:12.920 We can invert for the source, and we can do this in a simple way solving the 00:22:12.920 --> 00:22:16.080 least square problem in the frequency domain because it’s much easier. 00:22:16.090 --> 00:22:20.060 This convolution here became simple product. 00:22:20.060 --> 00:22:24.300 So we – as I said, we have u – our seismogram, or better, the spectrum – 00:22:24.300 --> 00:22:28.980 the power spectral density or the spectral amplitude of the seismogram. 00:22:28.980 --> 00:22:32.820 And we can compute the spectral amplitude for the Green’s function. 00:22:32.820 --> 00:22:39.060 So, for the Green’s function, in this case, I used a recent tool in seismology, 00:22:39.060 --> 00:22:43.500 which is called Instaseis, which means instantaneous seismogram. 00:22:43.500 --> 00:22:47.750 So basically it’s a catalog of Green’s function, and you can pull 00:22:47.750 --> 00:22:51.720 the Green’s function you want in between your source and your receiver. 00:22:51.720 --> 00:22:55.560 The advantage of using this is that it is nearly instantaneous. 00:22:55.560 --> 00:22:59.380 So thinking about, in the long term, thinking that something like this could 00:22:59.380 --> 00:23:05.280 be applied in a near real time, this could be a very interesting tool to be used. 00:23:05.280 --> 00:23:12.540 And those catalogs are accurate down to two seconds and so allows to really 00:23:12.550 --> 00:23:18.070 look at the very broadband signal. So solving this equation, we can 00:23:18.070 --> 00:23:23.600 compute the power spectrum – the spectral amplitude of the – of the source. 00:23:23.600 --> 00:23:27.720 And this is the normalized spectral amplitude as a function of period for 00:23:27.720 --> 00:23:33.290 the three forces – the vertical force, the north – the north-south component 00:23:33.290 --> 00:23:37.240 of the force, and the east-west component of the force. 00:23:37.240 --> 00:23:42.500 What you can note here is that there is a corner frequency, or a corner period, 00:23:42.500 --> 00:23:49.680 around 100 seconds. And 100 seconds is about the duration of the event. 00:23:49.680 --> 00:23:54.170 You can notice that there is a falloff as the period squared at shorter periods 00:23:54.170 --> 00:23:58.020 and a falloff as 1 over the period at longer periods. 00:23:58.020 --> 00:24:01.700 This reminds something that is being done for earthquakes and has been 00:24:01.700 --> 00:24:06.880 known for a long time – the spectrum of the source of an earthquake. 00:24:06.880 --> 00:24:11.370 So if you look at this, which is basically, again, the spectral amplitude as 00:24:11.370 --> 00:24:15.570 a function of period, but for an earthquake, you see that we have 00:24:15.570 --> 00:24:20.410 the same falloff for earthquake and the landslide at short period 00:24:20.410 --> 00:24:25.860 while the two events are pretty different for the long periods. 00:24:25.860 --> 00:24:29.980 And so this is interesting because other than looking at seismograms 00:24:29.980 --> 00:24:34.880 and trying to differentiate between those events and an earthquake, 00:24:34.880 --> 00:24:39.680 this could be another way to discriminate between 00:24:39.680 --> 00:24:44.510 mass wasting events and a regular tectonic event. 00:24:44.510 --> 00:24:48.580 So now let’s go back to the time domain, so from the amplitude 00:24:48.580 --> 00:24:52.960 of the spectrum, we go back to the force as a function of time. 00:24:52.960 --> 00:24:56.550 So nothing happened before the landslide, but nothing happened 00:24:56.550 --> 00:25:01.480 after the landslide. And you have forces – positive and negative means – 00:25:01.480 --> 00:25:04.820 for example, for the vertical, it means up and down, 00:25:04.820 --> 00:25:08.370 and so on for the other two components, during the event. 00:25:08.370 --> 00:25:14.350 Now, if this force that we computed is reliable, we can check that, going back 00:25:14.350 --> 00:25:19.930 to the data, and comparing what we get as synthetic seismograms with those – 00:25:19.930 --> 00:25:25.980 with this forces story and compare those synthetic seismograms with data. 00:25:25.980 --> 00:25:32.760 So here you have, for five stations around the event, the north component, 00:25:32.760 --> 00:25:36.400 east component, and vertical component – so you have 00:25:36.400 --> 00:25:42.170 data in red and synthetics in blue. And so you see that they match pretty well. 00:25:42.170 --> 00:25:47.120 So this means that we are able to explain, with this force history, the data. 00:25:47.120 --> 00:25:50.860 Now, what we can do with this. So we have the force history 00:25:50.860 --> 00:25:53.510 as a function of time. Now we know that, if we 00:25:53.510 --> 00:25:58.000 double-integrate, and we scale by the mass, we can obtain displacement. 00:25:58.000 --> 00:26:02.101 This minus here is because, remember, this is the reaction to the force. 00:26:02.101 --> 00:26:06.300 The force we are considering here is the opposite one, respect to the 00:26:06.300 --> 00:26:11.400 one the Earth exerts on you when you are on the roller coaster. 00:26:11.400 --> 00:26:16.800 So this simple equation, which is – accounts from Newton’s second law, 00:26:16.800 --> 00:26:22.480 is basically – tells you that having that force history allows you to reconstruct, 00:26:22.480 --> 00:26:28.620 really, the displacement of the center of mass of the landslide. 00:26:28.630 --> 00:26:34.380 So here we are considering, as the mass was concentrated in the center of mass. 00:26:34.380 --> 00:26:38.580 So we can reconstruct the displacement as a function of time. 00:26:38.580 --> 00:26:42.050 But there is a problem here because the force is an unknown 00:26:42.050 --> 00:26:45.790 is the one we just computed, the mass is also an unknown. 00:26:45.790 --> 00:26:50.110 So you see here that, if you change the mass, you change, 00:26:50.110 --> 00:26:54.550 pretty much, the displacement and the final runout here. 00:26:54.550 --> 00:26:59.300 So what we can do here to constrain this and to solve this problem – 00:26:59.300 --> 00:27:05.450 one equation, two unknowns – is get some help from satellite images. 