WEBVTT Kind: captions Language: en-US 00:00:01.882 --> 00:00:07.120 [silence] 00:00:07.120 --> 00:00:10.240 All right. Welcome, everybody. We’re happy to see everyone here 00:00:10.240 --> 00:00:13.680 for the seminar this week. Before we get started, just wanted 00:00:13.680 --> 00:00:18.320 to remind you that the USGS public lecture series is starting up again 00:00:18.320 --> 00:00:23.039 on March 25th – that’s tomorrow evening – at 7:00 p.m. 00:00:23.039 --> 00:00:28.696 A Jaguar’s Field of Dreams is the title of that seminar. 00:00:28.720 --> 00:00:32.575 The link is also provided in the chat. 00:00:34.800 --> 00:00:38.200 All right. Let’s see if I can stop sharing the screen. [chuckles] 00:00:41.040 --> 00:00:42.474 Okay. 00:00:44.240 --> 00:00:47.146 So [audio cuts out] … 00:00:49.615 --> 00:01:05.380 [silence] 00:01:05.380 --> 00:01:11.760 - Looks like we may have lost Tamara. So, Susan, did we have any other 00:01:11.760 --> 00:01:15.419 announcements that Tamara was going to go over? 00:01:15.443 --> 00:01:20.816 - She was going to announce the talk on Monday the 30th. 00:01:23.583 --> 00:01:26.456 - [laughs] 00:01:26.480 --> 00:01:29.840 In any case, there are a variety of events coming up that you’ll have 00:01:29.840 --> 00:01:33.760 received emails for, including, as she pointed out, the public lecture 00:01:33.760 --> 00:01:37.360 tomorrow and a series of other talks from the Innovation Center. 00:01:37.360 --> 00:01:44.080 I’ll hand it over now – or, next week, our seminar is Ian Stone, who was 00:01:44.080 --> 00:01:47.840 delayed from last week when we watched the all-hands meeting 00:01:47.840 --> 00:01:50.661 from the Department of the Interior. 00:01:51.760 --> 00:01:55.200 So please stick around and join us for that in a week. 00:01:55.200 --> 00:01:58.960 But, for now, I’ll hand it over to Jeanne Hardebeck, who will introduce 00:01:58.960 --> 00:02:03.680 today’s speaker, Ruth Harris. - Hi, thanks. It’s my pleasure 00:02:03.680 --> 00:02:08.696 to introduce Ruth, who many of you probably know. 00:02:08.720 --> 00:02:13.016 A brief summary of her career. She was an undergrad at MIT. 00:02:13.040 --> 00:02:16.960 She got a master’s degree at Cornell and earned her Ph.D. 00:02:16.960 --> 00:02:22.320 at UC-Santa Barbara. She’s been at the USGS ever since 00:02:22.320 --> 00:02:26.640 then, and for probably about 10 or 15 years now, has been the leader of the 00:02:26.640 --> 00:02:30.936 Earthquake Process, Probability, and Occurrence Megaproject. 00:02:30.960 --> 00:02:34.400 Ruth has won a number of awards, including the Department of Interior 00:02:34.400 --> 00:02:39.680 Meritorious Service Award, and in 2019, she was elected 00:02:39.680 --> 00:02:42.456 a fellow at AGU. 00:02:42.480 --> 00:02:47.200 Some of the highlights of her research – she was an early pioneer in the field 00:02:47.200 --> 00:02:53.600 of earthquake stress triggering, particularly in showing how large 00:02:53.600 --> 00:02:58.400 earthquakes, like the 1906 and other large earthquakes like that can cause 00:02:58.400 --> 00:03:02.960 stress shadows of areas where stress has been decreased and the number of 00:03:02.960 --> 00:03:06.480 earthquakes decreases because those earthquakes have been delayed. 00:03:06.480 --> 00:03:10.880 She also wrote a very influential early review paper on earthquake 00:03:10.880 --> 00:03:15.760 stress interactions that’s still frequently cited in that field. 00:03:15.760 --> 00:03:18.960 Much of her work has also been in dynamic rupture modeling, including 00:03:18.960 --> 00:03:22.720 some pioneer studies on fault geometries, and particularly what fault 00:03:22.720 --> 00:03:27.656 geometries will allow fault-to-fault jumping during earthquakes. 00:03:27.680 --> 00:03:31.440 She’s been the leader for a number of years now of the SCEC Dynamic 00:03:31.440 --> 00:03:37.360 Earthquake Rupture Code Validation Project, which is a large project that, 00:03:37.360 --> 00:03:40.560 even though it’s under the – under the auspices of SCEC, 00:03:40.560 --> 00:03:43.040 includes a lot of even international participants, 00:03:43.040 --> 00:03:45.040 where everybody validates their code. 00:03:45.040 --> 00:03:50.080 And all the hard work that Ruth has put into this really ensures the accuracy 00:03:50.080 --> 00:03:54.000 of all the results that come out of dynamic rupture simulations. 00:03:54.000 --> 00:03:56.880 Today she’s going to be talking about some dynamic rupture scenarios 00:03:56.880 --> 00:04:00.936 for the Hayward, Rodgers Creek, and Calaveras Fault systems. 00:04:00.960 --> 00:04:03.208 Take it away, Ruth. 00:04:04.150 --> 00:04:07.896 - Thank you very much, Jeanne, for the kind introduction. 