WEBVTT Kind: captions Language: en 00:00:01.660 --> 00:00:03.180 All right. Good morning, everyone. 00:00:03.180 --> 00:00:07.830 So just first, could we kind of move towards the front of the room? 00:00:07.830 --> 00:00:09.910 Ross will be having some demonstrations. 00:00:09.910 --> 00:00:11.790 So while you are moving … - [inaudible] 00:00:11.790 --> 00:00:15.680 [laughter] [inaudible] 00:00:15.680 --> 00:00:17.080 - It’ll be a good time. 00:00:17.080 --> 00:00:20.660 Okay, so while you guys are moving, so I feel a little strange introducing Ross 00:00:20.660 --> 00:00:23.890 because you guys probably know Ross better than you know me. [laughs] 00:00:23.890 --> 00:00:28.320 But, for those who are new, Ross is an emeritus scientist here at the USGS. 00:00:28.320 --> 00:00:31.089 He’s also an adjunct professor at Stanford. 00:00:31.089 --> 00:00:34.760 And more recently, he is the co-founder of Temblor, which is 00:00:34.760 --> 00:00:41.250 sort of a seismic hazard app and website that he’s been working very hard on. 00:00:41.250 --> 00:00:43.080 So Ross is a very accomplished scientist. 00:00:43.080 --> 00:00:46.020 He is currently the AGU Tectonic Physics president 00:00:46.020 --> 00:00:50.059 and the current GSA Distinguished Lecturer. 00:00:50.059 --> 00:00:55.540 So what we’ll be seeing today is a version of that Distinguished Lecture. 00:00:55.540 --> 00:00:58.990 So I learned that Ross is the second-highest-cited earthquake 00:00:58.990 --> 00:01:03.969 scientist in the ‘90s and the 10th-highest-cited in the past century. 00:01:03.969 --> 00:01:06.240 Which is pretty incredible. 00:01:06.240 --> 00:01:08.940 So there are many other accomplishments, most of which 00:01:08.940 --> 00:01:12.860 you probably already know. So I’ll hand it over to Ross now. 00:01:12.860 --> 00:01:15.100 Thank you, Ross. 00:01:18.320 --> 00:01:21.120 Okay. Great to be here. 00:01:21.120 --> 00:01:24.130 The argument I’m going to make in this talk is that the idea 00:01:24.130 --> 00:01:29.970 that earthquakes are random, independent events is a mirage. 00:01:29.970 --> 00:01:32.840 That earthquakes strongly interact. 00:01:32.840 --> 00:01:35.360 Aftershocks are the most obvious and the 00:01:35.360 --> 00:01:39.360 most abundant example of that interaction, but there are many others. 00:01:39.360 --> 00:01:43.750 But it’s hard to understand and pull out from the 00:01:43.750 --> 00:01:47.600 mass of earthquakes those strong interactions. 00:01:47.600 --> 00:01:53.760 So my argument is that we can do this and that understanding earthquake 00:01:53.760 --> 00:01:58.670 interactions is the easiest way in to the hardest problem in seismology, 00:01:58.670 --> 00:02:03.270 which is earthquake prediction. So we’re not going to predict the main shock. 00:02:03.270 --> 00:02:07.430 But after that main shock, we have tools and we have resources to tell us 00:02:07.430 --> 00:02:12.140 what could happen next in terms of aftershocks and subsequent main shocks. 00:02:12.140 --> 00:02:15.840 This may be the best we can ever do toward something that 00:02:15.840 --> 00:02:19.480 has predictive power. And the question is, how far can we take it? 00:02:19.480 --> 00:02:21.340 What is the evidence of its success? 00:02:21.340 --> 00:02:26.760 To what extent can we build that into our seismic hazard analyses? 00:02:27.410 --> 00:02:32.200 Now, everything I understand about earthquakes is represented 00:02:32.200 --> 00:02:34.260 in these demos. - No. 00:02:34.260 --> 00:02:36.620 [laughter] 00:02:36.620 --> 00:02:40.180 - And so you’re going to soon understand everything I do. 00:02:40.190 --> 00:02:43.700 And so the idea of this talk is I’m first – we’re going to spend half the time 00:02:43.700 --> 00:02:47.820 showing you these demos to explain what the principles are 00:02:47.820 --> 00:02:50.740 of earthquakes and earthquake interaction. 00:02:50.740 --> 00:02:52.400 And then we’re going to go to slides. 00:02:52.400 --> 00:02:57.850 And the goal of the slides is just to say, can we confirm that in the Earth itself? 00:02:57.850 --> 00:03:03.901 Okay, so my first preposition – proposition to you is that there are 00:03:03.901 --> 00:03:07.930 four things you need – four things you need to create earthquakes. 00:03:07.930 --> 00:03:10.950 If you have all four, you only get earthquakes. 00:03:10.950 --> 00:03:13.850 If you’re missing any one of them, you never get earthquakes. 00:03:13.850 --> 00:03:16.680 Okay, what are those? One is the steady motion 00:03:16.680 --> 00:03:20.860 of the plate’s interiors, which I’m going to represent by this crank. 00:03:20.860 --> 00:03:23.560 I’m going to just steadily crank my casting reel. 00:03:23.560 --> 00:03:25.660 That’s why it’s called QuakeCaster. 00:03:25.660 --> 00:03:32.200 That’s connected by a high-tech non-stretch line to a rubber band. 00:03:32.200 --> 00:03:34.630 That rubber band represents the rubberiness, 00:03:34.630 --> 00:03:36.720 the elasticity, of the Earth’s crust. 00:03:36.720 --> 00:03:41.280 Okay, the Earth’s crust turns out to be rubbery, it’s just very stiff rubber. 00:03:41.680 --> 00:03:45.840 That’s connected to a mass – my kitchen countertop samples. 00:03:45.850 --> 00:03:50.210 And they’re sitting on top of a frictional surface. That’s the fault. 00:03:50.210 --> 00:03:54.940 Okay? Mass, friction, elasticity, steady motion. 00:03:54.940 --> 00:03:57.110 So now I’m just going to crank as steadily as I can, 00:03:57.110 --> 00:03:59.820 and I want you just to see what happens. 00:03:59.820 --> 00:04:11.840 [intermittent scratching sounds] 00:04:11.840 --> 00:04:13.140 Okay. 00:04:13.140 --> 00:04:16.260 There’s nothing I can do, if I’m cranking steadily, 00:04:16.260 --> 00:04:20.120 to prevent earthquakes from occurring. They’re always going to occur, right? 00:04:20.120 --> 00:04:24.240 Okay, so let me ask you some questions. Were all the earthquakes the same size? 00:04:24.240 --> 00:04:27.840 If we define size as how they slip? 00:04:27.840 --> 00:04:29.040 Come on. - [multiple responses] No. 00:04:29.050 --> 00:04:30.050 - No. 00:04:30.050 --> 00:04:31.900 Were they all separated by the same amount of time? 00:04:31.900 --> 00:04:34.509 - [multiple responses] No. - Was I cranking steadily? 00:04:34.509 --> 00:04:35.840 - Yeah. - Sort of. 00:04:35.840 --> 00:04:38.420 - More or less. - Okay. That’s very bad news. 00:04:38.420 --> 00:04:41.300 If I can’t make regular repeating earthquakes with Home Depot 00:04:41.300 --> 00:04:45.540 sandpaper, kitchen countertop samples, and rubber bands, we’re never 00:04:45.550 --> 00:04:48.460 going to get them in the Earth. And that’s the bitter pill 00:04:48.460 --> 00:04:52.620 that seismologists have had to swallow for 50 years of research. 00:04:52.620 --> 00:04:58.160 Earthquakes are not predictable in terms of regularity. 00:04:58.160 --> 00:05:01.560 Everywhere we see this. This is an amazing thing when you 00:05:01.570 --> 00:05:07.410 think about it because, if I was cranking steadily, and if the rubber band 00:05:07.410 --> 00:05:11.400 was always within its elastic range, and the mass never changed, there’s 00:05:11.400 --> 00:05:16.700 only one explanation for that remarkable aperiodicity and irregularity. 00:05:17.420 --> 00:05:21.240 It’s got to be the sandpaper, right? And yet – this is industrial nonskid – 00:05:21.240 --> 00:05:25.590 if I put my hand over this, it feels very, very regular. 00:05:25.590 --> 00:05:28.710 And that tells us that tiny changes in friction 00:05:28.710 --> 00:05:32.870 exert a huge influence on the earthquake process. 00:05:32.870 --> 00:05:37.180 Lowly friction controls the whole show. That’s what you get out of this. 00:05:37.900 --> 00:05:41.780 All right. Now let me ask you another question. 00:05:41.780 --> 00:05:45.200 Do I have a longer wait after a bigger earthquake? 00:05:45.200 --> 00:05:47.460 Stands to reason, right? Because if I have 00:05:47.470 --> 00:05:51.930 a bigger earthquake, presumably, the rubber band now gets 00:05:51.930 --> 00:05:55.680 much more relaxed, and I’m going to have to crank more afterwards. 00:05:56.760 --> 00:05:59.720 Like everything else in this experiment, there’s no right answers. 00:05:59.720 --> 00:06:07.440 [intermittent scratching sounds] 00:06:07.960 --> 00:06:10.820 Not very convincing, right? And if we did it again and asked 00:06:10.830 --> 00:06:15.090 the question, well, do we get a larger earthquake after a longer wait, 00:06:15.090 --> 00:06:17.919 which seems quite reasonable, right – because then the rubber band 00:06:17.919 --> 00:06:21.090 will stretch much more – we don’t see that, either. 00:06:21.090 --> 00:06:24.800 What we used to call time predictability and slip predictability of earthquakes 00:06:24.800 --> 00:06:30.270 cratered. They make perfect sense. They’re rational. They don’t work. 00:06:30.270 --> 00:06:33.200 What this means is, when we say a fault is overdue, 00:06:33.200 --> 00:06:36.860 it does not mean that the next earthquake is going to be large. 