00:27:05.450 --> 00:27:09.190 This satellite image shows you the source region. 00:27:09.190 --> 00:27:13.990 Here you can see pretty well the region where this landslide happened. 00:27:13.990 --> 00:27:16.870 You see also the scarp of an old landslide. 00:27:16.870 --> 00:27:19.380 You can see the color here is pretty different. 00:27:19.380 --> 00:27:21.650 This is new. This is old. 00:27:21.650 --> 00:27:24.700 And then you can see where is the deposit zone. 00:27:24.700 --> 00:27:33.560 So what you can do is to compute this displacement as in the previous slide and 00:27:33.560 --> 00:27:40.500 compute the runout distance from this image and constrain this runout distance. 00:27:40.500 --> 00:27:45.350 So the runout distance here is about 900 meters. And so we can 00:27:45.350 --> 00:27:51.600 compute the displacement. This series of dots here shows you 00:27:51.600 --> 00:27:55.510 the 3D displacement of the center of mass when you 00:27:55.510 --> 00:27:59.450 constrain the final runout to 900 meters. 00:27:59.450 --> 00:28:03.580 And so, going back to that previous equation, this landslide 00:28:03.580 --> 00:28:06.780 we can estimate – we can get an estimate of the mass. 00:28:06.780 --> 00:28:10.690 And the mass in this case was 150 million tons – 00:28:10.690 --> 00:28:14.550 25 Great Pyramids of Giza. So really a massive landslide. 00:28:14.550 --> 00:28:20.540 So that’s why was so visible and was basically seen worldwide 00:28:20.540 --> 00:28:23.100 in the – in the seismic data. 00:28:23.100 --> 00:28:28.020 So now, is this reliable? Can we check again if this trajectory is – 00:28:28.020 --> 00:28:35.730 actually fits with some data? We can actually try to see what’s 00:28:35.730 --> 00:28:39.150 the fit between this trajectory and the topography. 00:28:39.150 --> 00:28:44.710 And you can see here the elevation as a function of distance along the profile. 00:28:44.710 --> 00:28:50.010 You can see that we have a mismatch of a few meters between the two. 00:28:50.010 --> 00:28:54.900 So this actually, trajectory of the center of mass is pretty reliable. 00:28:54.900 --> 00:28:59.880 So, to summarize this first part, I show you that we have ways to 00:28:59.880 --> 00:29:05.440 distinguish, but we’re still working on it. We need to find ways to distinguish 00:29:05.440 --> 00:29:10.120 even more and in near real time, signal due to earthquakes 00:29:10.130 --> 00:29:13.460 and signal due to mass wasting events. 00:29:13.460 --> 00:29:17.780 We are able nowadays to model the force history, and with this force history, 00:29:17.780 --> 00:29:22.890 we can invert for characteristics of the event, like the mass and the runout and 00:29:22.890 --> 00:29:27.460 the trajectory of the center of mass. And, in this case, we need – 00:29:27.460 --> 00:29:30.760 to estimate the mass – satellite images to constrain. 00:29:30.760 --> 00:29:36.750 And so this framework I show you could be actually extended, and this is 00:29:36.750 --> 00:29:43.040 what I’m doing, to basically build some near real-time identification 00:29:43.040 --> 00:29:47.250 of landslide in the seismic records. And this could be done using – 00:29:47.250 --> 00:29:51.340 integrating different techniques, like machine learning techniques, forward 00:29:51.340 --> 00:29:56.990 seismic modeling, remote sensing, and the geological analysis to find out 00:29:56.990 --> 00:30:01.650 the events in the seismic records, infer the dynamics and the location 00:30:01.650 --> 00:30:05.500 of the event, estimate mass and characteristics of the event, 00:30:05.500 --> 00:30:09.600 and then compare with geological findings. 00:30:10.910 --> 00:30:15.000 So now I’m moving from land to the ocean, basically. 00:30:15.000 --> 00:30:18.520 And I’m going to tell you about the seismology due to 00:30:18.520 --> 00:30:20.960 the ocean and atmospheric activity. 00:30:24.840 --> 00:30:28.940 So I show at the beginning – at the very beginning, that all the time, we record 00:30:28.940 --> 00:30:34.380 this small vibration, which looks like this, which sends really a random signal. 00:30:34.380 --> 00:30:38.130 And, in fact, it was discarded for a very long time. 00:30:38.130 --> 00:30:42.780 So it turned out that this signal is generated by the interaction between 00:30:42.780 --> 00:30:45.850 the atmosphere, the ocean, and the solid Earth. 00:30:45.850 --> 00:30:51.810 So, during storms, you have some mechanisms which allows to generate 00:30:51.810 --> 00:30:55.120 seismic waves – a lot of seismic waves basically everywhere 00:30:55.120 --> 00:30:59.940 on Earth in the oceans. And that’s why you record those signals everywhere on 00:30:59.940 --> 00:31:05.660 Earth – in the middle of continents, on islands, during summer, during winter. 00:31:05.660 --> 00:31:11.560 So this is pretty much a random signal, but when you move from the time 00:31:11.560 --> 00:31:16.380 domain to the frequency domain, you get a shape like this for 00:31:16.380 --> 00:31:20.220 the power spectral density as a function of period. 00:31:20.220 --> 00:31:24.100 So each one of those black lines is the power spectral density 00:31:24.100 --> 00:31:29.430 over a day – computed over one day of a month. 00:31:29.430 --> 00:31:33.080 Here we are in January for a station in Europe. 00:31:33.080 --> 00:31:36.860 And here, this pink line shows you the average spectrum. 00:31:36.860 --> 00:31:41.150 So what you can see here is that you have differences – day by day, 00:31:41.150 --> 00:31:44.760 you can have a different amplitude for the power spectral density, 00:31:44.760 --> 00:31:49.200 but the shape – the overall shape remains the same. 00:31:49.200 --> 00:31:53.571 And let’s look more broadly at the entire frequency band 00:31:53.571 --> 00:31:58.690 between 10 hertz and 10,000 seconds for the three components. 00:31:58.690 --> 00:32:02.870 So this is power spectral density again as a function of period. 00:32:02.870 --> 00:32:07.560 Blue means vertical component. Red is the north component. 00:32:07.560 --> 00:32:10.050 And green is the east component. 