00:04:07.920 --> 00:04:13.840 And now I seamlessly switch over to my slides, and we’ll see how this goes. 00:04:16.224 --> 00:04:20.318 [silence] 00:04:20.357 --> 00:04:25.154 And can you see my slides? - Looks great. 00:04:25.154 --> 00:04:31.896 - Okay. And then I will try to get my pointer to work – the laser pointer. 00:04:31.920 --> 00:04:35.661 And that – yes. All right. Whoo. First time. 00:04:36.640 --> 00:04:40.160 So, welcome, everybody. And I want to give a big thank you 00:04:40.160 --> 00:04:44.160 to Tamara Jeppson, who hopefully won’t be big-blued forever – 00:04:44.160 --> 00:04:47.680 will be able to come back in – Austin Elliott, and Susan Garcia, 00:04:47.680 --> 00:04:51.120 who are doing all the hard work of the seminar, organizing, 00:04:51.120 --> 00:04:53.976 coordinating, and then helping all of us with Teams. 00:04:54.000 --> 00:04:57.736 And thank you, again, to Jeanne for the kind introduction. 00:04:57.760 --> 00:05:01.040 So today I’m going to talk about a project that I started in an almost 00:05:01.040 --> 00:05:05.176 embarrassingly long time ago – I think almost 20 years ago now. 00:05:05.200 --> 00:05:09.680 And, at that time, we didn’t know as much about these faults as we do now. 00:05:09.680 --> 00:05:13.440 So I started on this project, and I think it was actually originally 00:05:13.440 --> 00:05:17.816 suggested by Art McGarr. So I started on this project, 00:05:17.840 --> 00:05:20.800 and I finished the project and wrote a paper, 00:05:20.800 --> 00:05:22.880 and it went through internal review and did fine. 00:05:22.880 --> 00:05:26.320 But then I hesitated before sending it to the journal because I really wanted 00:05:26.320 --> 00:05:30.000 the model to be bigger and better. So, in the time of our doing the 00:05:30.000 --> 00:05:35.120 first half of it, the Hayward and – the Hayward and Calaveras Faults 00:05:35.120 --> 00:05:38.640 connected up to each other. And then we also learned a lot 00:05:38.640 --> 00:05:41.656 more about the Hayward and the Rodgers Creek Faults. 00:05:41.680 --> 00:05:47.096 So that made me realize that the model I was doing wasn’t quite sufficient. 00:05:47.120 --> 00:05:50.480 So then I put the work aside for a few years and then started it up again. 00:05:50.480 --> 00:05:52.480 And then here is the culmination of it. 00:05:52.480 --> 00:05:56.240 So, of course, those two faults didn’t, in life – real life connect to each other, 00:05:56.240 --> 00:05:59.068 but we learned so much more about them. 00:06:00.560 --> 00:06:05.120 So I wanted to give many thanks to my JGR paper co-authors. 00:06:05.120 --> 00:06:07.600 And this new paper just came out last week. 00:06:07.600 --> 00:06:11.496 It was finally printed in its fully, glossy form. 00:06:11.520 --> 00:06:15.520 Co-authors Michael Barall, Dave Lockner, Diane Moore, 00:06:15.520 --> 00:06:20.240 Dave Ponce, Russ Grayer, and Carolyn Morrow all work at the USGS. 00:06:20.240 --> 00:06:23.840 Gareth Funning from UC-Riverside. Christos Kyriakopoulos, who is 00:06:23.840 --> 00:06:27.120 at University of Memphis. And then Donna Eberhart-Phillips, who 00:06:27.120 --> 00:06:32.776 is at GNS Science New Zealand and also has an affiliation with UC-Davis. 00:06:32.800 --> 00:06:36.000 Next – internal review. All right. A lot of people mutter and 00:06:36.000 --> 00:06:39.440 mumble about USGS internal reviews, and that it slows down our work, 00:06:39.440 --> 00:06:44.320 but it actually makes it so much better. So I want to acknowledge internal 00:06:44.320 --> 00:06:48.480 reviewers for this manuscript, including really – got really helpful 00:06:48.480 --> 00:06:51.176 comments from Evan Hirakawa and Jess Murray. 00:06:51.200 --> 00:06:55.336 And then the previous iteration of this manuscript benefited a lot 00:06:55.360 --> 00:06:58.240 from comments by Jim Lienkaemper and John Langbein. 00:06:58.240 --> 00:07:03.600 So internal review works, and it helps us make our science even better and 00:07:03.600 --> 00:07:09.256 helps us find subtle flaws that we then improve and make things look okay. 00:07:09.280 --> 00:07:11.600 We received funding – or, I received funding from 00:07:11.600 --> 00:07:15.520 the PG&E/USGS CRADA for this work, and then also for some related 00:07:15.520 --> 00:07:19.365 work that I’m going to talk about – mention during this talk too. 00:07:20.560 --> 00:07:23.680 And one of the great things about working at the USGS – one of the 00:07:23.680 --> 00:07:26.640 many great things about working at the USGS is all the colleagues 00:07:26.640 --> 00:07:31.520 that we have easy access to and can easily talk to about our work. 00:07:31.520 --> 00:07:34.616 And then we also have, of course, our colleagues in academia. 