00:06:37.860 --> 00:06:40.360 We say this all the time. Or some people say this all the time. 00:06:40.360 --> 00:06:42.360 And it cannot be right. 00:06:42.360 --> 00:06:46.610 Because there’s so much variability in this, even when the – when we have 00:06:46.610 --> 00:06:50.520 those situations, it doesn’t produce it. And another thing that’s interesting. 00:06:51.360 --> 00:06:57.540 If I look at the spring right after an earthquake, it’s still really taut. 00:06:58.160 --> 00:07:01.520 It’s not much different than right before the earthquake. 00:07:01.520 --> 00:07:05.669 In fact, something like 85% of the tautness of the spring 00:07:05.669 --> 00:07:08.350 is there right after the earthquake. 00:07:08.350 --> 00:07:12.790 That’s also important. That says earthquakes rarely, 00:07:12.790 --> 00:07:18.270 if ever, drop all the stress. We don’t get this after the earthquake. 00:07:18.270 --> 00:07:24.480 That means all faults live very close to failure for most of their lives. 00:07:24.480 --> 00:07:28.919 Earthquakes are really only fluctuating a small amount of the total stress 00:07:28.919 --> 00:07:32.889 the system is under. And that makes our life difficult as well. 00:07:32.889 --> 00:07:38.380 Because we’re not seeing the entire process of stress accumulation. 00:07:38.380 --> 00:07:43.080 Probably only the largest faults, and very close to the faults, do we ever see 00:07:43.080 --> 00:07:48.009 much more than 15% drop. And this is exactly what we see in the Earth. 00:07:48.009 --> 00:07:50.289 Stress drops on – typical stress drops 00:07:50.289 --> 00:07:53.040 for earthquakes are on the order of 30 bars. 00:07:53.040 --> 00:07:54.889 But the total stress those faults are under 00:07:54.889 --> 00:07:58.330 is probably much more like 300, 500 bars. 00:07:58.330 --> 00:08:00.710 So yet another element. 00:08:00.710 --> 00:08:02.539 Okay. 00:08:02.540 --> 00:08:07.280 What if – and I’m going to get rid – use my mustache brush, get rid of that. 00:08:07.280 --> 00:08:08.440 Okay. 00:08:08.440 --> 00:08:13.060 What if I, instead, do a one-bricker rather than a two-bricker? 00:08:13.070 --> 00:08:16.600 So what’s your prediction? If I’m cranking at the same 1-inch-a-year 00:08:16.600 --> 00:08:21.479 San Andreas rate, am I going to get fewer earthquakes or more? 00:08:21.479 --> 00:08:22.509 - More. - More. 00:08:22.509 --> 00:08:25.229 - And are they going to have smaller slip or larger? 00:08:25.229 --> 00:08:26.330 - Smaller. - Smaller. 00:08:26.330 --> 00:08:28.780 - Okay. So people have a pretty clear prediction about this. 00:08:28.780 --> 00:08:31.800 Let me show you again what the two-bricker looked like. 00:08:31.800 --> 00:08:33.279 [intermittent scratching sounds] 00:08:33.280 --> 00:08:35.660 Okay. And I’m going to take that off. 00:08:35.660 --> 00:08:38.280 And I’m going to try to crank at the same speed. 00:08:38.289 --> 00:08:41.719 [intermittent scratching sounds] 00:08:41.719 --> 00:08:44.670 So you’re right. We’re seeing about twice the rate of 00:08:44.670 --> 00:08:48.100 earthquakes, and they have less slip. Okay, and that makes sense. 00:08:48.100 --> 00:08:51.369 Because this is a force balance, right? The force-resisting motion 00:08:51.369 --> 00:08:55.500 is just the change in length of the spring times the spring stiffness. 00:08:55.500 --> 00:08:59.080 I mean, the force acting on the motion. The force-resisting motion is the 00:08:59.089 --> 00:09:03.279 weight of this brick times the friction coefficient, which is about a half. 00:09:03.279 --> 00:09:07.629 So if we halve the weight, we halve the resistance, so it goes sooner. 00:09:07.629 --> 00:09:10.939 Okay. As we all know, we have a one-bricker place 00:09:10.939 --> 00:09:14.980 on the San Andreas at Parkfield. It produces earthquakes much more 00:09:14.980 --> 00:09:20.079 frequently than northern California, which at least can produce two-brickers, 00:09:20.080 --> 00:09:24.580 such as in 1906, and southern California, such as in 1857. 00:09:24.580 --> 00:09:29.460 Why we only have a one-bricker earthquake there is another question, 00:09:29.470 --> 00:09:34.819 but it can occur. And what’s interesting about that is everything else is the same. 00:09:34.819 --> 00:09:38.899 Same friction. Same spring stiffness. Same rate. 00:09:38.900 --> 00:09:42.880 And yet, we’re producing earthquakes at twice the rate and half the slip. 00:09:43.700 --> 00:09:46.400 All right, let’s go one step farther. 00:09:46.410 --> 00:09:51.399 What we’ve done on the bottom of these guys is we sandblasted them to be rough. 00:09:51.399 --> 00:09:54.319 On the top is how your kitchen countertop is supposed to be. 00:09:54.319 --> 00:09:59.740 It’s been polished. So what if I put this guy upside-down? 00:09:59.740 --> 00:10:02.139 Now what do you think is going to happen? 00:10:02.140 --> 00:10:04.080 So who thinks I’m going to get 00:10:04.080 --> 00:10:07.440 no earthquakes at all, and it’s just going to creep? 00:10:07.449 --> 00:10:11.239 Who thinks I’m going to get more earthquakes than before? 00:10:11.239 --> 00:10:13.660 Who thinks it’s going to have no difference? 00:10:13.660 --> 00:10:14.699 Okay. 00:10:14.699 --> 00:10:17.660 So what’s interesting about this is the friction on this side is 00:10:17.660 --> 00:10:20.709 identical to what it was before. I’m only changing the friction 00:10:20.709 --> 00:10:24.040 on one side of the fault. And I’m going to crank at the same speed. 00:10:24.500 --> 00:10:27.020 [quiet scratching sounds] 00:10:27.020 --> 00:10:32.300 And it’s very interesting. It’s not quite, but almost, creeping. 00:10:32.300 --> 00:10:34.579 That was just the rubber band getting underneath it. 00:10:34.580 --> 00:10:36.460 Let’s try this again. 00:10:37.829 --> 00:10:41.279 So now we get kind of a staccato very high-frequency 00:10:41.279 --> 00:10:46.699 earthquakes because we brought the friction down very, very low. 00:10:46.699 --> 00:10:50.679 And that’s exactly what the creeping section of the San Andreas behaves like. 00:10:50.679 --> 00:10:55.249 It has the highest rate of earthquakes everywhere, but they’re small. 00:10:55.249 --> 00:10:58.129 And it’s only got low friction on one side of the fault. 00:10:58.129 --> 00:11:02.790 One side is granite, just like this guy. And the other side has got 00:11:02.790 --> 00:11:05.560 very, very clay materials with very, very low friction. 00:11:05.560 --> 00:11:07.660 And yet, it has this behavior. 00:11:09.320 --> 00:11:11.239 Now, you might say, yeah, yeah, yeah. 00:11:11.239 --> 00:11:13.990 But the San Andreas is 1,000 kilometers long. You’re talking about 00:11:13.990 --> 00:11:17.940 an earthquake that’s 20 kilometers long. Faults don’t work in isolation. 00:11:17.940 --> 00:11:19.540 Right? That’s true. 00:11:19.540 --> 00:11:24.500 So let’s put a few more earthquakes into the system. 00:11:24.500 --> 00:11:27.820 So now I’m going to put these guys together. 00:11:28.560 --> 00:11:34.180 [rusting sounds] 00:11:34.180 --> 00:11:38.540 Okay. Here comes the next prediction. Which guy is going first? 00:11:38.540 --> 00:11:41.260 Who thinks it’s this one? Okay. 00:11:41.260 --> 00:11:44.180 Who’s going second? Is it this one? 00:11:44.800 --> 00:11:46.959 Who’s going third? Is it this one? 00:11:46.959 --> 00:11:49.519 This could be a trick question. [laughter] Okay? 00:11:49.520 --> 00:11:51.020 So here we go. 00:11:51.689 --> 00:11:57.260 [intermittent scratching sounds] 00:11:57.269 --> 00:11:59.940 It gets really interesting really fast, right? 00:11:59.940 --> 00:12:02.620 Yes. Steve was right. This guy went first. 00:12:02.620 --> 00:12:05.949 But he probably also went second until he had put enough strain on this 00:12:05.949 --> 00:12:10.429 second spring that it loaded this guy. And when this guy went, eventually 00:12:10.429 --> 00:12:13.610 this guy – and this guy unloaded this, and then they all went together. 00:12:13.610 --> 00:12:16.189 So sometimes we had three independent earthquakes 00:12:16.189 --> 00:12:19.310 that might be separated by 30 years, 100 years. 00:12:19.310 --> 00:12:24.249 And sometimes we had them all go together, separated by seconds. 00:12:24.249 --> 00:12:26.749 And that’s exactly what we see in the Earth. 00:12:26.749 --> 00:12:33.549 The 1992 Landers earthquake strung together three little faults that, 00:12:33.549 --> 00:12:36.769 on a good day, could have produced a 6-1/2. 00:12:36.769 --> 00:12:40.730 And yet, within a matter of 30 seconds, they all ruptured together. 00:12:40.730 --> 00:12:43.439 And the paleoseismic evidence for that and many other faults 00:12:43.440 --> 00:12:45.940 show that isn’t always the behavior. 00:12:45.940 --> 00:12:50.699 Sometimes these guys go together when the conditions are just right. 00:12:50.699 --> 00:12:52.639 We never know what those conditions are. 00:12:52.639 --> 00:12:54.649 We can’t measure the tension of the spring. 