00:32:10.050 --> 00:32:14.820 So there are names here. Because, over the years, 00:32:14.820 --> 00:32:18.430 people gave name to the different part of the spectrum. 00:32:18.430 --> 00:32:21.030 So you can see there are two major peaks. 00:32:21.030 --> 00:32:23.950 The first one here, which is the dominant one, everywhere 00:32:23.950 --> 00:32:28.580 on Earth occurred at about 7 seconds. Or you can say the range in between 00:32:28.580 --> 00:32:32.950 1 and 10 second. And it’s called the secondary microseism. 00:32:32.950 --> 00:32:39.000 The second one – primary microseism – is in between 10 and 20 second. 00:32:39.000 --> 00:32:43.620 And the long period instead is called the seismic hum of the Earth. 00:32:43.630 --> 00:32:46.930 The reason why people gave different names is because 00:32:46.930 --> 00:32:52.680 the cause of those different frequency bands is different. 00:32:52.680 --> 00:32:58.220 So looking at the very high frequency, what you find there is the signature 00:32:58.220 --> 00:33:04.060 of lakes, rivers, human activities. At very long period, you have 00:33:04.060 --> 00:33:10.480 instead atmospheric pressure having a direct effect on the solid Earth. 00:33:10.480 --> 00:33:15.550 You have tides. So very long-period phenomena. 00:33:15.550 --> 00:33:20.430 On the end of this very broad band here between 1 or 2 seconds 00:33:20.430 --> 00:33:25.350 and the 300 seconds is generated by the ocean. 00:33:25.350 --> 00:33:30.650 And in particular by ocean gravity waves in between 1 or 2 seconds and 00:33:30.650 --> 00:33:34.030 20 seconds, ocean gravity waves are the waves that you see 00:33:34.030 --> 00:33:37.660 when you go to the beach. So the waves that are restored by 00:33:37.660 --> 00:33:44.290 gravity and up at the very surface – the very shallow part of the ocean. 00:33:44.290 --> 00:33:50.660 Those have wavelengths between tenths to 100 meters. 00:33:50.660 --> 00:33:53.940 So really the first layer on top of the ocean. 00:33:53.940 --> 00:33:59.270 Then these longer-period parts, so between about 20 second and 00:33:59.270 --> 00:34:03.570 300 second, is generated by ocean infragravity waves. 00:34:03.570 --> 00:34:08.310 Those are ocean waves still but very long-period waves generated 00:34:08.310 --> 00:34:14.480 in turn by the interaction of ocean gravity waves. So are actually an active 00:34:14.480 --> 00:34:19.639 field of research in oceanography because their cause is not entirely clear. 00:34:19.639 --> 00:34:23.520 But what we know today is that those waves can generate 00:34:23.520 --> 00:34:27.519 seismic waves – the seismic waves. So how this happens? 00:34:27.519 --> 00:34:32.710 So let’s look at the dominant signal here – the secondary microseismics. 00:34:32.710 --> 00:34:37.639 So the theory was generated – was developed in the – in the ’50s 00:34:37.639 --> 00:34:42.869 and in the ’60s by those people here – Longuet-Higgins and Hasselmann. 00:34:42.869 --> 00:34:46.269 They developed analytically the generation – the theory 00:34:46.269 --> 00:34:50.259 for the generation of secondary microseismic. 00:34:50.259 --> 00:34:54.629 And what they said is basically summarized in this cartoon. 00:34:54.629 --> 00:34:59.009 So you have two waves – two ocean waves, or in the real world, 00:34:59.009 --> 00:35:03.109 the [inaudible] of waves, coming from nearly opposite directions 00:35:03.109 --> 00:35:07.040 and occurring at nearly similar frequency content. 00:35:07.040 --> 00:35:13.010 When they interact – they meet up, they generate a pressure at the 00:35:13.010 --> 00:35:15.589 second order approximation in terms of wave height. 00:35:15.589 --> 00:35:19.119 So if you develop the equation, you find the term in terms of 00:35:19.119 --> 00:35:25.130 ocean waves squared – sorry – ocean waves squared. 00:35:25.130 --> 00:35:29.980 And this pressure has the characteristics that it doesn’t attenuate with depth. 00:35:29.980 --> 00:35:35.019 So one single wave would not generate anything at the seafloor, so in the 00:35:35.019 --> 00:35:38.829 solid Earth, because the particle motion would be pretty big here, 00:35:38.829 --> 00:35:43.109 and then it will decrease, decrease, decrease up to zero at the seafloor, 00:35:43.109 --> 00:35:46.559 and decrease exponentially. But two waves interacting and 00:35:46.559 --> 00:35:52.470 generating a standing wave at the ocean surface would generate basically a signal 00:35:52.470 --> 00:35:55.940 that is able to arrive at the seafloor. If you want to see this as a – 00:35:55.940 --> 00:36:00.880 in a seismology – as a seismological point of view, it will generate acoustic 00:36:00.880 --> 00:36:04.970 or compressional waves in the ocean, which will be able – you see in this 00:36:04.970 --> 00:36:11.859 small insert here – to reach the seafloor and generate the wavefield below. 00:36:11.859 --> 00:36:15.890 And it will go again and again, up and down, between the ocean 00:36:15.890 --> 00:36:20.559 surface and the seafloor, generating – every time the P wave impacts 00:36:20.559 --> 00:36:24.869 the seafloor, generating this wavefield. 00:36:24.869 --> 00:36:30.890 This mechanism is able to explain the – mostly everything about the 00:36:30.890 --> 00:36:36.160 vertical component of noise records. It is not able to explain entirely 00:36:36.160 --> 00:36:38.849 the horizontal components, so this is a 100-year mystery 00:36:38.849 --> 00:36:41.740 in ambient noise seismology. 00:36:41.740 --> 00:36:46.579 Because this would only generate P and SP waves, this mechanism. 00:36:46.579 --> 00:36:49.519 And so you won’t explain Love waves with this. 00:36:49.519 --> 00:36:53.190 And we’ll come back to this at the very, very end of my presentation. 00:36:53.190 --> 00:36:56.860 Let’s go farther in period. 00:36:56.860 --> 00:37:00.720 Let’s talk about the other peak here – so the primary microseismic. 00:37:00.720 --> 00:37:04.519 In this case, this mechanism of ocean wave interaction became 00:37:04.520 --> 00:37:07.720 ineffective because you are moving to longer period. 