00:07:34.640 --> 00:07:38.640 So I want to just – this is a list of some of the many people who, 00:07:38.640 --> 00:07:42.800 over the years, got to have me ask them questions and give – 00:07:42.800 --> 00:07:45.336 and then they gave me thoughtful answers. 00:07:45.360 --> 00:07:48.640 I want to most particularly point out Bob Simpson, 00:07:48.640 --> 00:07:51.920 whose name is first on the list. He’s a – has been a long-time mentor 00:07:51.920 --> 00:07:56.720 to so many of us at the USGS – a kind, wonderful, very smart, 00:07:56.720 --> 00:08:00.080 and modest person who contributed a whole bunch to this work, 00:08:00.080 --> 00:08:05.458 especially in the first – in the first year or so I was working on it. 00:08:06.560 --> 00:08:09.816 All right. So here we are. San Francisco Bay area. 00:08:09.840 --> 00:08:12.160 Pretty nice place. We’ve got the ocean. 00:08:12.160 --> 00:08:16.216 We’ve got some mountains. We have the bay. 00:08:16.240 --> 00:08:18.400 So that makes it kind of good. Sometimes we go on fire, 00:08:18.400 --> 00:08:22.400 so that’s not so fun. But we also have a pretty high 00:08:22.400 --> 00:08:27.120 probability of large earthquakes. This figure is from the Fact Sheet – 00:08:27.120 --> 00:08:31.096 Aagaard et al. USGS Fact Sheet published in 2016. 00:08:31.120 --> 00:08:35.680 And the results are based on work by Ned Field and colleagues that 00:08:35.680 --> 00:08:38.776 were published in 2013 and 2014. 00:08:38.800 --> 00:08:41.600 And that’s the California earthquake probability. 00:08:41.600 --> 00:08:45.200 So this is just for the Bay Area – focusing on the Bay Area. 00:08:45.200 --> 00:08:50.080 And here we have a 72% probability of one or more magnitude greater than 00:08:50.080 --> 00:08:53.360 or equal to 6.7. And that number 6.7 is due to 00:08:53.360 --> 00:08:56.400 a damaging earthquake in southern California – the Northridge earthquake 00:08:56.400 --> 00:09:03.016 that occurred in ’94. So we know 6.7s and bigger are bad – bad for us. 00:09:03.040 --> 00:09:10.056 And the 72% probability is for the 30-year period 2014 to 2043. 00:09:10.080 --> 00:09:12.720 Now, we think a lot about the San Andreas Fault. 00:09:12.720 --> 00:09:14.480 That’s really famous. Everyone knows about the 00:09:14.480 --> 00:09:16.800 San Andreas Fault who knows about faults. 00:09:16.800 --> 00:09:19.976 So we think about it as having a pretty high probability. 00:09:20.000 --> 00:09:22.960 But we also have these other faults that even have higher probabilities 00:09:22.960 --> 00:09:25.280 than the San Andreas. And these include the combined 00:09:25.280 --> 00:09:30.480 Hayward and Rodgers Creek Faults with a 33% chance of one or more 00:09:30.480 --> 00:09:33.496 magnitude greater than or equal to 6.7 earthquakes. 00:09:33.520 --> 00:09:38.640 And then down here, we have the Calaveras Fault with a 26% probability – 00:09:38.640 --> 00:09:42.536 Central Calaveras – that then moves over to the Northern Calaveras. 00:09:42.560 --> 00:09:44.960 So we also have these other faults that have high chances of 00:09:44.960 --> 00:09:48.160 having large earthquakes. We know that they’ve had them 00:09:48.160 --> 00:09:51.656 in either historic or geologic times. 00:09:51.680 --> 00:09:56.856 Rodgers Creek Fault had large earthquakes that is evidenced from the 00:09:56.880 --> 00:10:01.920 geological studies by Suzanne Hecker, Dave Schwartz, and others have 00:10:01.920 --> 00:10:04.856 worked a lot on the Rodgers Creek Faults. 00:10:04.880 --> 00:10:08.000 The Hayward Fault – we know the best – the reason we have 00:10:08.000 --> 00:10:11.600 the HayWired earthquake scenario is because we know that there was 00:10:11.600 --> 00:10:16.960 a large earthquake in 1868 – about a magnitude 6.8 or so – on this fault. 00:10:16.960 --> 00:10:20.000 We didn’t, of course, have a lot of seismic recordings of that, 00:10:20.000 --> 00:10:23.976 but we do have some other information about its effects. 00:10:24.000 --> 00:10:28.456 And the HayWired scenario has been worked on by – Ken Hudnut led that. 00:10:28.480 --> 00:10:30.880 And then Anne Wein is also a collaborator on that – 00:10:30.880 --> 00:10:33.760 one of the key leads on that. And they’ve looked at everything 00:10:33.760 --> 00:10:38.480 from ground shaking to landslides, liquefaction, all the way to societal 00:10:38.480 --> 00:10:43.341 effects, and that’s work that Anne Wein is particularly focused on. 00:10:44.320 --> 00:10:49.600 Calaveras Fault – Calaveras Fault has also had large-ish earthquakes – 00:10:49.600 --> 00:10:53.600 not quite as big, but the Northern Calaveras Fault, there’s geologic 00:10:53.600 --> 00:10:58.480 evidence of maybe one or two events that Keith Kelson discovered. 