00:12:54.649 --> 00:12:56.920 And if we did, we’d have to do it really accurately 00:12:56.920 --> 00:13:01.470 because it’s already 90% of the way there almost all the time. 00:13:01.470 --> 00:13:03.410 And that further makes the problem of 00:13:03.410 --> 00:13:06.730 earthquake prediction enormously difficult. 00:13:06.730 --> 00:13:11.829 Because probably no little earthquake knows it’s marked for future greatness. 00:13:11.829 --> 00:13:14.980 It doesn’t know if it’s going to bring its sisters along. 00:13:14.980 --> 00:13:17.619 And yet it does happen, and when that happens is 00:13:17.620 --> 00:13:21.489 when we get the really big earthquakes or the really unexpected earthquakes. 00:13:22.440 --> 00:13:28.240 Now, at this point, you might say, well, yes, okay, now you’ve gotten a long fault 00:13:28.240 --> 00:13:31.800 interacting, but that’s not the whole story because there are parallel faults. 00:13:31.810 --> 00:13:35.709 Like, in the Bay Area, we have four or five parallel strike-slip faults. 00:13:35.709 --> 00:13:38.149 On a big subduction zone, we have a large surface 00:13:38.149 --> 00:13:41.709 on which earthquakes occur. So this is still a little limited, 00:13:41.709 --> 00:13:44.329 and so I wanted to go one step farther. 00:13:44.329 --> 00:13:46.080 So I’m going to take these guys off. 00:13:47.140 --> 00:13:49.680 [rusting sounds] 00:13:50.060 --> 00:13:52.820 I’m going to put this on. 00:13:54.020 --> 00:13:58.900 And this is something that they’re going to be able to follow for the streaming, 00:13:58.900 --> 00:14:01.840 but what I’d like you to do is all come up and get around it. 00:14:01.840 --> 00:14:04.180 Because you need to see it from the top. 00:14:13.020 --> 00:14:18.380 Okay. So what I have here is this fishnet stocking, which is an elastic membrane, 00:14:18.389 --> 00:14:21.639 which is the crust of the Earth. So we no longer have a rubber band here 00:14:21.639 --> 00:14:26.350 because the rubberiness – the elasticity of the crust – is now in the medium. 00:14:26.350 --> 00:14:30.720 And so I have this little slider, which has got a little kind of sandpaper bottom. 00:14:30.720 --> 00:14:32.769 And I’m going to put it under here. 00:14:32.769 --> 00:14:35.829 And this is my two-bricker. And this is connected. 00:14:35.829 --> 00:14:39.420 And now we’re going to see how the two-bricker behaves. 00:14:40.020 --> 00:14:44.100 [intermittent quiet scratching sounds] 00:14:44.100 --> 00:14:46.520 So we still get a long wait between earthquakes, but now 00:14:46.520 --> 00:14:49.920 you’re seeing something else, right? In addition to the wait between 00:14:49.920 --> 00:14:53.899 earthquake, look at these wings here of high shear stress. 00:14:53.899 --> 00:14:56.639 We see this region here of tension, 00:14:56.639 --> 00:14:59.579 and behind it, we see the region of compression. 00:14:59.579 --> 00:15:04.769 So what we’re now seeing is this field of influence of this big earthquake. 00:15:04.769 --> 00:15:07.759 He’s no longer just transmitting in a linear fashion. 00:15:07.760 --> 00:15:11.020 We’re seeing this spatial relationship. 00:15:11.549 --> 00:15:13.560 So you can see right away that this guy 00:15:13.560 --> 00:15:16.860 has an influence about a couple fault lengths away. 00:15:17.280 --> 00:15:21.680 All right. So now let’s say – that’s our two-bricker. 00:15:21.680 --> 00:15:26.460 Now let’s put a one-bricker in here, and let’s get them far away from each other. 00:15:27.760 --> 00:15:31.619 And so, if things behave like they did here, this guy 00:15:31.620 --> 00:15:36.220 would go less frequently than this. So let’s see if that happens. 00:15:36.840 --> 00:15:42.960 [intermittent quiet scratching sounds] 00:15:42.960 --> 00:15:48.360 About two or three times more frequent, right, than this guy, and smaller always. 00:15:48.370 --> 00:15:50.699 So these are not interacting. 00:15:50.699 --> 00:15:54.209 This is exactly what you’d expect based on everything we just saw before. 00:15:54.209 --> 00:15:58.100 Okay, but what happens if I start to move them close together? 00:16:00.420 --> 00:16:05.080 So now I’m going to take these guys – and by the way, notice that my fishnet 00:16:05.089 --> 00:16:08.169 stocking has some runs and holes in it. That’s perfect. 00:16:08.169 --> 00:16:10.319 That’s just what the crust looks like. [laughter] 00:16:10.320 --> 00:16:13.620 Okay? So that’s intentional, kind of. 00:16:14.820 --> 00:16:19.059 Okay, so now they’re about one source – one fault dimension away. 00:16:19.059 --> 00:16:20.399 Now let’s see what happens. 00:16:20.400 --> 00:16:24.300 [intermittent quiet scratching sounds] 00:16:24.300 --> 00:16:28.850 So the little guys does maybe have one earthquake on its own. 00:16:28.850 --> 00:16:32.309 But whenever the big guy goes, the little guy goes with it. 00:16:32.309 --> 00:16:36.960 And in general, its behavior is heavily influenced by this. 00:16:36.960 --> 00:16:39.000 This guy has become a captive. 00:16:39.009 --> 00:16:41.220 That’s what the Hayward Fault is. 00:16:41.220 --> 00:16:44.639 It’s got – you know, the San Andreas is 1,000 kilometers long. 00:16:44.639 --> 00:16:46.749 The Hayward Fault is 100 kilometers long. 00:16:46.749 --> 00:16:50.660 San Andreas is 2 centimeters-per-year slip, or 20 millimeters. 00:16:50.660 --> 00:16:51.980 This is 10 millimeters. 00:16:51.980 --> 00:16:57.500 And yet, they have earthquake inter-event times that are roughly similar. 00:16:58.220 --> 00:17:01.140 That wouldn’t happen if the Hayward were on its own. 00:17:01.149 --> 00:17:05.209 The Hayward is feeling the influence of the San Andreas. 00:17:05.209 --> 00:17:09.520 And this kind of interaction is very important. 00:17:09.520 --> 00:17:12.120 And what we have to do is read that. 00:17:12.120 --> 00:17:16.880 Now, let me do another thing. So, okay, those are just two little guys. 00:17:16.880 --> 00:17:21.330 That isn’t really what a seismic catalog or an earthquake region looks like. 00:17:21.330 --> 00:17:27.410 So now what I want to do is I kind of just want to randomly put sliders in here. 00:17:27.410 --> 00:17:30.980 So here is one. 00:17:31.460 --> 00:17:32.980 Here’s another. 00:17:32.980 --> 00:17:37.000 And I don’t really care how big they are. 00:17:37.960 --> 00:17:43.860 [rustling and banging noises] 00:17:43.860 --> 00:17:45.900 Just load it up one more. 00:17:47.600 --> 00:17:48.920 Okay. 00:17:49.240 --> 00:17:55.800 [rustling and banging noises] 00:17:56.620 --> 00:18:01.820 God, these – neodymium magnets, just – like, you – it’s impossible to 00:18:01.820 --> 00:18:05.000 even get them – all right. So now let’s see what we see. 00:18:08.360 --> 00:18:12.790 This guy is so close to the edge that it’s feeling the frame, and I don’t want that. 00:18:13.880 --> 00:18:15.080 Okay. 00:18:15.920 --> 00:18:18.500 [intermittent scratching sounds] 00:18:18.500 --> 00:18:23.460 Now, those two big ones are now so strongly interacting, they’re in lock-step. 00:18:24.320 --> 00:18:28.800 This guy feels a little of these guys. These guys are on their own. 00:18:28.800 --> 00:18:31.450 So we can see that interactions are occurring. 00:18:31.450 --> 00:18:33.860 Whenever this one goes, this one goes. 00:18:33.860 --> 00:18:38.990 Now, if you close your eyes, it just sounds like random noise. 00:18:38.990 --> 00:18:41.890 That’s what we think we’re looking at in a catalog, 00:18:41.890 --> 00:18:45.530 but we can see here that there are a lot of interactions. 00:18:45.530 --> 00:18:47.440 And those are the interactions we care about. 00:18:47.440 --> 00:18:49.920 And that’s what we need to figure out. 00:18:49.920 --> 00:18:54.750 Reading those is our – is the challenge and is the opportunity in Earth science 00:18:54.750 --> 00:18:57.240 in terms of understanding earthquake occurrence. 00:18:57.900 --> 00:19:01.040 Now, I want to do one more thing. [loud bang] 00:19:01.040 --> 00:19:06.850 Okay, what did I do? I just put a seismic wave through the table. 00:19:06.850 --> 00:19:09.500 And I think we saw – let’s just do it one more time. 00:19:09.500 --> 00:19:11.760 [scratch sound] [bang] 00:19:11.770 --> 00:19:12.850 We saw an earthquake. 00:19:12.850 --> 00:19:17.130 Okay, so I didn’t add any shear stress to this system. 00:19:17.130 --> 00:19:21.720 What I did is I put a wave in. And right now, it’s doing nothing. 00:19:21.720 --> 00:19:25.210 Most of the time, it might do nothing. Sometimes it’ll do something. 00:19:25.210 --> 00:19:30.060 And this is what we call dynamic stressing. 00:19:32.300 --> 00:19:35.560 Just try this again. I’ll just stop it at random. 