00:37:07.720 --> 00:37:10.460 Something I forget to say for the previous mechanism, 00:37:10.460 --> 00:37:13.460 which is important, is that, this pressure will be at the double 00:37:13.460 --> 00:37:17.329 of the frequency of the ocean waves. So, to give you an example, 00:37:17.329 --> 00:37:23.569 an ocean wave with a period – so seismic noise at 7 seconds 00:37:23.569 --> 00:37:27.619 is generated by 14-second period in terms of ocean waves. 00:37:27.619 --> 00:37:30.289 So you need really big storms in the ocean. 00:37:30.289 --> 00:37:33.779 In this case, you need to have longer periods. 00:37:33.779 --> 00:37:37.500 And so this mechanism became ineffective because having 00:37:37.500 --> 00:37:43.630 so long period ocean waves became very, very rare. 00:37:43.630 --> 00:37:46.619 So what happened in this case, and this was mostly developed – 00:37:46.619 --> 00:37:52.460 the theory – in the ’60s by Hasselmann and then was extended recently, 00:37:52.460 --> 00:37:57.980 and was also tested with computer models, is that you – when you have 00:37:57.980 --> 00:38:04.230 one single wave at the same frequency of the seismic waves you generate 00:38:04.230 --> 00:38:08.500 approaching the coast, the bottom pressure, which is nearly negligible, 00:38:08.500 --> 00:38:12.630 as I explained to you in the previous slides, in the deep water, 00:38:12.630 --> 00:38:15.650 this bottom pressure becomes pretty large in shallow water. 00:38:15.650 --> 00:38:20.240 And actually became very large. And what you can do is to basically 00:38:20.240 --> 00:38:26.130 estimate the moment when this long wavelength pressure – 00:38:26.130 --> 00:38:34.740 this purple line here – is not zero. And, in that case, you can basically 00:38:34.740 --> 00:38:38.000 compute the wavelength, and so the wave number. 00:38:38.010 --> 00:38:43.220 And so you can compute the location of the sources, which happens in 00:38:43.220 --> 00:38:47.440 shallow water. So only shallow water you can generate this. 00:38:47.440 --> 00:38:50.609 So today, thanks to computer model, as I said, we can actually 00:38:50.609 --> 00:38:52.450 test those mechanisms. 00:38:52.450 --> 00:38:56.200 And we can compute the sources. You see here on the left – 00:38:56.200 --> 00:39:01.319 the sources during January on top and July on the bottom for 00:39:01.319 --> 00:39:03.700 secondary microseism. 00:39:03.700 --> 00:39:07.680 And on the right, for primary – January, again, on the top row 00:39:07.680 --> 00:39:12.250 and July bottom row. So the main difference is that – 00:39:12.250 --> 00:39:15.230 between those two is that, for the primary, as I said, 00:39:15.230 --> 00:39:19.930 you get sources only along the coast or close to islands. 00:39:19.930 --> 00:39:24.089 Instead, in the other case, you have sources everywhere in deep water 00:39:24.089 --> 00:39:27.680 environment as well as in shallow – in a shallow-water environment. 00:39:27.680 --> 00:39:32.180 You see strong seasonality in the secondary microseismic sources, 00:39:32.180 --> 00:39:35.840 and this is something that we have been observing for a long time in the data 00:39:35.840 --> 00:39:39.840 as well. You see less seasonality for the primary microseismic, 00:39:39.849 --> 00:39:42.859 but you can see some as well. If you focus, for example, 00:39:42.859 --> 00:39:46.190 on the north Atlantic Ocean, you’ll see that the sources 00:39:46.190 --> 00:39:49.920 get weaker during the local summer. 00:39:49.920 --> 00:39:55.140 So using those sources, basically what you can do is to make models, 00:39:55.140 --> 00:40:00.349 like we do for earthquakes, simulate with all those sources together acting, 00:40:00.349 --> 00:40:04.940 the power spectral density, as I show you here – this is an example for the 00:40:04.940 --> 00:40:09.759 primary microseismic as a function – each inset here shows you the 00:40:09.759 --> 00:40:12.069 power spectral density as a function of frequency, 00:40:12.069 --> 00:40:16.880 data versus synthetics, at 24 stations. 00:40:16.880 --> 00:40:21.510 This is an average over a year, but you can do this over time. 00:40:21.510 --> 00:40:25.640 So this is displacement as a function of time. 00:40:25.640 --> 00:40:29.079 And you can see this is the root-mean-square over time. 00:40:29.079 --> 00:40:35.450 You can see that basically these red and blue curves overlap pretty well, 00:40:35.450 --> 00:40:39.220 and you have very nice overlaps, like here, at peaks. 00:40:39.220 --> 00:40:45.040 A peak means a storm. This means that we are able to 00:40:45.040 --> 00:40:49.280 model the data due to a storm – a single storm. 00:40:49.280 --> 00:40:54.580 And so we are able to track and to go back to the storm itself. 00:40:55.490 --> 00:41:00.940 Why it is interesting to do all this? Because the first observation is 00:41:00.940 --> 00:41:05.470 that those storms are everywhere in the ocean. 00:41:05.470 --> 00:41:09.910 Earthquakes happened mostly along plate boundaries, and so we are – 00:41:09.910 --> 00:41:14.080 when we use earthquakes, we are constrained by this geometry. 00:41:14.080 --> 00:41:19.930 So adding a way to basically use both of those sources, which are still sources 00:41:19.930 --> 00:41:24.840 of seismic waves, would basically increase the resolution and the 00:41:24.840 --> 00:41:29.700 possibility of having a coverage which is much denser than 00:41:29.700 --> 00:41:34.910 using just earthquakes. The other interesting things is that, in seismology, 00:41:34.910 --> 00:41:40.400 as we know, we have – we have data – digital data since the ’60s. 00:41:40.400 --> 00:41:45.980 This is the WWSSN global scale seismic network. 00:41:45.990 --> 00:41:51.260 And even before, we had analog data since the beginning of the 20th century. 00:41:51.260 --> 00:41:54.990 This is interesting because the satellite era for monitoring 00:41:54.990 --> 00:41:59.779 the environment started about in the ’70s or in the ’60s with 00:41:59.779 --> 00:42:04.190 very few satellites at the beginning. So here’s it’s a very small – 00:42:04.190 --> 00:42:07.319 sorry, you have years here, and those are the satellites 00:42:07.319 --> 00:42:09.849 that are being used for monitoring the Earth. 