00:10:58.480 --> 00:11:01.040 And I think those are smaller than magnitude 6, 00:11:01.040 --> 00:11:03.096 or maybe around magnitude 6. 00:11:03.120 --> 00:11:07.680 And then, in our time, or some of our times, when we did have good 00:11:07.680 --> 00:11:11.680 recordings of earthquakes, central Calaveras has produced events. 00:11:11.680 --> 00:11:15.336 There was the Alum Rock event – I think that was in 2010. 00:11:15.360 --> 00:11:18.320 But then, 1984 is the biggest one that we’ve had – 00:11:18.320 --> 00:11:23.976 low magnitude 6, and that was well-recorded. 00:11:24.000 --> 00:11:27.280 So these faults produce large earthquakes, but in addition to this, 00:11:27.280 --> 00:11:31.440 they also creep. So we have to try to figure out, 00:11:31.440 --> 00:11:35.576 what is the fact that they’re creeping all the time, or slowly slipping all the time, 00:11:35.600 --> 00:11:37.440 what does that do to the large earthquakes? 00:11:37.440 --> 00:11:41.760 Does it make them different or the same as our garden variety 00:11:41.760 --> 00:11:43.680 larger earthquake on a locked fault, 00:11:43.680 --> 00:11:46.776 such as the San Andreas Fault in this part of California? 00:11:46.800 --> 00:11:52.000 So that’s – and that is part of the study, is to try to understand how the fact that 00:11:52.000 --> 00:11:57.920 they are aseismic and slipping, what does that do to the ground motions 00:11:57.920 --> 00:12:00.800 and extent of rupture that will be produced when we do 00:12:00.800 --> 00:12:04.568 eventually have large earthquakes on these faults again. 00:12:06.080 --> 00:12:10.240 Okay, and this is just showing two photos of the effects of 00:12:10.240 --> 00:12:14.080 fault creep on the Hayward Fault. And these are – these two photos 00:12:14.080 --> 00:12:18.760 are one of many in an Open-File report that was written by Phil Stoffer – 00:12:18.760 --> 00:12:24.536 the late Phil Stoffer and published in a Open-File report in 2008. 00:12:24.560 --> 00:12:27.520 The top is just showing what happens when you have fault creep – 00:12:27.520 --> 00:12:30.320 creeping fault intersecting a school. That doesn’t go so well. 00:12:30.320 --> 00:12:32.456 They had to close the school. 00:12:32.480 --> 00:12:36.240 And then the bottom is kind of more just a fun thing to look at, but it’s 00:12:36.240 --> 00:12:39.760 what happens when you have creeping fault go through a city park. 00:12:39.760 --> 00:12:42.800 Your stone wall, which used to be pretty connected, 00:12:42.800 --> 00:12:45.520 becomes disconnected. So sometimes you can have, 00:12:45.520 --> 00:12:49.760 like, interesting and curiosity kind of features like this, but other times, 00:12:49.760 --> 00:12:53.440 when you’re intersecting your infrastructure, then it doesn’t – 00:12:53.440 --> 00:12:58.080 it doesn’t work so well with fault creep. So fault creep, overall, isn’t quite 00:12:58.080 --> 00:13:01.760 as dangerous – isn’t as dangerous as large earthquakes. 00:13:01.760 --> 00:13:06.456 But it still does cause some problems. 00:13:06.480 --> 00:13:12.160 So creeping faults are rare, but, in the grand scheme of things, they’re 00:13:12.160 --> 00:13:17.520 pretty abundant in northern California. And Diane Moore, Bob McLaughlin – 00:13:17.520 --> 00:13:20.400 I think there are one or two other co-authors – have a paper in, I think, 00:13:20.400 --> 00:13:26.160 Tectonics in 2018, where they go and explain why this is the case 00:13:26.160 --> 00:13:29.656 for this part of the state – why we have so many creeping faults. 00:13:29.680 --> 00:13:33.200 And these creeping faults include the Maacama Fault, Bartlett Springs Fault, 00:13:33.200 --> 00:13:35.896 Rodgers Creek, which isn’t quite labeled here, 00:13:35.920 --> 00:13:38.616 Greenville, Hayward, Calaveras. 00:13:38.640 --> 00:13:41.520 And then, jumping down to central California, of course, we have the 00:13:41.520 --> 00:13:45.920 famous San Andreas Fault creep, and that’s maybe even the one place 00:13:45.920 --> 00:13:50.080 in the world where fault creep – on an active continental tectonic fault 00:13:50.080 --> 00:13:55.176 where fault creep is almost, if not the whole, slip budget. 00:13:55.200 --> 00:13:58.320 Then we have to head all the way down to southern California – 00:13:58.320 --> 00:14:01.280 the southern part of southern California before we encounter 00:14:01.280 --> 00:14:04.640 creeping faults again. And these are the Superstition Hills and Imperial Faults. 