00:19:35.560 --> 00:19:37.900 [banging sounds] 00:19:37.910 --> 00:19:38.910 Okay. 00:19:38.910 --> 00:19:41.180 So it’s not very efficient, is it? 00:19:41.180 --> 00:19:45.180 Sometimes we get something to jump. That guy went this time. 00:19:45.180 --> 00:19:47.580 Didn’t go very far, either, did it? [loud bang] 00:19:47.580 --> 00:19:51.500 So this is an interesting phenomenon. And it’s the phenomenon in which, 00:19:51.500 --> 00:19:54.860 in addition to these permanent or static stresses that are set up in the 00:19:54.860 --> 00:20:00.760 elastic medium, that sending a seismic wave can, under some circumstances, 00:20:00.760 --> 00:20:05.390 take a guy that is very close to failure and do something which is probably 00:20:05.390 --> 00:20:11.580 not the application of shear stress – maybe this – and allow it to go. 00:20:11.580 --> 00:20:14.850 And that is also part of the puzzle that we have to read, and it’s 00:20:14.850 --> 00:20:19.690 a much harder and fascinating part. And we’re going to look at that too. 00:20:19.690 --> 00:20:23.310 So now you know everything I do. [laughter] 00:20:23.310 --> 00:20:26.820 And so now the question is, can we show that this is actually 00:20:26.820 --> 00:20:29.380 how the Earth works? And that’s what the slides are for. 00:20:29.380 --> 00:20:31.290 So thank you for standing around and seeing this. 00:20:31.290 --> 00:20:34.220 And now we’ll see it in the next part. 00:20:50.120 --> 00:20:53.260 I mentioned the Landers earthquake to you. 00:20:53.260 --> 00:20:57.090 Landers earthquake is kind of the Velveteen Rabbit of seismic science. 00:20:57.090 --> 00:21:00.470 We’ve loved it so much, its ears and eyes have fallen out. 00:21:00.470 --> 00:21:04.470 But this is this earthquake where three faults got together and ruptured. 00:21:04.470 --> 00:21:08.280 And here’s a calculation that my colleagues and I made 00:21:08.280 --> 00:21:15.120 after the earthquake of places where the Coulomb stress – and I want to just – 00:21:15.120 --> 00:21:16.540 even though we just dropped the lights, 00:21:16.540 --> 00:21:25.270 I want to just clarify what that Coulomb stress is, and I want to use this to do it. 00:21:25.270 --> 00:21:29.740 [rustling and banging sounds] 00:21:29.740 --> 00:21:32.060 Okay, what is the Coulomb stress? 00:21:32.060 --> 00:21:35.960 It’s the conditions that promote failure or inhibit failure. 00:21:35.960 --> 00:21:41.500 So, if this guy is just about ready to go, and I pull a little bit more on the spring, 00:21:41.500 --> 00:21:46.610 I’m increasing the shear stress, and it goes to failure. 00:21:46.610 --> 00:21:49.570 If it were not close to failure, pulling on the spring wouldn’t matter. 00:21:49.570 --> 00:21:52.710 It only works if it’s close to failure. By the same token, if it’s 00:21:52.710 --> 00:21:57.080 close to failure, and I lift the top brick, I unclamp it and bring it close to failure. 00:21:57.080 --> 00:22:00.450 So the Coulomb stress is just the combination of increasing the 00:22:00.450 --> 00:22:04.390 tension on the spring and lifting a brick. That’s red up here. 00:22:04.390 --> 00:22:10.030 And the opposite – if I throw on another brick or I loosen the spring, that’s blue. 00:22:10.030 --> 00:22:13.440 That’s what we call the stress shadow. Okay. That’s all that is. 00:22:16.430 --> 00:22:20.940 So you can see calculations of where the stress increased and decreased 00:22:20.940 --> 00:22:26.710 seemed to correspond to the first three hours or so of the aftershocks. 00:22:26.710 --> 00:22:30.300 And it’s interesting that over here at Big Bear, we have earthquakes 00:22:30.300 --> 00:22:34.280 that are well off the fault. There’s one up here – one or two. 00:22:34.280 --> 00:22:37.400 There’s one over here. And the rest are along the fault. 00:22:37.400 --> 00:22:41.280 Okay, now what’s interesting is that a 6.5 occurred there 00:22:41.280 --> 00:22:45.290 3 hours and 8 minutes later. So if this calculation had been made 00:22:45.290 --> 00:22:51.210 in advance, maybe it would have had some societally useful purposes. 00:22:51.210 --> 00:22:52.781 But you could still say, now, wait a minute. 00:22:52.781 --> 00:22:57.070 I could throw a dart at this map, and I have kind of a 50/50 chance of hitting 00:22:57.070 --> 00:23:03.030 a red zone, so big deal. But we have 10,000 aftershocks of Landers. 00:23:03.030 --> 00:23:06.310 And the vast majority of them do occur in these regions where 00:23:06.310 --> 00:23:10.560 we’ve calculated that we’ve lifted a brick or pulled on the spring. 00:23:11.220 --> 00:23:13.620 But, again, if you’re a skeptical reader, 00:23:13.620 --> 00:23:16.710 and you should be, you’d say, but wait a minute. You also have 00:23:16.710 --> 00:23:20.860 some earthquakes in the stress shadows. So that doesn’t make sense. 00:23:20.860 --> 00:23:24.310 Because, no matter how close it was to failure, you’ve just done something 00:23:24.310 --> 00:23:27.980 to reduce it from failure, and yet you’re getting earthquakes there. 00:23:27.980 --> 00:23:30.440 So doesn’t that falsify the whole hypothesis? 00:23:31.760 --> 00:23:34.090 And our answer is, no. What you expect in the 00:23:34.090 --> 00:23:38.300 stress shadows is a lower rate of earthquakes, not no earthquakes. 00:23:38.300 --> 00:23:41.150 And that turns out to be what we see. 00:23:41.150 --> 00:23:45.890 Now, about 10 years ago, we produced software so that everybody 00:23:45.890 --> 00:23:49.590 could make these kind of calculations. It’s MATLAB software. 00:23:49.590 --> 00:23:53.640 And a lot of analyses have been done on many earthquakes, 00:23:53.640 --> 00:23:56.620 and many people in the room have been involved in those. 00:23:57.340 --> 00:24:02.400 But early in my career, I had a set of interactions with David Jackson 00:24:02.400 --> 00:24:05.230 David Jackson was my Groundhog Day. 00:24:05.230 --> 00:24:08.150 Any great idea I had, he poured cold water on it. 00:24:08.150 --> 00:24:10.710 Every meeting I went to, no matter what part of the world, 00:24:10.710 --> 00:24:14.110 he would show up to pour cold water on it. [laughter] 00:24:14.110 --> 00:24:19.700 We had public debates. He basically wrecked all my nice days. 00:24:19.700 --> 00:24:22.740 And this went on for just about 30 years. 00:24:22.740 --> 00:24:26.470 And then, about five years ago, I sat back, and I thought, okay, 00:24:26.470 --> 00:24:31.630 in all of the arguments that Dave and I have had, he’s been mostly right. 00:24:31.630 --> 00:24:33.300 And I’ve been mostly wrong. 00:24:33.300 --> 00:24:36.130 And what I should be doing is working with him. 00:24:36.130 --> 00:24:40.820 And David’s principle point is that, what I just showed you at Landers, 00:24:40.820 --> 00:24:43.250 because the calculation is made after the earthquake, 00:24:43.250 --> 00:24:48.340 he would say, it doesn’t count. Only prospective testing counts. 00:24:48.960 --> 00:24:51.140 And, as we all know, that’s very hard. 00:24:51.140 --> 00:24:53.570 Right? Because you can’t cook the books. 00:24:53.570 --> 00:24:57.820 You can’t do anything to make the data look like the model. 00:24:57.820 --> 00:25:01.880 But it is the direction that clearly one needs to go. 00:25:01.880 --> 00:25:04.130 So I want to show you a couple of studies where we made 00:25:04.130 --> 00:25:07.360 10-year forecasts. This is the other advantage of being emeritus. 00:25:07.360 --> 00:25:11.950 I actually made a 10-year forecast, and I’m still in my career. [laughter] 00:25:11.950 --> 00:25:17.640 Okay, so what we’ve done to Coulomb analysis is to add another layer, 00:25:17.640 --> 00:25:23.210 which is borrowed from the work of great USGS scientist Jim Dieterich, 00:25:23.210 --> 00:25:27.870 and that is that the places where we have these stress trigger zones, 00:25:27.870 --> 00:25:31.720 they will amplify the background rate of seismicity. 00:25:31.720 --> 00:25:35.800 Places where have shadow zones will suppress the background rate. 00:25:35.800 --> 00:25:39.230 So you don’t – you’re not – just knowing what the Coulomb 00:25:39.230 --> 00:25:41.650 stress changes alone aren’t enough. 00:25:41.650 --> 00:25:44.190 You need to know what the background seismicity is. 00:25:44.190 --> 00:25:46.900 And here’s such a calculation for Landers. 00:25:46.900 --> 00:25:51.610 The red areas are where we forecast we should see a higher rate or see earthquakes. 00:25:51.610 --> 00:25:54.070 And the white areas are where we expect no earthquakes. 00:25:54.070 --> 00:25:58.990 And you can see there’s a fairly good correlation after the fact. 00:25:58.990 --> 00:26:01.370 So here’s the area for which we made a forecast. 00:26:01.370 --> 00:26:04.000 And here is the forecast. 00:26:04.000 --> 00:26:08.530 So we said, for magnitude 5 in this 10-year period, you can see 00:26:08.530 --> 00:26:13.170 that no box has anything like 100% chance. It’s up to 15%. 