00:42:09.849 --> 00:42:14.299 If you look at the beginning – this is the 1972 at the beginning – 00:42:14.299 --> 00:42:19.119 you have very, very few satellites. So you have a huge gap in knowledge 00:42:19.119 --> 00:42:24.019 going back in time when you want to monitor the global environment. 00:42:24.019 --> 00:42:29.559 And this became pretty important if you – if you care about looking at 00:42:29.559 --> 00:42:32.770 the long time series of what’s going on, for example, 00:42:32.770 --> 00:42:36.799 in the atmosphere in relation with global warming. 00:42:36.799 --> 00:42:40.240 So I show you two examples of what we can do. 00:42:40.240 --> 00:42:44.999 The first one is how we can track a tropical cyclone using seismic data. 00:42:44.999 --> 00:42:47.650 Those are two images. The first one on the left 00:42:47.650 --> 00:42:52.900 is one of the premier study on showing sources of a tropical cyclone. 00:42:52.900 --> 00:42:55.549 You can see this was Hurricane Katrina. 00:42:55.549 --> 00:42:59.849 And you can see that this source was pretty large – part of land, 00:42:59.849 --> 00:43:04.730 part in the ocean, at a particular moment. 00:43:04.730 --> 00:43:09.430 On the right, you’ll see an example of a model source. 00:43:09.430 --> 00:43:13.109 Compare with the data, this white circle is the cyclone 00:43:13.109 --> 00:43:19.360 at that time, and this patch here is the seismic source that has been modeled. 00:43:19.360 --> 00:43:22.320 But we can go farther. We can try to test this 00:43:22.329 --> 00:43:26.460 over time for a tropical cyclone. And this is an example. 00:43:26.460 --> 00:43:30.020 I’m going to show you a comparison between data – 00:43:30.020 --> 00:43:36.480 satellite data in red and observations from seismology in green. 00:43:36.480 --> 00:43:42.900 And I’m going to show you this example for strong tropical 00:43:42.900 --> 00:43:48.440 cyclones that occurred in 2006 in the western Pacific. So this is a typhoon. 00:43:48.440 --> 00:43:55.499 Was pretty intense in the sense that it was a Category 5 multiple times. 00:43:55.499 --> 00:43:58.529 And I’m going to show you a movie. So remember, red, you have data 00:43:58.529 --> 00:44:02.720 from satellites, and in green, you have data from seismology. 00:44:02.720 --> 00:44:04.700 So, at the beginning, you see that seismology is 00:44:04.700 --> 00:44:09.980 not really able to track anything. And this time is when a typhoon 00:44:09.980 --> 00:44:14.730 was actually a tropical storm, but very, very weak. But then you see that, 00:44:14.730 --> 00:44:20.100 with the findings of seismology, we are able to track the entire event over time. 00:44:20.100 --> 00:44:24.660 And so you see that – I can play it again. 00:44:24.670 --> 00:44:28.599 You can see that basically the sources follows the event. 00:44:28.599 --> 00:44:34.550 If you look carefully, the green dots follows the last red dot. 00:44:34.550 --> 00:44:38.110 And this is because basically this mechanism of ocean wave 00:44:38.110 --> 00:44:42.970 interaction happen in a tropical cyclone, this is known – this has been known for 00:44:42.970 --> 00:44:50.049 50 years about – this happen actually on the tail of the event itself all the time. 00:44:50.049 --> 00:44:55.579 So another thing that we can do is to look at the amplitude of the 00:44:55.579 --> 00:44:59.619 seismic records to say something about the event themselves. 00:44:59.619 --> 00:45:04.269 And this is – those are two example of people studying the amplitude. 00:45:04.269 --> 00:45:08.010 This is with the TA Array during Hurricane Sandy. 00:45:08.010 --> 00:45:10.579 Here is the event. And the red shows you the 00:45:10.579 --> 00:45:16.210 station with the loud signal, blue with a very low signal. 00:45:16.210 --> 00:45:18.259 And you see that there is some correlation 00:45:18.259 --> 00:45:20.799 with the location of the event. 00:45:20.799 --> 00:45:24.730 On the right, you see people showing what happened during the landfall 00:45:24.730 --> 00:45:29.299 when the event goes on land. In that case, you see that this is 00:45:29.299 --> 00:45:32.269 the spectral amplitude as a function of frequency. 00:45:32.269 --> 00:45:35.980 You see there is a shift in frequency and in amplitude. 00:45:35.980 --> 00:45:40.640 So the amplitude itself also carries an interesting – 00:45:40.640 --> 00:45:44.730 some interesting imprints of the events. 00:45:44.730 --> 00:45:48.040 So what we have been doing is to look at long time series. 00:45:48.040 --> 00:45:52.599 Here we analyze 13 years of data in the western Pacific. 00:45:52.599 --> 00:45:55.529 This is an example for a station in Taiwan here. 00:45:55.529 --> 00:46:02.119 Here, you have all the typhoons going on over 13 years. 00:46:02.119 --> 00:46:05.049 And you see the spectrogram over time. 00:46:05.049 --> 00:46:11.900 If you zoom over one hurricane season – one typhoon season, you see that 00:46:11.900 --> 00:46:15.630 you have those bursts of energy, and those correspond pretty well 00:46:15.630 --> 00:46:17.460 with those black lines. 00:46:17.460 --> 00:46:24.170 Those black lines represent the typhoons scaled in the sense of showing the 00:46:24.170 --> 00:46:29.160 tropical cyclone intensity – so the categories associated with the event. 00:46:29.160 --> 00:46:33.809 So what we thought to do is to basically analyze the data, dividing 00:46:33.809 --> 00:46:39.010 the data in the – data in the absence of tropical cyclones, data in the presence 00:46:39.010 --> 00:46:44.160 of tropical cyclones, selecting the data with some criteria, like storms within 00:46:44.160 --> 00:46:48.019 a certain range from the stations. Or strong storms – 00:46:48.019 --> 00:46:54.529 so larger than Category 1 events. And looking at the oceanic part of 00:46:54.529 --> 00:46:58.750 the track only because there we know what the mechanism is. 00:46:58.750 --> 00:47:03.680 So dividing this and looking at the data during tropical cyclone, computing the 00:47:03.680 --> 00:47:08.829 power spectral density as a function of tropical cyclone density, we can 00:47:08.829 --> 00:47:13.430 see that there is a nearly linear relationship between 00:47:13.430 --> 00:47:18.569 those two variables here. Those three shows basically 00:47:18.569 --> 00:47:22.930 the seismic data featured in three different period bands. 