00:14:04.640 --> 00:14:08.000 So this is a review article I wrote about large earthquakes, 00:14:08.000 --> 00:14:15.200 creeping faults, published in 2017. Creeping faults also occur 00:14:15.200 --> 00:14:18.056 in some places in the rest of the world. 00:14:18.080 --> 00:14:22.000 They’re still quite rare. And this figure shows some of the 00:14:22.000 --> 00:14:26.376 other locations where creeping faults have also been – also been noted. 00:14:26.400 --> 00:14:30.640 Some faults do respond after an earthquake, and they will – 00:14:30.640 --> 00:14:33.760 there will be, like, noticeable postseismic slip. 00:14:33.760 --> 00:14:37.440 But, for here, the creeping faults that are being mentioned are 00:14:37.440 --> 00:14:39.600 those where they’re slowly slipping all the time. 00:14:39.600 --> 00:14:44.497 So it’s not just postseismic, but it’s always happening. 00:14:46.080 --> 00:14:50.720 All right. So the question is, how are creeping-fault earthquakes 00:14:50.720 --> 00:14:55.040 different or the same – different from or the same as locked-fault 00:14:55.040 --> 00:14:57.576 earthquakes – your regular earthquakes? 00:14:57.600 --> 00:15:01.496 So, for the – in the Reviews of Geophysics paper, did a comparison 00:15:01.520 --> 00:15:05.600 looking at two things – looking at the rupture areas – fault surface 00:15:05.600 --> 00:15:08.080 rupture areas and then also looking at the ground shaking effects. 00:15:08.080 --> 00:15:12.160 So here’s looking at the fault surface at the rupture areas of earthquakes 00:15:12.160 --> 00:15:17.040 that have occurred on creeping faults. So what I’ve done here is compared 00:15:17.040 --> 00:15:23.200 the predicted rupture area based on empirical relation – an empirical 00:15:23.200 --> 00:15:28.080 relation published in Tom Hanks’ and Bill Bakun’s BSSA 2008 paper. 00:15:28.080 --> 00:15:31.360 So there – it’s a magnitude – a paper about magnitude-log area 00:15:31.360 --> 00:15:34.720 for crustal earthquakes. So look at what the predicted rupture 00:15:34.720 --> 00:15:39.176 area should be, and that’s based on primarily locked-fault earthquakes. 00:15:39.200 --> 00:15:42.376 And then what the actual rupture area was. 00:15:42.400 --> 00:15:47.360 Anywhere where it’s a red circle is a creeping-fault earthquake. 00:15:47.360 --> 00:15:54.240 Any part of the plot where you see a blue – a blue square is a locked-fault 00:15:54.240 --> 00:15:57.840 earthquake, so just mostly look to see where the red circles are. 00:15:57.840 --> 00:16:02.216 And you’ll notice the red circles falling within the scatter of the blue squares. 00:16:02.240 --> 00:16:05.520 So, at least in terms of rupture areas, it looks like the creeping-fault 00:16:05.520 --> 00:16:08.640 earthquakes and the locked-fault earthquakes are behaving the same. 00:16:08.640 --> 00:16:12.480 But the caveat is that the biggest creeping-fault earthquake here that we 00:16:12.480 --> 00:16:17.040 looked at, that we had good data for – actually for the ground motion – 00:16:17.040 --> 00:16:19.840 was a magnitude less than 6.7 earthquake. 00:16:19.840 --> 00:16:22.880 So we don’t have information for magnitude greater than 6.7 00:16:22.880 --> 00:16:25.656 creeping-fault earthquakes. 00:16:25.680 --> 00:16:28.960 Okay. We did – also looked at the ground shaking and first published 00:16:28.960 --> 00:16:31.840 this is a paper with Norm Abrahamson 00:16:31.840 --> 00:16:36.376 and published this one in 2014 – a GRL paper. 00:16:36.400 --> 00:16:40.800 And then I noticed there was a glitch on one point on the plot, so anyway, 00:16:40.800 --> 00:16:44.720 showed it again in the 2017 Reviews of Geophysics paper. 00:16:44.720 --> 00:16:48.960 So this is just a quick synopsis of looking at peak ground acceleration 00:16:48.960 --> 00:16:53.200 versus earthquake magnitude for creeping-fault earthquakes 00:16:53.200 --> 00:16:56.640 and locked-fault earthquakes. The main takeaway from this figure – 00:16:56.640 --> 00:16:59.176 creeping-fault earthquakes are the red diamonds. 00:16:59.200 --> 00:17:01.656 Locked-fault earthquakes are the blue and greens. 00:17:01.680 --> 00:17:04.720 So the main thing to notice is that the red diamonds are falling within 00:17:04.720 --> 00:17:08.640 the scatter of the blue diamonds for peak ground acceleration, 00:17:08.640 --> 00:17:13.896 falling within the scatter of the green diamonds for peak ground velocity. 00:17:13.920 --> 00:17:17.416 In other words, you can’t tell the difference in the peak ground shaking 00:17:17.440 --> 00:17:20.480 between the creeping-fault earthquakes and the locked-fault earthquakes. 00:17:20.480 --> 00:17:24.240 One again, this is just for magnitude less than 6.7 earthquakes. 00:17:24.240 --> 00:17:29.818 Because that’s all the information that we had for creeping faults. 