00:26:13.170 --> 00:26:17.460 But as a whole, this whole area is 92%. And that box in the 00:26:17.460 --> 00:26:21.890 lower left-hand corner is 82%. Okay, how’d we do? 00:26:21.890 --> 00:26:25.500 So this is a published forecast. We can’t do anything to change it. 00:26:25.500 --> 00:26:31.150 Well, there was a magnitude 5.1 in the box. So that 92% looks good. 00:26:31.150 --> 00:26:35.540 And we almost got to 5 in that corner. We have a couple of near-5s. 00:26:35.540 --> 00:26:37.880 So that’s maybe okay. 00:26:37.880 --> 00:26:41.850 If we go a little deeper and say, well, how does this forecast compare to 00:26:41.850 --> 00:26:47.120 the actual seismicity that occurred, you can see, well, it does pretty well 00:26:47.120 --> 00:26:50.580 in the lower left-hand corner, but we seem to miss a lot of – 00:26:50.580 --> 00:26:54.300 there were a lot more earthquakes out at the Hector Mine area 00:26:54.310 --> 00:26:59.350 because we low-balled the rate at which it should produce earthquakes. 00:26:59.350 --> 00:27:03.000 So I would say, overall, this forecast did okay. 00:27:03.920 --> 00:27:06.630 All right. We have another one we can do, and that was one 00:27:06.630 --> 00:27:11.160 we produced after the 2008 Wenchuan earthquake. 00:27:11.160 --> 00:27:14.900 And a year ago, we had an earthquake in Gansu, China, 00:27:14.900 --> 00:27:18.750 which sadly destroyed this national park. 00:27:18.750 --> 00:27:22.940 But here is the – here is the background rate of earthquakes 00:27:22.940 --> 00:27:25.850 during the eight years before the main shock. 00:27:25.850 --> 00:27:31.410 So if Wenchuan didn’t occur, we would, more or less, expect that to continue. 00:27:31.410 --> 00:27:35.080 And then here is the stress impact of the Wenchuan earthquake. 00:27:35.080 --> 00:27:37.280 So the difference between these two represents 00:27:37.280 --> 00:27:43.120 the effect of the stretching of that fishnet stocking. 00:27:43.120 --> 00:27:45.210 And you can see that there are some areas here that get 00:27:45.210 --> 00:27:48.890 much redder than they were before down here. 00:27:48.890 --> 00:27:52.760 And other areas that look more or less the same in both places. 00:27:52.760 --> 00:27:53.980 And here’s what happened. 00:27:53.980 --> 00:28:01.080 We’ve had two earthquakes larger than 6 in that 10-year period, both in red zones. 00:28:01.080 --> 00:28:04.420 So, again, this is one where we couldn’t cook the books. 00:28:04.420 --> 00:28:07.710 All right, here’s one more on a much faster time scale – 00:28:07.710 --> 00:28:13.120 a magnitude 6.2 earthquake in Kumamoto, Japan, on April 15th. 00:28:13.120 --> 00:28:18.000 My colleague and co-author Shinji Toda did this analysis, and he said, well, 00:28:18.000 --> 00:28:24.600 both the Futagawa Fault and the Hinagu Fault are now in stress-triggering lobes. 00:28:24.600 --> 00:28:28.940 So he said, well, this looks like it could be a source of a trigger. 00:28:28.940 --> 00:28:34.340 And he presented this to the press in a press conference, and it was published. 00:28:34.340 --> 00:28:39.220 And the next day, there was a magnitude 7 rupturing both of those faults. 00:28:41.640 --> 00:28:47.320 So, in a nutshell, this is my – this was my worldview up to 2012, 00:28:47.320 --> 00:28:53.240 that, as far as I could tell, most of what we were seeing was static stress-related, 00:28:53.240 --> 00:28:56.830 even though there was a lot of work done by many USGS and other scientists 00:28:56.830 --> 00:29:01.690 on the dynamic triggering – the banging. Didn’t look very convincing. 00:29:01.690 --> 00:29:04.520 But I had my comeuppance by this earthquake. 00:29:05.160 --> 00:29:10.420 So this event – largest strike-slip earthquake ever recorded, 00:29:10.420 --> 00:29:13.700 probably by a factor 10 – very, very strange-looking earthquake. 00:29:13.700 --> 00:29:18.620 It does not look like a San Andreas rupture. Off in the Indian Ocean. 00:29:18.620 --> 00:29:23.420 And I was at dinner with Sharon, my wife, at a Turkish restaurant 00:29:23.420 --> 00:29:27.210 in Menlo Park, and I kept getting buzzed by earthquakes. 00:29:27.210 --> 00:29:29.930 And I thought, well, I’ve just been sensitized 00:29:29.930 --> 00:29:33.200 to earthquakes as a result of this main shock. 00:29:33.200 --> 00:29:35.490 But, over the next six days, 00:29:35.490 --> 00:29:40.140 here are the magnitude 5.5-and-larger earthquakes that occurred. 00:29:40.140 --> 00:29:43.880 Now, you look at this, and you go, well, first of all, they’re uniformly 00:29:43.880 --> 00:29:47.330 spread around the world. They’re not clustered, 00:29:47.330 --> 00:29:50.880 or they don’t tend to be close to the main shock. That looks odd. 00:29:50.880 --> 00:29:53.790 And they're – the places – these are the usual suspects. 00:29:53.790 --> 00:29:56.210 These are places where earthquakes always occur. 00:29:56.210 --> 00:30:00.140 So could these possibly be, in any sense, global aftershocks? 00:30:00.140 --> 00:30:05.480 Well, here’s the convincing – here’s what convinced me. 00:30:05.490 --> 00:30:09.510 That the rate of magnitude 5.5 earthquakes, 00:30:09.510 --> 00:30:13.730 even if we exclude anything within 1,500 kilometers of the main shock, 00:30:13.730 --> 00:30:17.260 rocketed up and then quickly decayed. 00:30:17.260 --> 00:30:21.000 Now, what’s interesting about this plot is I went into my office, and I compared 00:30:21.000 --> 00:30:25.650 the previous week of global earthquakes to the day or two afterwards. 00:30:25.650 --> 00:30:29.440 I saw it was higher. Maybe as high as a factor of 10. 00:30:29.440 --> 00:30:32.640 And I thought, this is all within the noise. 00:30:32.640 --> 00:30:34.210 Fred Pollitz came into my office, 00:30:34.210 --> 00:30:35.820 and he said, did you see these global aftershocks? 00:30:35.820 --> 00:30:38.730 No, they’re not real. And he went into his office, and he 00:30:38.730 --> 00:30:43.090 showed me the same plot that I did, and suddenly I realized, I’m wrong. 00:30:43.090 --> 00:30:48.320 I’m wrong because this was one of the moments when all of your judgment 00:30:48.320 --> 00:30:54.020 and wisdom turns out to be bias. And this is a risk that we all run. 00:30:54.020 --> 00:30:57.120 Everything I thought I knew, I wasn’t willing to accept 00:30:57.120 --> 00:31:00.670 something different when it was staring at me in the face. 00:31:00.670 --> 00:31:03.000 And had I not had Fred Pollitz as a colleague, 00:31:03.000 --> 00:31:05.400 I would have missed it altogether. 00:31:06.120 --> 00:31:08.930 So we should all be on alert for this. 00:31:09.320 --> 00:31:13.600 Now, you look at this, and you go, okay, if a measly 8.6 can do this, 00:31:13.610 --> 00:31:16.360 then what about the 9.2? What about the 9.0? 00:31:16.360 --> 00:31:20.250 We’ve had a lot of big earthquakes in the last several years. 00:31:20.250 --> 00:31:24.170 And even if you stack all those earthquakes, 00:31:24.170 --> 00:31:28.480 now you get kind of a global aftershock Rorschach test. 00:31:28.480 --> 00:31:32.290 If you believe it, maybe there’s something here, but it’s pretty subtle. 00:31:32.290 --> 00:31:34.670 Maybe there’s nothing there. 00:31:34.670 --> 00:31:36.860 Why is this earthquake different? 00:31:36.860 --> 00:31:41.120 Well, the earthquake is different because it’s a strike-slip event. 00:31:41.120 --> 00:31:44.760 And strike-slip events launch huge Love waves. 00:31:44.760 --> 00:31:51.360 The Love waves of this 8.6 are 2 to 3 times larger than the 9.2. 00:31:52.150 --> 00:31:57.500 And so what strikes me – struck me about that is, okay, 00:31:57.500 --> 00:32:00.070 as you know, I’m a very sophisticated scientist, so I have 00:32:00.070 --> 00:32:04.100 a lot of – I have a lot of rubber bands, and I have a lot of sticky dots. 00:32:04.100 --> 00:32:08.520 So I put sticky dots on – where all these global aftershocks were, 00:32:08.520 --> 00:32:12.460 and I found I could put a rubber band through all of them – 00:32:12.460 --> 00:32:15.850 perpendicular rubber band that goes through the main shock. 00:32:15.850 --> 00:32:20.510 So I ran into Fred’s office. This is very big. I have a lot of globes. 00:32:20.510 --> 00:32:23.770 I ran into Fred’s office with my sticky dots and said, Fred, look at this! 00:32:23.770 --> 00:32:29.460 And he goes, oh, yeah, that’s the Love wave elastodynamic propagation. 00:32:29.460 --> 00:32:33.380 And he – then he comes over to my office, like, the next day, with this. 00:32:33.380 --> 00:32:34.540 [laughter] 00:32:34.550 --> 00:32:39.680 Now, I ask you, which is better? [laughter] 00:32:39.680 --> 00:32:43.150 His showoff-y thing or my high school science fair – 00:32:43.150 --> 00:32:46.150 maybe middle school science fair version? [laughter] 00:32:46.150 --> 00:32:50.309 Okay, but Fred has done a very interesting trick here. 00:32:50.309 --> 00:32:53.770 This is somehow the strain, but notice how the strain 00:32:53.770 --> 00:32:58.580 isn’t very much higher at the source than very far away. 00:32:58.580 --> 00:33:02.410 So what Fred has done is something more devious than that. 00:33:02.410 --> 00:33:07.290 He’s just saying, what is the duration of the strain wave above a rather 00:33:07.290 --> 00:33:11.160 low threshold? And that’s why you see it distributed around the world. 