00:47:22.930 --> 00:47:26.630 So what you can think is that, well, we have the power spectral density. 00:47:26.630 --> 00:47:30.059 We have a linear relationship. We can compute the tropical cyclone 00:47:30.060 --> 00:47:34.220 intensity remotely using seismic data. It’s not the simple. 00:47:34.220 --> 00:47:38.259 Because the tropical cyclone intensity is not a Gaussian distribution, 00:47:38.259 --> 00:47:40.440 which would allow you to do something like this. 00:47:40.440 --> 00:47:45.790 It’s a gamma distribution. You have a lot of very weak events. 00:47:45.790 --> 00:47:48.809 Here it shows you the histogram of the typhoons in that region 00:47:48.809 --> 00:47:51.260 as a function of the tropical cyclone intensity. 00:47:51.260 --> 00:47:55.380 There are a lot of them very weak. There are few very strong. 00:47:55.390 --> 00:48:00.509 So what we need to do is to change our linear relationship, our linear 00:48:00.509 --> 00:48:04.799 regression to take into account this. So what with this was implying 00:48:04.799 --> 00:48:09.230 a generalized linear model. So generalization of a simple linear 00:48:09.230 --> 00:48:12.940 regression where you can take into account the gamma distribution for the 00:48:12.940 --> 00:48:18.180 tropical cyclone intensity is a statistical measure – a very basic machine 00:48:18.180 --> 00:48:23.780 learning technique, if you want. So we selected 11 years to train our 00:48:23.780 --> 00:48:31.080 machine – so to compute basically those parameters here, which shows you 00:48:31.089 --> 00:48:35.299 the relationship between the power spectral density and the tropical 00:48:35.299 --> 00:48:38.880 cyclone intensity. And then we use those to predict 00:48:38.880 --> 00:48:42.650 tropical cyclone intensity for two other seasons. 00:48:42.650 --> 00:48:47.230 And here I show you the comparison between tropical cyclone intensity – 00:48:47.230 --> 00:48:51.980 you have the categories of the event with colors as a function of time. 00:48:51.980 --> 00:48:57.290 You have the observed from satellites in black, 00:48:57.290 --> 00:49:02.220 and the estimated one from seismology in red, for three events. 00:49:02.220 --> 00:49:06.730 So you see that the overall increase and decreases of the lifecycle of the 00:49:06.730 --> 00:49:12.820 event is well-captured by seismic data. We have some interesting features, 00:49:12.820 --> 00:49:16.329 like a gap between the two. And this is likely due to the fact 00:49:16.329 --> 00:49:20.480 that basically winds need time to grow ocean waves. 00:49:20.480 --> 00:49:27.190 So this gap – this delay here actually is related to the interaction between 00:49:27.190 --> 00:49:31.279 the atmosphere and the ocean. This is an active field of research. 00:49:31.279 --> 00:49:36.749 So it could be very interesting to use seismology to basically 00:49:36.749 --> 00:49:38.859 say something about this. 00:49:38.859 --> 00:49:43.749 We have also some mismatch at the beginning and at the end of the event, 00:49:43.749 --> 00:49:48.269 and this is related to what I showed you in the movie of the – of the track. 00:49:48.269 --> 00:49:51.701 At the very beginning, we’re not able to track the event. 00:49:51.701 --> 00:49:54.590 We’re not able to reconstruct the intensity because the event is weak. 00:49:54.590 --> 00:49:59.799 It’s a tropical depression, actually. But overall, the correlation coefficient 00:49:59.799 --> 00:50:04.839 is pretty high. And, if you’re not convinced that this is good, 00:50:04.839 --> 00:50:10.400 I can show you what’s the tropical cyclone intensity in the re-analysis 00:50:10.400 --> 00:50:14.910 data set. So the re-analysis data set is what atmospheric scientists use for 00:50:14.910 --> 00:50:19.420 describing the state of the atmosphere. And they know – is, again, 00:50:19.420 --> 00:50:23.650 another very active field of research. They know that tropical cyclones, 00:50:23.650 --> 00:50:27.140 like every other extreme, they are not well-captured. 00:50:27.140 --> 00:50:32.680 And you see that the – those event are massively underestimated. 00:50:32.680 --> 00:50:34.859 And actually, we do better. 00:50:34.860 --> 00:50:40.340 So we could actually use something like this to help 00:50:40.340 --> 00:50:44.780 constraining atmospheric data sets for tropical cyclones. 00:50:44.780 --> 00:50:50.079 So, to summarize this second part, I show you that, today, we have a 00:50:50.079 --> 00:50:54.930 pretty good knowledge of the generation mechanism and where the sources 00:50:54.930 --> 00:51:00.089 are in the ocean during storms. We can model the spectrum. 00:51:00.089 --> 00:51:04.210 We can make computer models for the spectrum of ambient noise. 00:51:04.210 --> 00:51:10.260 And we can track strong events like tropical cyclone and reconstruct the 00:51:10.260 --> 00:51:15.400 amplitude – the intensity of those events remotely using seismology. 00:51:15.400 --> 00:51:20.499 And so this is – this can be very informative, especially going back in 00:51:20.499 --> 00:51:25.589 time and using some old data that we have at the global scale for constraining 00:51:25.589 --> 00:51:31.140 characteristics of those strong events. And so the other – the other interesting 00:51:31.140 --> 00:51:36.279 thing is that we can use computer model to monitor the solid Earth and solve 00:51:36.279 --> 00:51:45.529 mysteries in the – in the ambient noise fields, like the one I described before, 00:51:45.529 --> 00:51:48.519 how Love waves are generated. We are working on this. 00:51:48.519 --> 00:51:54.269 We are working right now on using a realistic computer model for the Earth 00:51:54.269 --> 00:52:00.999 which accounts for topography, 3D structure, elasticity, gravity, and so on. 00:52:00.999 --> 00:52:06.289 And we are implementing sources in those models, like we have been doing 00:52:06.289 --> 00:52:11.470 for years for earthquakes, to, on one end, trying to understand where Love waves 00:52:11.470 --> 00:52:17.369 come from, and also as a first step for monitoring the solid Earth. 