00:17:31.440 --> 00:17:35.760 So, in the first figure that I showed you that had earthquake probabilities 00:17:35.760 --> 00:17:40.720 for magnitude greater than or equal or 6.7 in the Bay Area – so the stuff 00:17:40.720 --> 00:17:43.680 I just talked about was for magnitude less than 6.7. 00:17:43.680 --> 00:17:46.640 But what about if you want to know about how creeping-fault 00:17:46.640 --> 00:17:50.216 earthquakes behave if they’re magnitude 6.7 or larger? 00:17:50.240 --> 00:17:54.056 So we don’t – we do not have empirical evidence for this. 00:17:54.080 --> 00:17:55.520 We just know that they have occurred, 00:17:55.520 --> 00:17:59.521 but we don’t know what they look like or anything. 00:18:01.280 --> 00:18:04.616 So this takes us over to computational modeling. 00:18:04.640 --> 00:18:09.360 So two tools that we could use to solve this problem of how creeping fault 00:18:09.360 --> 00:18:13.040 earthquakes behave and what they look like are these kinematic 00:18:13.040 --> 00:18:17.200 earthquake rupture simulations or dynamic, or spontaneous, 00:18:17.200 --> 00:18:20.296 earthquake rupture simulations. I want to use the words “spontaneous” 00:18:20.320 --> 00:18:22.480 meaning dynamic – dynamic, spontaneous. 00:18:22.480 --> 00:18:26.080 So the main thing about them is that the earthquake is not pre-decided. 00:18:26.080 --> 00:18:31.708 Instead, it happens through a physical interaction of all the processes. 00:18:32.960 --> 00:18:37.920 So here is a schematic – so other people’s talks, including, I think, 00:18:37.920 --> 00:18:41.256 the talk next week by Mendenhall postdoc Ian Stone, 00:18:41.280 --> 00:18:45.372 will use the kinematic earthquake rupture simulation method. 00:18:45.372 --> 00:18:48.720 You’ve also maybe heard talks by Artie Rodgers, who’s showed 00:18:48.720 --> 00:18:51.440 simulations of Hayward Fault earthquakes – high-frequency, 00:18:51.440 --> 00:18:54.880 really nice simulations looking at high-frequency ground motions 00:18:54.880 --> 00:18:57.256 from Hayward earthquakes. 00:18:57.280 --> 00:19:01.760 Next week’s talk, and also Artie Rodgers’ talks, these are all using 00:19:01.760 --> 00:19:04.800 this kinematic earthquake rupture simulation tool. 00:19:04.800 --> 00:19:09.096 For these, it’s already pre-decided what the earthquake source is going to be. 00:19:09.120 --> 00:19:13.336 It’s already pre-decided how far along strike the earthquake will be, 00:19:13.360 --> 00:19:19.120 how far along dip it’ll reach, how much slip there will be, how fast it will slip – 00:19:19.120 --> 00:19:21.416 everything has already been predetermined. 00:19:21.440 --> 00:19:23.920 And then they’re trying to see what the ground shaking will look like 00:19:23.920 --> 00:19:26.960 based on that predetermined earthquake source. 00:19:26.960 --> 00:19:31.440 So that’s other people’s talks. My talk is about spontaneous 00:19:31.440 --> 00:19:34.056 or dynamic earthquake rupture simulation. 00:19:34.080 --> 00:19:37.120 We are trying to figure out what the earthquakes are going to look like, 00:19:37.120 --> 00:19:42.616 so we have that as an output of all the assumptions that we make. 00:19:42.640 --> 00:19:48.582 So we have to assume the fault geometry, as these guys do – and gals, 00:19:48.616 --> 00:19:51.840 the rock properties, as these people do. 00:19:51.840 --> 00:19:54.880 But they do not have to assume the initial stresses are friction because 00:19:54.880 --> 00:19:58.240 they’ve already – they already have pre-decided the earthquake source. 00:19:58.240 --> 00:20:01.736 So we’re going to put all these pieces together. 00:20:01.760 --> 00:20:06.160 And it’ll have interaction among them, but physically self-consistent results. 00:20:06.160 --> 00:20:08.400 But beforehand, we don’t know what the result’s going to be. 00:20:08.400 --> 00:20:12.616 Instead, it is an outcome of our simulations. 00:20:12.640 --> 00:20:14.480 And also these models predict ground motions, 00:20:14.480 --> 00:20:17.725 and our models also predict ground motions. 00:20:19.280 --> 00:20:23.280 All right. So step number 1. Well, if you have all the stuff, 00:20:23.280 --> 00:20:25.280 but you don’t have a code, you’re not going to go anywhere. 00:20:25.280 --> 00:20:30.287 So how about, first of all, let’s go find a well-tested code. 00:20:31.760 --> 00:20:36.160 This brings us to the SCEC-USGS Dynamic Rupture – Earthquake 00:20:36.160 --> 00:20:40.560 Rupture Code Comparison Group, and this is a congenial group of people. 00:20:40.560 --> 00:20:43.840 These are showing photos of everyone whose photos I could find, 00:20:43.840 --> 00:20:46.376 and I think I’m still missing some people here. 00:20:46.400 --> 00:20:48.480 But these are our collaborators over the years. 