00:33:11.160 --> 00:33:14.480 So if this were really what matters, it means it’s not so much 00:33:14.480 --> 00:33:20.910 how hard you hit. It’s how long you hit that plays a role here. 00:33:20.910 --> 00:33:26.400 Now, if Love waves really were the reason why this earthquake was unique, 00:33:26.400 --> 00:33:28.870 then you have to think about, well, what do Love waves do 00:33:28.870 --> 00:33:30.380 They do this. 00:33:30.380 --> 00:33:34.110 So if you’re a vertical strike-slip fault, you can promote failure. 00:33:34.110 --> 00:33:36.390 If you’re right- or left-lateral, sooner or later, it’s going to push 00:33:36.390 --> 00:33:40.929 you one way or the other. So that suggests that, if these really 00:33:40.929 --> 00:33:47.320 are triggered by Love waves, then these global aftershocks should be strike-slip. 00:33:47.320 --> 00:33:49.200 And that’s what you see. 00:33:49.200 --> 00:33:53.600 Typically, only a quarter of the global shocks are strike-slip. 00:33:53.600 --> 00:34:00.300 But during the days after this event, up to three-quarters of them are strike-slip. 00:34:00.300 --> 00:34:04.150 So that seems to hang together. And notice too that, although we think in 00:34:04.150 --> 00:34:10.249 terms of about a week, even 12 days in, we’re still seeing some action here. 00:34:10.249 --> 00:34:14.860 Maybe isn’t over as quick as one might think. 00:34:14.860 --> 00:34:19.190 And there’s one more picture to this story, and it’s this. 00:34:19.190 --> 00:34:23.669 If you look at the number of 5.5s around the world globally. 00:34:23.669 --> 00:34:26.129 You can see that, right before this earthquake, 00:34:26.129 --> 00:34:32.680 there was a very, very low rate – the lowest rate in about 10 years. 00:34:32.680 --> 00:34:34.620 And none of the other big earthquakes 00:34:34.620 --> 00:34:38.139 occurred during a time of a very low rate. 00:34:38.740 --> 00:34:40.839 Okay, so what does that suggest? 00:34:40.839 --> 00:34:45.720 Imagine you have an apple tree. And the apple tree is ripening apples. 00:34:45.720 --> 00:34:47.800 And as the apples ripen, they drop. 00:34:47.800 --> 00:34:51.540 So let’s say you’re normally dropping at one or two a day. 00:34:52.480 --> 00:34:56.620 For whatever reason, you go through a week where no apples drop at all. 00:34:56.629 --> 00:34:59.029 And now I run up and I shake the trunk. 00:34:59.029 --> 00:35:02.090 Shaking the trunk is that Love wave. 00:35:02.090 --> 00:35:05.990 Those ripening apples are faults getting closer and closer to failure. 00:35:05.990 --> 00:35:11.109 So maybe, if you’ve, by accident, had a period where you don’t have 00:35:11.109 --> 00:35:15.640 many earthquakes occurring, then the effect is much more pronounced. 00:35:16.770 --> 00:35:21.840 So that’s what I thought at the end of this work, that, okay, 00:35:21.859 --> 00:35:26.359 dynamic triggering of large earthquakes is possible but extremely rare. 00:35:26.359 --> 00:35:28.980 We will probably never see it again in our lifetime because it takes 00:35:28.980 --> 00:35:32.489 a very, very large strike-slip earthquake. We looked at every other large 00:35:32.489 --> 00:35:36.230 historical strike-slip earthquake we could, obviously including 1906. 00:35:36.230 --> 00:35:39.260 None of them showed anything. 00:35:39.620 --> 00:35:41.920 And then I was wrong again. 00:35:41.920 --> 00:35:47.180 Because look at this. This is the Kos-Bodrum earthquake 00:35:47.180 --> 00:35:52.430 that occurred last summer. And, if you look at the first 16 hours, 00:35:52.430 --> 00:35:55.430 you see this kind of amazing jet of earthquakes. 00:35:55.430 --> 00:35:59.700 These are small, but up to 400 kilometers away. 00:36:00.240 --> 00:36:03.540 And Fred did a dynamic model for that and said, well, under certain 00:36:03.540 --> 00:36:09.700 circumstances, you can get dynamic stresses that propagate in that direction. 00:36:10.420 --> 00:36:15.400 Now, Tom Parsons and Joan Gomberg and Debi Kilb, 00:36:15.410 --> 00:36:19.640 and many people have looked at other evidence for dynamic triggering, 00:36:19.640 --> 00:36:24.509 most of that for very small earthquakes. What’s unique about the 2012 case 00:36:24.509 --> 00:36:28.589 is that some of these are large – as large as magnitude 7. 00:36:28.589 --> 00:36:31.480 And I should just make the other comment that you may not know, 00:36:31.480 --> 00:36:35.630 that if you get earthquake insurance, it’s backed by re-insurance companies 00:36:35.630 --> 00:36:40.400 whose entire strategy is to globally diversify your earthquake risk 00:36:40.400 --> 00:36:42.970 so no one earthquake can affect another. 00:36:42.970 --> 00:36:46.890 Which, during this six days after the 2012 earthquake, 00:36:46.890 --> 00:36:51.029 was completely destroyed, right? Because they were all correlated. 00:36:51.029 --> 00:36:53.029 All right. 00:36:53.029 --> 00:36:55.859 And then, another thing happened that we were all aware of – 00:36:55.860 --> 00:37:01.940 this huge sequence of earthquakes in Mexico in September. 00:37:01.940 --> 00:37:06.640 And many of these shocks can be explained by static stress, 00:37:06.650 --> 00:37:12.400 such as this aftershock of the Tehuantepec earthquake, 00:37:12.400 --> 00:37:15.289 this aftershock of the Pinotepa Nacional. 00:37:15.289 --> 00:37:18.630 This itself is kind of a late aftershock of another earthquake of about the same 00:37:18.630 --> 00:37:24.269 size that occurred here. But what about the relationship between these two? 00:37:24.269 --> 00:37:28.960 They are 615 kilometers apart and 12 days apart. 00:37:28.960 --> 00:37:31.360 Static stress is out. 00:37:31.369 --> 00:37:34.609 We calculated – Tom Parsons independently calculated something like 00:37:34.609 --> 00:37:38.630 a 1 in 3,500 chance that these are coincidental, just because of the 00:37:38.630 --> 00:37:42.279 background rate of occurrence of earthquakes of this size. 00:37:42.279 --> 00:37:45.769 That might seem like a large number to you, but do you remember 00:37:45.769 --> 00:37:51.829 that the Tehuantepec earthquake – the magnitude 8.2 – occurred two hours 00:37:51.829 --> 00:37:59.890 after the national earthquake emergency alarm – you know, the training exercise? 00:37:59.890 --> 00:38:02.720 You know what the chances that an earthquake of that size would 00:38:02.720 --> 00:38:07.059 occur within two hours of an annual alarm? It’s 1 in a million. 00:38:07.059 --> 00:38:10.630 So 1 in 3,500 may not be a big chance. 00:38:10.630 --> 00:38:13.550 But years ago, when I was a grad student, my mother, 00:38:13.550 --> 00:38:19.180 who’s a calligrapher, calligraphed something from Alfred North Whitehead. 00:38:20.480 --> 00:38:25.200 And so maybe we shouldn’t just leave it at that. 00:38:25.200 --> 00:38:28.109 So again, I’d go back to the source, Fred. 00:38:28.109 --> 00:38:33.240 And I say, Fred, do a dynamic stress animation of this earthquake. 00:38:34.520 --> 00:38:38.880 And what’s interesting – it’s kind of like a gun pointed at Puebla. 00:38:38.890 --> 00:38:44.650 And the dynamic unclamping is 6-1/2 bars – twice the shear stress. 00:38:44.650 --> 00:38:49.079 It’s a very large number. Of course, the total dynamic – 00:38:49.079 --> 00:38:55.140 the total clamping that that fault is under at 50 kilometers’ depth is on the order of 00:38:55.140 --> 00:39:00.400 5 or 10 times higher. It’s not enough to actually pull any of these faces apart. 00:39:00.400 --> 00:39:07.880 But it’s a large stress change that lasts for something like 20 or 30 seconds. 00:39:07.880 --> 00:39:10.540 And the other thing that’s interesting – whoop – 00:39:10.540 --> 00:39:15.360 about it is that there are lots of sources for this Puebla. 00:39:15.369 --> 00:39:18.589 There are a lot of other focal mechanisms all like that here. 00:39:18.589 --> 00:39:22.200 And it chose this one, which is the one in the line of fire. 00:39:23.220 --> 00:39:28.690 So could this be evidence of delayed dynamic triggering? 00:39:28.690 --> 00:39:32.549 You know, the hypothesis one could advance is that, if you do this, 00:39:32.549 --> 00:39:36.720 maybe you can get fluids into the fault. And if you can get fluids into the fault 00:39:36.720 --> 00:39:42.730 and then start diffusing out, maybe that would explain the 12-day delay. 00:39:43.940 --> 00:39:47.200 I don’t know what the answer is here, but I’ve learned enough to know 00:39:47.200 --> 00:39:50.670 that I don’t have the right answer, and I can often be wrong. 00:39:50.670 --> 00:39:55.220 So our goal is, what we understand, build into our models. 00:39:55.220 --> 00:39:58.340 What we don’t understand, keep turning it over. 00:39:58.340 --> 00:40:00.519 Keep looking to see if there’s a relationship 00:40:00.519 --> 00:40:03.440 right in front of you that we might have missed. 00:40:03.440 --> 00:40:08.599 So I want to leave you with the idea that it seems to me that most earthquakes 00:40:08.599 --> 00:40:12.289 communicate by static stress – everything we saw over there – 00:40:12.289 --> 00:40:18.640 but that under unique circumstances, dynamic stress also plays a role. 00:40:18.640 --> 00:40:22.259 And I’d finally like to thank my wonderful colleagues and 00:40:22.260 --> 00:40:28.320 co-authors who helped build the demo tools and the papers that I’m describing. 