00:52:17.369 --> 00:52:21.759 So with this, I want to thank you for your attention, and I want to also 00:52:21.760 --> 00:52:27.460 acknowledge all my collaborators I had over the years. Thank you. 00:52:29.640 --> 00:52:33.200 [Silence] 00:52:33.200 --> 00:52:38.779 - All right. Thank you, Lucia, for a really nice talk. 00:52:38.779 --> 00:52:43.559 We want to open it up to questions, and if you have a question, 00:52:43.560 --> 00:52:47.020 you can type it into the meeting chat. 00:52:48.020 --> 00:52:55.500 And we’ll see what questions come up, but in the meantime, I wanted to 00:52:55.500 --> 00:52:59.340 ask one to Lucia. Can you kind of … 00:52:59.340 --> 00:53:01.160 - Yes? - … speculate a little bit? 00:53:01.160 --> 00:53:07.529 Do you think that this research could be useful in interpreting data 00:53:07.529 --> 00:53:12.559 from the INSIGHT lander on Mars? - That’s a – that’s a very interesting 00:53:12.559 --> 00:53:17.470 question. I know there are people working on this, on the specific aspect 00:53:17.470 --> 00:53:23.900 of studying the ambient background vibration due to Mars. 00:53:23.900 --> 00:53:29.599 So we have no ocean on Mars. But we could have a signal due to 00:53:29.599 --> 00:53:35.759 small rock movements or even signal due to the atmosphere on Mars. 00:53:35.759 --> 00:53:39.200 So I’m not personally working on this, but I know there is – there is an 00:53:39.200 --> 00:53:43.829 effort of studying this. And so this has been – this signal 00:53:43.829 --> 00:53:49.829 has been pretty useful, and will be useful, for studying the Earth. 00:53:49.829 --> 00:53:56.360 And so it’s probably a good signal for studying the interior of Mars as well. 00:53:58.880 --> 00:54:11.960 [Silence] 00:54:11.970 --> 00:54:15.230 - And if anyone wants to ask their question, 00:54:15.230 --> 00:54:22.060 you can also just unmute yourself and ask to the group. 00:54:24.020 --> 00:54:33.080 [Silence] 00:54:33.080 --> 00:54:35.980 - This is Andy. I’ll ask a question. 00:54:35.980 --> 00:54:39.720 When you go back to the older data, the seismic data is also degraded. 00:54:39.720 --> 00:54:44.979 Obviously it precedes satellites, but particularly in terms of the periods 00:54:44.979 --> 00:54:47.089 that are available in the old data, it is digitized. 00:54:47.089 --> 00:54:51.440 How would that limit your ability to track and really – or, I think it’s really 00:54:51.440 --> 00:54:56.200 more important to get the overall quantity of storms. 00:54:56.200 --> 00:55:01.220 - Yes. Yeah. That’s a very interesting question. 00:55:01.220 --> 00:55:08.180 So, of course, it will be much more difficult to track events going back in 00:55:08.180 --> 00:55:13.710 time, especially because, to do it in a reliable way, you need arrays of stations. 00:55:13.710 --> 00:55:18.799 So many stations in – close each other in a small area. 00:55:18.799 --> 00:55:24.229 And so, going back in time, this is – this becomes pretty rare. 00:55:24.229 --> 00:55:27.319 So the first array, I think, was applied around the ’60s. 00:55:27.319 --> 00:55:32.230 So it’s not bad because you can go back a little bit, but not back in time. 00:55:32.230 --> 00:55:37.849 On the other end, you have a – basically single stations since the 00:55:37.849 --> 00:55:42.210 beginning of the 20th century. And the study I showed you for 00:55:42.210 --> 00:55:48.049 the tropical cyclone intensity was done using one single station. 00:55:48.049 --> 00:55:53.450 So there is a potential for using single stations to study on 00:55:53.450 --> 00:55:57.039 one end of the intensity. And this is one big question in 00:55:57.039 --> 00:56:01.220 atmospheric science, is the intensity of tropical cyclone increasing over time 00:56:01.220 --> 00:56:04.380 due to global warming? The answer is, we don’t know, 00:56:04.380 --> 00:56:07.950 in the sense that there are evidences that probably this is true, but the 00:56:07.950 --> 00:56:13.789 time series that atmospheric scientists are using now is pretty short. 00:56:13.789 --> 00:56:19.900 So everything that can actually allow to go back in time 00:56:19.900 --> 00:56:22.500 with this will be very useful. And, on the other hand, 00:56:22.500 --> 00:56:28.910 you can actually try to count events – try to see if the number of those events 00:56:28.910 --> 00:56:31.340 is increasing in time. 00:56:32.140 --> 00:56:35.220 - Great. Thanks. - Thanks. 00:56:37.540 --> 00:56:40.920 - Are there any other – are there more questions? 00:56:43.340 --> 00:56:49.080 [Silence] 00:56:49.080 --> 00:56:51.260 Don’t be shy. 00:56:54.200 --> 00:57:05.040 [Silence] 00:57:05.049 --> 00:57:09.329 I was also curious in that … - Sure. 00:57:09.329 --> 00:57:17.410 - You may have mentioned, but why is the secondary microseism 00:57:17.410 --> 00:57:22.940 a larger amplitude than the primary microseism? 00:57:22.940 --> 00:57:27.749 - That’s a good question. So, well, first of all, as you have seen 00:57:27.749 --> 00:57:32.380 when I showed the maps of the sources, you see that those sources are 00:57:32.380 --> 00:57:36.480 basically everywhere compared the primary microseismic. 00:57:36.499 --> 00:57:43.950 So you have a lot of sources going on. And so, you have basically a large 00:57:43.950 --> 00:57:48.450 amount of sources acting at the same time, which gives you 00:57:48.450 --> 00:57:51.609 a larger amplitude. On the other hand, 00:57:51.609 --> 00:57:54.880 this secondary microseismic is very seasonal. 00:57:54.880 --> 00:57:59.970 So, in summer – during summertime, you’ll get the amplitude decreasing 00:57:59.970 --> 00:58:04.529 due to the fact that the storms around are weak or absent. 00:58:04.529 --> 00:58:09.360 And so you are sensitive to only storms far away. 00:58:12.340 --> 00:58:17.580 - Any final questions? - I have one. 00:58:17.580 --> 00:58:20.560 Hi, this is Alan Yong. Nice talk, Lucia. 