00:20:48.480 --> 00:20:51.440 Some people have moved into other subjects. 00:20:51.440 --> 00:20:54.296 A few people went over to induced seismicity. 00:20:54.320 --> 00:20:57.600 Some people went into the oil industry. Some people went into the banking 00:20:57.600 --> 00:21:00.880 industry or re-insurance. But many of these people are 00:21:00.880 --> 00:21:06.000 still dynamic rupture modelers. And people often start as students. 00:21:06.000 --> 00:21:08.640 Sometimes that’s their thesis project, where they have to write a code 00:21:08.640 --> 00:21:11.120 or use someone else’s code. And then some people have made it 00:21:11.120 --> 00:21:14.880 all the way up to be senior researchers, senior faculty, etc. 00:21:14.880 --> 00:21:17.440 But it’s a evolving group. There are always new people 00:21:17.440 --> 00:21:21.120 joining, and it’s very exciting. And the main point of our group 00:21:21.120 --> 00:21:24.640 has been to check and make sure our codes are working okay 00:21:24.640 --> 00:21:28.376 and then also to talk about dynamic rupture science. 00:21:28.400 --> 00:21:32.800 So we’ve been around – I think we started in late 2003, early 2004. 00:21:32.800 --> 00:21:36.800 We’ve now been around for about 17 years, and it’s just a really 00:21:36.800 --> 00:21:39.520 good collaboration. And we’ve had people from – 00:21:39.520 --> 00:21:44.505 by now, at least nine countries or 10 countries participating. 00:21:46.720 --> 00:21:50.240 So these are the exercises that we did. These types of codes do not have 00:21:50.240 --> 00:21:53.576 any analytic solutions with which to test them. 00:21:53.600 --> 00:21:57.040 So really, I could just do anything and chose some pretty movie, 00:21:57.040 --> 00:22:00.560 and you would have no idea if I did it right or just made up something. 00:22:00.560 --> 00:22:06.640 You wouldn’t know unless you could also go and check – use the code – 00:22:06.640 --> 00:22:11.840 your code or my code and go re-run all those – use all the same assumptions 00:22:11.840 --> 00:22:15.600 and then see if we get the same results. So, when we’re doing this, 00:22:15.600 --> 00:22:18.640 we wandered a little bit, maybe for a few months, and then we finally 00:22:18.640 --> 00:22:21.920 settled on one benchmark where we could get it all together. 00:22:21.920 --> 00:22:26.960 And that was a very simple model of a vertical strike-slip fault 00:22:26.960 --> 00:22:29.600 in homogeneous fullspace. So we didn’t even have the 00:22:29.600 --> 00:22:32.776 Earth’s surface. Then we tacked on the Earth’s surface. 00:22:32.800 --> 00:22:35.920 So we went through all these different exercises, and for every one of them, 00:22:35.920 --> 00:22:40.000 we had everyone use the same assumptions – so same fault geometry, 00:22:40.000 --> 00:22:45.040 initial stresses, friction, rock properties. And then we’d see if we’d both get 00:22:45.040 --> 00:22:47.960 the same – if we’d all get the same ground shaking at the Earth’s surface – 00:22:47.960 --> 00:22:50.160 predicted the same ground shaking at the Earth’s surface, and actually 00:22:50.160 --> 00:22:52.136 throughout the whole 3D model. 00:22:52.160 --> 00:22:56.536 And then also if we would get the same rupture progress. 00:22:56.560 --> 00:22:59.440 So we went through all these exercises, slowly making changes. 00:22:59.440 --> 00:23:02.720 Because, as soon as you make a drastic change, guarantee it’s all going to fall 00:23:02.720 --> 00:23:04.960 apart, and no one’s going to match. So we did it gradually. 00:23:04.960 --> 00:23:08.240 We did the bimaterial problem. Some of these things we did because 00:23:08.240 --> 00:23:13.200 they were – they were studies that individual scientists had done in our 00:23:13.200 --> 00:23:16.160 group, and then we just wanted to see if our codes worked okay with them. 00:23:16.160 --> 00:23:19.760 Others, we had kind of more funding motivations or other 00:23:19.760 --> 00:23:22.800 reasons for doing them. So, in the middle part, we are stepping 00:23:22.800 --> 00:23:27.360 slowly towards the bottom two cases, which was doing simulations 00:23:27.360 --> 00:23:30.320 of extreme ground motion. And the reason we were doing these 00:23:30.320 --> 00:23:34.240 is there had been a model that Joe Andrews, who was then a member 00:23:34.240 --> 00:23:37.440 of our group before he retired, he had done some simulations 00:23:37.440 --> 00:23:42.400 for extreme ground shaking at Yucca Mountain, the site of our 00:23:42.400 --> 00:23:46.240 nation’s formerly proposed high-level nuclear waste repository. 