00:40:28.320 --> 00:40:29.800 Thank you. 00:40:29.800 --> 00:40:36.200 [Applause] 00:40:36.380 --> 00:40:39.160 - Any questions for Ross? 00:40:40.400 --> 00:40:43.300 [Silence] 00:40:43.940 --> 00:40:46.700 None? Everyone understood everything? 00:40:46.700 --> 00:40:48.420 [laughter] 00:40:49.560 --> 00:40:57.500 [Silence] 00:40:58.420 --> 00:41:00.200 - [inaudible] - I’m sorry. 00:41:00.200 --> 00:41:03.920 All right, Ross. It’s a great talk, and there’s lot of interesting things there. 00:41:03.920 --> 00:41:08.729 I just want to sort of publicly criticize one piece of 00:41:08.729 --> 00:41:13.150 statistical comments that you made, which is, like, oh, the odds of 00:41:13.150 --> 00:41:17.059 these two earthquakes happening together is 1 in 3,500. 00:41:17.060 --> 00:41:20.900 Several years ago, you know, there was a – we had two times that – 00:41:20.900 --> 00:41:22.999 or maybe it just once that, you know, there was, like, 00:41:23.000 --> 00:41:27.130 two magnitude 7s in the world within a day. 00:41:27.660 --> 00:41:29.940 And, you know, people were talking about how odd it was. 00:41:29.940 --> 00:41:34.660 But the fact of the matter is, the chance of that happening was – 00:41:34.660 --> 00:41:36.769 you know, I forget, it was, like, a few times in a century. 00:41:36.769 --> 00:41:39.150 And it was happening a few times in a century. 00:41:39.150 --> 00:41:43.700 So, I mean, the interesting thing is that almost anything we observe 00:41:43.700 --> 00:41:45.829 is very unlikely. That’s why, when people 00:41:45.829 --> 00:41:49.440 calculate likelihoods, they’re incredibly small numbers. 00:41:49.440 --> 00:41:53.720 Because they’re the probability of an exact occurrence happening. 00:41:53.720 --> 00:41:56.380 And really, whenever we look at these sort of coincidences – 00:41:56.380 --> 00:41:58.520 and I’m not saying there isn’t a physical link, but whenever we 00:41:58.520 --> 00:42:00.970 look at a coincidence, we have to look at it 00:42:00.970 --> 00:42:04.069 across a much bigger spectrum of possibilities. 00:42:04.069 --> 00:42:09.000 Do these sorts of coincidences happen more often than we would expect? 00:42:09.000 --> 00:42:11.420 Not what is the probability of this coincidence. 00:42:11.420 --> 00:42:13.450 And I think that’s something that people tend to do 00:42:13.450 --> 00:42:17.820 and just sort of – we need to try to stop doing that. 00:42:18.440 --> 00:42:21.940 - Okay, this is a very good point. Let me – I … 00:42:26.620 --> 00:42:29.800 I have … - I did like – I did like your analogy 00:42:29.809 --> 00:42:35.089 to the two-hour gap after the clearly unrelated [laughs] warning. 00:42:35.089 --> 00:42:38.309 - Yeah. So point well-taken. I’m not arguing that I know 00:42:38.309 --> 00:42:41.739 that these are related. I’m saying, we should just 00:42:41.740 --> 00:42:47.660 explore what possible relationship – if they are related, how? 00:42:47.660 --> 00:42:52.630 Now, here’s the calculation we made for whether or not they’re coincidental. 00:42:52.630 --> 00:42:58.309 So we shuffled the catalog origin times, searching for pairs of 00:42:58.309 --> 00:43:04.819 a 7 followed by an 8 within 300 to 700 kilometers over two weeks. 00:43:04.819 --> 00:43:07.260 Broad numbers – not the exact case. 00:43:07.260 --> 00:43:10.660 We did 10,000 shuffles, and we got 1 in 3,500. 00:43:10.660 --> 00:43:14.380 Then we said, okay, what’s the chance that the 7 would strike 00:43:14.380 --> 00:43:19.180 within 100 kilometers of Puebla within two weeks of some other event? 00:43:19.180 --> 00:43:22.860 And we got 1 chance in 500 – much, much lower. 00:43:22.860 --> 00:43:27.960 Now, a sanity check is if you multiply the birthday probability, 00:43:27.970 --> 00:43:34.079 or the birthday paradox problem, then the numbers come about the same. 00:43:34.079 --> 00:43:36.569 So what is the birthday paradox problem? 00:43:36.569 --> 00:43:40.559 I was born on September 28th. Anybody in the room have my birthday? 00:43:41.340 --> 00:43:43.640 Okay, for a group of this size, there’s a very high probability 00:43:43.640 --> 00:43:47.440 someone will have my birthday, okay? Even though there aren’t 365 of us. 00:43:47.440 --> 00:43:49.930 Didn’t happen here, but it happens quite a bit. 00:43:49.930 --> 00:43:54.839 So that’s another case – this is really the birthday paradox problem, 00:43:54.839 --> 00:43:59.589 but the birthday is 12 days long, and the room is 600 kilometers wide. 00:43:59.589 --> 00:44:04.960 So these numbers may not be correct. They’re our best effort to try to 00:44:04.960 --> 00:44:09.400 answer that question. And I am not saying that they are not coincidental. 00:44:10.240 --> 00:44:12.920 - So, okay, is that the global catalog you shuffled? 00:44:12.920 --> 00:44:14.540 - The … 00:44:16.640 --> 00:44:18.340 No. It’s global. 00:44:18.340 --> 00:44:21.029 - So – yeah, I just – it was the global catalog. 00:44:21.029 --> 00:44:25.749 And so, for – over what period of time? For how – over 100 years, or … 00:44:25.749 --> 00:44:30.660 - So this is – we’re looking at two-week periods, and we just took the full 00:44:30.660 --> 00:44:33.710 catalog, and we did 10,000 shuffles of it. - Okay. 00:44:33.710 --> 00:44:35.819 I’ll take a look at it. All right. 00:44:35.819 --> 00:44:39.740 - So it’s not that important. I’m not saying – you know, 00:44:39.740 --> 00:44:44.840 as we’ve said, yes, we need to worry about unlikely events. 00:44:46.880 --> 00:44:51.460 [Silence] 00:44:52.500 --> 00:44:54.100 - Hi, Ross. That was a great talk. 00:44:54.100 --> 00:44:57.720 And I now know what I’m going to do for my next home construction project. 00:44:57.720 --> 00:45:01.440 I’m going to build your slider. 00:45:01.440 --> 00:45:04.700 I think that, you know, you’ve raised a lot of interesting points. 00:45:04.700 --> 00:45:08.150 I’m going to make a case for understanding earthquake physics 00:45:08.150 --> 00:45:11.279 a little better than represented by your slider. 00:45:11.280 --> 00:45:14.380 And I think it is important – I’m glad you mentioned the possible 00:45:14.380 --> 00:45:18.729 dampening effect of water diffusing into the faults for Puebla earthquake. 00:45:18.729 --> 00:45:21.989 We know that fluids are involved in faulting a lot. 00:45:21.989 --> 00:45:26.489 And so it’s important in looking at these dynamic and static stress models 00:45:26.489 --> 00:45:30.880 to include the fluid pressure response to stress changes – mean stress, 00:45:30.880 --> 00:45:34.470 normal stress in particular. And also to think about frictional melting. 00:45:34.470 --> 00:45:38.170 You know, we see – my hackles got raised a little bit when you 00:45:38.170 --> 00:45:40.819 mentioned that, you know, stress drops are a tiny fraction 00:45:40.819 --> 00:45:43.250 of the resolved shear stress. That’s not true for mature faults 00:45:43.250 --> 00:45:46.819 in which they can be a great fraction – most of that stress gets relieved 00:45:46.819 --> 00:45:48.799 during earthquakes, as shown by Art Lachenbruch 00:45:48.799 --> 00:45:50.970 and John Sass in their heat flow observations. 00:45:50.970 --> 00:45:54.089 So I think it is important to think about physics. 00:45:54.089 --> 00:45:56.849 I know it’s important to think about physics. 00:45:56.849 --> 00:45:58.960 Field evidence shows fluids are involved in faulting. 00:45:58.960 --> 00:46:01.890 We have induced seismicity to demonstrate this. 00:46:01.890 --> 00:46:04.869 We have various time-dependent effects that are best explained by it. 00:46:04.869 --> 00:46:08.769 And also to think about melting. Frictional melting, fluid pressurization, 00:46:08.769 --> 00:46:11.190 dynamic weakening. What would make a big earthquake 00:46:11.190 --> 00:46:14.789 behave very differently than a small one might influence – 00:46:14.789 --> 00:46:17.460 probably would influence some of the inferences drawn from 00:46:17.460 --> 00:46:20.380 a sliding block room-temperature experiment. 00:46:20.380 --> 00:46:22.779 So just putting that in and interested in your thoughts on that. 00:46:22.779 --> 00:46:26.619 - Yes. I think those are all important elements. 00:46:26.619 --> 00:46:31.779 I would say that, if most earthquakes had complete stress drops, 00:46:31.779 --> 00:46:35.890 then aftershock focal mechanisms would almost be random. 00:46:35.890 --> 00:46:38.200 And that’s only seen in a few earthquakes, 00:46:38.200 --> 00:46:41.930 like Loma Prieta or Tohoku. So I would still argue that 00:46:41.930 --> 00:46:46.690 most of the time, the earthquake stress drop is still rather small. 00:46:46.690 --> 00:46:51.359 The problem, of course, with fluids is it’s very difficult to have any direct 00:46:51.359 --> 00:46:58.789 observation of the fluid and its role in the process of either rupture or restressing. 