00:58:20.560 --> 00:58:24.680 - Thank you. - My question for us is – 00:58:24.680 --> 00:58:30.800 actually, I’d like you to give us your thoughts on the whole diffused fuel 00:58:30.809 --> 00:58:37.200 concept and, like, partition of energy. This is, of course, you know, work 00:58:37.200 --> 00:58:40.820 that’s been recently promoted by Hiroshi Kawase 00:58:40.820 --> 00:58:44.660 and Paco Sánchez-Sesma. 00:58:46.120 --> 00:58:52.140 - Yeah. So this is – this question is pretty much related to what I’m doing 00:58:52.140 --> 00:58:57.099 now in the sense that the study of the Love waves. 00:58:57.099 --> 00:59:02.869 So, at the moment, we are able to explain one-third of the data. 00:59:02.869 --> 00:59:07.799 So it’s a lot in the sense that, before the ’50s, we didn’t know 00:59:07.799 --> 00:59:12.400 literally anything about seismic noise. We were just discarding noise. 00:59:12.400 --> 00:59:16.769 Then, in the ’50s and ’60s, Longuet-Higgins and Hasselmann 00:59:16.769 --> 00:59:19.410 developed this theory, which works pretty well to 00:59:19.410 --> 00:59:22.819 explain the vertical component of the data. 00:59:22.819 --> 00:59:26.950 We don’t know anything about the generation of the other two components, 00:59:26.950 --> 00:59:32.170 and especially for the Love waves. We know how Rayleigh waves generate, 00:59:32.170 --> 00:59:37.210 and so for the three components, but we don’t know anything about Love waves. 00:59:37.210 --> 00:59:42.530 And so the equal partition of energy will be pretty much related to 00:59:42.530 --> 00:59:46.130 the generation mechanism of where the sources are for Love waves. 00:59:46.130 --> 00:59:51.049 In the sense that the possible – let me explain – the possible 00:59:51.049 --> 00:59:54.180 mechanism for the generation of Love waves are two. 00:59:54.180 --> 00:59:58.349 One is due to the bathymetry. So basically, you have this pressure 00:59:58.349 --> 01:00:03.349 force which, in absence of any roughnesses that discontinue this, 01:00:03.349 --> 01:00:08.249 will generate only P and SP waves. But if add the roughnesses – so if you 01:00:08.249 --> 01:00:12.440 add the bathymetry, you basically will generate a splitting of the force. 01:00:12.440 --> 01:00:15.900 You generate a horizontal component of the force, and so Love waves. 01:00:15.900 --> 01:00:19.390 This is one possibility. The other one is, instead, 01:00:19.390 --> 01:00:23.460 related to the structure itself. And the structure means that this 01:00:23.460 --> 01:00:27.369 wavefield – those sources – those pressure sources generate 01:00:27.369 --> 01:00:32.999 P and SP waves, which will arrive at the 3D – at the [inaudible] of the Earth. 01:00:32.999 --> 01:00:37.249 And by scattering the focusing – the focusing effect will be partially 01:00:37.249 --> 01:00:40.960 converted to Love waves. So, as you can understand, 01:00:40.960 --> 01:00:44.869 those two mechanisms are pretty different and will lead to different 01:00:44.869 --> 01:00:47.259 conclusions in terms of equal partition of energy. 01:00:47.259 --> 01:00:51.420 And, in particular, the second one is an ergodic vision of the Earth – 01:00:51.420 --> 01:00:55.750 if you want the Earth to act as an ergodic system. 01:00:55.750 --> 01:01:01.519 So you leave those sources generating waves, and if you wait enough, 01:01:01.519 --> 01:01:05.569 you will generate all the possible waves that are expected 01:01:05.569 --> 01:01:10.320 for the Earth. So discriminating this will give us 01:01:10.320 --> 01:01:15.910 understanding what are the sources for this wave, which is two-thirds of the 01:01:15.910 --> 01:01:21.520 data, will give us an important answer on the equal partition of energy. 01:01:23.820 --> 01:01:26.979 - Thank you. That’s a – that’s a very interesting answer. 01:01:26.979 --> 01:01:29.700 And it seems like there’s a lot of caveats, 01:01:29.700 --> 01:01:34.880 as represented by what you’ve remarked. Thank you. 01:01:34.880 --> 01:01:36.980 - Thank you. 01:01:38.410 --> 01:01:42.320 - So we have a question from Clara Yoon, who asks, how close to the 01:01:42.339 --> 01:01:47.390 landslide does the seismic station need to be to reliably do the analysis 01:01:47.390 --> 01:01:51.590 and infer the mass? - So it depends on the event. 01:01:51.590 --> 01:01:55.000 It depends how big the event was. 01:01:55.080 --> 01:02:01.540 I showed you at a certain point an example of a landslide in Taiwan, 01:02:01.540 --> 01:02:06.560 and you showed the data really coming up from the ground noise. 01:02:06.569 --> 01:02:10.630 That was a big landslide. [inaudible] to magnitude 5. 01:02:10.630 --> 01:02:13.589 And in that case, was basically worldwide seen. 01:02:13.589 --> 01:02:17.190 As the one I showed you – that was another big landslide 01:02:17.190 --> 01:02:21.070 was worldwide seen. But, of course, this depends on the event. 01:02:21.070 --> 01:02:26.640 I studied, as well, a very small, tiny rockfall that occurred in the – 01:02:26.640 --> 01:02:29.749 along the Hudson River in New York. 01:02:29.749 --> 01:02:35.250 And, in that case, it was a very small event, and so, going farther away, 01:02:35.250 --> 01:02:38.690 you couldn’t see much. But, in that case, we were lucky enough 01:02:38.690 --> 01:02:41.749 to have a seismometer 2 kilometers away, 01:02:41.749 --> 01:02:47.749 and that was enough to estimate the mass of the event and reconstruct 01:02:47.749 --> 01:02:51.500 the dynamics of the event. So it depends. 01:02:53.540 --> 01:02:56.480 - All right. Any last questions? 01:02:58.860 --> 01:03:02.400 [Silence] 01:03:02.400 --> 01:03:07.020 All right. If not, well, thank you to Lucia. 01:03:07.030 --> 01:03:11.950 And a reminder to everyone on the seminar that we will 01:03:11.950 --> 01:03:17.229 have another seminar next week at 10:30 a.m. on Wednesday. 01:03:17.229 --> 01:03:23.010 And another plug that, if you have suggestions for speakers for the summer, 01:03:23.010 --> 01:03:27.220 send them to me and Noha. Thank you a lot, Lucia. 01:03:27.220 --> 01:03:29.260 - Thank you. Thank you, everyone. 01:03:29.260 --> 01:03:30.820 - Thank you. - All right. Thank you. 01:03:30.820 --> 01:03:32.820 - Thanks. - Bye. 01:03:34.040 --> 01:03:37.640 [Silence]