00:46:58.789 --> 00:47:01.119 But I agree with you that there’s an enormous amount 00:47:01.119 --> 00:47:05.329 of evidence that it is important. Our software does allow you to 00:47:05.329 --> 00:47:11.089 play with Skempton’s coefficient. And some others have made plug-in modules 00:47:11.089 --> 00:47:15.380 for the software so that it – actual fluid diffusion could be included in it. 00:47:15.380 --> 00:47:19.540 So I think, you know, that’s just one of the areas where we can do better. 00:47:20.880 --> 00:47:25.320 [Silence] 00:47:26.280 --> 00:47:31.360 - Ross, your fishnet stocking demonstration caught my eye 00:47:31.369 --> 00:47:36.930 when you said that the Hayward Fault is really determined by what the 00:47:36.930 --> 00:47:42.029 larger fault next to it does. So that makes sense when I look 00:47:42.029 --> 00:47:45.619 at your model, but historically, do you see that happening? 00:47:45.619 --> 00:47:49.700 I mean, does the Hayward follow big earthquakes 00:47:49.700 --> 00:47:55.249 on the San Andreas with any … - So I think it’s difficult to know. 00:47:55.249 --> 00:48:00.239 Because we can’t – I think if you lined up prehistoric San Andreas events 00:48:00.239 --> 00:48:03.910 and prehistoric Hayward events, the uncertainties would kind of 00:48:03.910 --> 00:48:05.390 overwhelm you, although maybe somebody 00:48:05.390 --> 00:48:08.009 in the room can tell me the answer to that question. 00:48:08.009 --> 00:48:14.250 I would say, just in general, that the repeat – the average inter-event 00:48:14.250 --> 00:48:19.289 times are – appear to be rather similar despite the fact that one fault’s short 00:48:19.289 --> 00:48:23.800 and one fault’s long. Because they’re relatively close to each other. 00:48:23.800 --> 00:48:27.020 So I’m not saying that every time the San Andreas goes, 00:48:27.029 --> 00:48:31.829 the Hayward goes, or vice versa, but that they are influencing 00:48:31.829 --> 00:48:37.480 each other and that we can’t think of Hayward in isolation. 00:48:39.340 --> 00:48:47.920 [Silence] 00:48:48.920 --> 00:48:50.920 - Any other questions? 00:48:52.220 --> 00:48:55.440 I had a more general one. So you showed some successes 00:48:55.440 --> 00:48:59.839 of models, you know, seeming to correlate with the aftershocks. 00:48:59.840 --> 00:49:02.500 What about failures? [laughter] 00:49:02.500 --> 00:49:05.460 Well, that’s why prospective testing is so important. 00:49:05.479 --> 00:49:08.369 David Jackson would say, oh, if you don’t get a good result, 00:49:08.369 --> 00:49:15.120 you don’t publish it. And so therefore, you kind of bias yourselves. 00:49:15.120 --> 00:49:20.940 Tom Parson has done several global studies that show that at least shear 00:49:20.950 --> 00:49:26.509 stress is correlated with aftershocks. There was, just published this month, 00:49:26.509 --> 00:49:30.349 a study called Deep Learning of Aftershock Patterns Following 00:49:30.349 --> 00:49:34.140 Large Earthquakes. Very interesting study in which 00:49:34.140 --> 00:49:38.309 they tested Coulomb stress against several other potential 00:49:38.309 --> 00:49:42.320 stress components to explain aftershock occurrence in, I think, 00:49:42.320 --> 00:49:47.960 165 global earthquakes, putting in the best source model for each one of them. 00:49:47.969 --> 00:49:51.599 And they argued that Coulomb stress really didn’t do very well, 00:49:51.599 --> 00:49:55.809 although it was a component of their neural network output, and that 00:49:55.809 --> 00:50:00.980 maximum shear stress and the second invariant actually did better. 00:50:00.980 --> 00:50:05.819 So this is that kind – even though it’s not prospective, it’s a global study. 00:50:05.819 --> 00:50:08.670 So very, very important. What’s interesting is, if you take 00:50:08.670 --> 00:50:14.950 a look at some of their cases, and you look at how Coulomb 00:50:14.950 --> 00:50:16.479 does compared to the neural network, 00:50:16.479 --> 00:50:20.509 you’re not overwhelmed that the neural network has really hit it here. 00:50:20.509 --> 00:50:25.650 And the problem with what they’ve done is, by assuming that the receiver planes – 00:50:25.650 --> 00:50:29.869 the faults on which stress is calculated – are all parallel to the main fault, 00:50:29.869 --> 00:50:34.700 that automatically produces a very large negative area on the fault, 00:50:34.700 --> 00:50:37.520 or near the fault, where most aftershocks are occurring. 00:50:37.520 --> 00:50:41.019 So you’ve set yourself up to penalize that. 00:50:41.019 --> 00:50:43.999 Also, Coulomb stress is the only one that has negative lobes, 00:50:44.000 --> 00:50:46.480 not just positive lobes. And if you’re only measuring 00:50:46.480 --> 00:50:50.640 aftershocks, you can’t actually see a seismicity rate decrease. 00:50:50.640 --> 00:50:55.279 So I think they’ve set up the problem in a manner that makes it very hard to see it, 00:50:55.279 --> 00:51:00.219 in part because the world isn’t populated by Venetian blind faults. 00:51:00.219 --> 00:51:02.160 And particularly when it comes to little earthquakes, 00:51:02.160 --> 00:51:04.930 they’re occurring on all kinds of crazy faults. 00:51:04.930 --> 00:51:12.170 So our approach is, instead, to use focal mechanisms as proxies for the faults. 00:51:12.170 --> 00:51:15.769 Because focal mechanisms capture that true variability. 00:51:15.769 --> 00:51:19.009 So here are all the focal mechanisms that we have 00:51:19.009 --> 00:51:23.390 from GeoNet for the Kaikoura magnitude 7.6 earthquake. 00:51:23.390 --> 00:51:27.089 The mechanisms are red where we calculated stress to have increased 00:51:27.089 --> 00:51:30.509 and blue where it has decreased. And now you see a lot of red. 00:51:30.509 --> 00:51:33.450 Not exclusively red, but a lot of red close to the fault, 00:51:33.450 --> 00:51:36.369 in addition to some places very far apart. 00:51:36.369 --> 00:51:40.190 If we compare this to the seismicity rate change, the rate of earthquakes 00:51:40.190 --> 00:51:43.869 after Kaikoura to the preceding four years, 00:51:43.869 --> 00:51:47.460 you can see that these are very well-correlated. 00:51:47.460 --> 00:51:51.980 For example, the Darfield, Canterbury, area turns blue. 00:51:51.980 --> 00:51:53.640 And it turns blue here. 00:51:53.640 --> 00:51:55.900 This is red and this is red. 00:51:55.900 --> 00:52:00.779 We see it red around Wellington up here. 00:52:00.779 --> 00:52:06.720 So I think how you create these tests is important. 00:52:06.720 --> 00:52:12.869 And so the nominal failure of Coulomb in this test is really just a springboard 00:52:12.869 --> 00:52:17.640 for trying to design better tests as opposed to the end of the road. 00:52:20.680 --> 00:52:23.380 - Any final comments or questions? 00:52:23.380 --> 00:52:25.660 - Okay. I have a final comment. [laughter] 00:52:26.520 --> 00:52:31.579 You know, since I left the Survey, I’ve been building Temblor. 00:52:31.579 --> 00:52:35.049 And I spend a lot of time in – among engineers and a lot of time 00:52:35.049 --> 00:52:39.190 in the insurance world. And fortunately, through this 00:52:39.190 --> 00:52:45.140 international lecture, I was – six or seven countries this year. 00:52:45.140 --> 00:52:51.519 The one constant I see wherever I go is the high regard the USGS is held in. 00:52:51.519 --> 00:52:56.009 I never hear anything but appreciation and respect. 00:52:56.009 --> 00:52:59.760 And I – because I’m not so much associated with the Survey anymore, 00:52:59.760 --> 00:53:02.589 I would hear that if that weren’t the case. 00:53:02.589 --> 00:53:05.940 It’s something that probably, inside this room, you don’t realize 00:53:05.940 --> 00:53:11.700 as much as I can from my position outside how well regarded it is. 00:53:11.700 --> 00:53:15.920 Years ago, Mark Zoback said, you know, everybody thinks 00:53:15.930 --> 00:53:20.869 the Survey is a 500-pound gorilla. He says, I’m not – I’m here. 00:53:20.869 --> 00:53:25.549 I really know it’s 500 one-pound monkeys. [laughter] 00:53:25.549 --> 00:53:30.349 But actually, the idea that it’s a 500-pound gorilla isn’t correct. 00:53:30.349 --> 00:53:35.269 That everywhere you turn, people appreciate and use 00:53:35.269 --> 00:53:39.829 the research and the tools and the guidance that the Survey provides. 00:53:39.829 --> 00:53:43.219 So, you know, in addition to a shout-out to Jack Boatwright, 00:53:43.219 --> 00:53:47.969 I just want to say it’s important that you know how highly regarded 00:53:47.969 --> 00:53:50.759 all of you are and the work of the Survey. 00:53:50.760 --> 00:53:53.500 And that’s what I’d like to leave you with. Thank you. 00:53:53.500 --> 00:54:00.140 [Applause] 00:54:00.140 --> 00:54:01.500 - So if you’d like to join us for lunch, 00:54:01.500 --> 00:54:04.180 probably just hang out here for a few minutes, and we’ll coordinate that. 00:54:04.860 --> 00:54:12.300 [Silence]