00:23:46.240 --> 00:23:49.920 So we wanted to check and make sure that his code and our codes were all 00:23:49.920 --> 00:23:53.840 working the same and producing the same results. And that did work well. 00:23:53.840 --> 00:23:56.800 And we also published a paper on that in 2011 00:23:56.800 --> 00:23:59.976 in Seismological Research Letters. 00:24:00.000 --> 00:24:03.440 Next, we wandered into the theme of friction. 00:24:03.440 --> 00:24:06.776 So we looked at different forms of rate-state friction. 00:24:06.800 --> 00:24:10.640 And then the bottom two we did also included thermal pressurization – 00:24:10.640 --> 00:24:14.080 modeling thermal pressurization. And this last benchmark over here 00:24:14.080 --> 00:24:19.520 is one that we just finished in October and was presented in October at 00:24:19.520 --> 00:24:22.560 our most recent SCEC workshop. 00:24:23.360 --> 00:24:27.920 Next, we also looked a bunch at fault geometry. 00:24:27.920 --> 00:24:31.120 So the upper left example here was fault branches. 00:24:31.120 --> 00:24:36.616 And fault branches are related to our specific Rodgers Creek-Hayward- 00:24:36.640 --> 00:24:39.040 Calaveras-Northern Calaveras simulations because there, too, 00:24:39.040 --> 00:24:41.280 we have a branch that’s a lot fancier than this branch. 00:24:41.280 --> 00:24:45.040 But we did branches, and the reason that we did branches for the code 00:24:45.040 --> 00:24:50.400 comparison exercise had to do with Jeanne Hardebeck discovering a branch 00:24:50.400 --> 00:24:54.240 fault very close to Diablo Canyon Nuclear Power Plant in central 00:24:54.240 --> 00:24:59.440 California. So suddenly, the owner of that power plant was very interested 00:24:59.440 --> 00:25:03.760 in knowing whether or not a rupture could go from the main fault there 00:25:03.760 --> 00:25:08.136 and go onto a fault branch, and thereby, get closer to the power plant. 00:25:08.160 --> 00:25:11.280 That also was very helpful and led us to some extra funding, so that 00:25:11.280 --> 00:25:15.256 was pretty nice. We looked at heterogeneous initial stresses. 00:25:15.280 --> 00:25:19.520 These are often used in simulations and have been since the 1980s, 00:25:19.520 --> 00:25:24.136 sometimes to represent other things, such as fault surface roughness. 00:25:24.160 --> 00:25:26.000 We looked at the fault step-over problem. 00:25:26.000 --> 00:25:29.200 That’s one of my favorites just because it’s the first one I ever did 00:25:29.200 --> 00:25:33.336 for my own research – the first-ever multi-fault rupture study. 00:25:33.360 --> 00:25:37.176 We looked at elastic and viscoplastic yield and what happens when the 00:25:37.200 --> 00:25:41.840 off-fault material, or the off-fault rocks, don’t behave elastically. 00:25:41.840 --> 00:25:46.720 This one was motivated by a study that two of our group members 00:25:46.720 --> 00:25:52.000 had done for southern California, where they’re simulating large 00:25:52.000 --> 00:25:56.080 San Andreas Fault ruptures and their effects on the L.A. Basin. 00:25:56.080 --> 00:25:59.120 But the simulated ground shaking that they were getting was way too high. 00:25:59.120 --> 00:26:01.920 It was much higher than seemed reasonable. 00:26:01.920 --> 00:26:05.440 So they wanted to know if their code was implementing – 00:26:05.440 --> 00:26:08.560 could implement viscoplasticity to try to tamp down those 00:26:08.560 --> 00:26:11.416 ground motions, and that did, indeed, work. 00:26:11.440 --> 00:26:14.936 Then we ventured over to rough faults. Rough faults are really popular. 00:26:14.960 --> 00:26:17.840 People have thought about fault roughness for years, but it’s only 00:26:17.840 --> 00:26:21.336 in more recent times that they’ve been able to put it into their codes. 00:26:21.360 --> 00:26:25.040 So we did that. And then we looked at a range of velocity structures – 00:26:25.040 --> 00:26:29.040 1D vertical, 1D horizontal, and then 3D – sort of like the 00:26:29.040 --> 00:26:33.736 SCEC Community Velocity Model – the Harvard version of it. 00:26:33.760 --> 00:26:35.840 Then we looked at one specific earthquake. 00:26:35.840 --> 00:26:42.240 We modeled the 2004 magnitude 6 Parkfield earthquake and did synthetic 00:26:42.240 --> 00:26:45.920 ground motions at each of the seismic – at the stations that actually recorded 00:26:45.920 --> 00:26:49.280 ground shaking during the earthquake. Although didn’t do that great at 00:26:49.280 --> 00:26:51.680 matching the data, but we did a really good job of matching 00:26:51.680 --> 00:26:54.936 each other, so partly encouraging. 00:26:54.960 --> 00:26:57.840 All right. So, so far, we’ve tested the codes for a variety of 00:26:57.840 --> 00:27:04.296 generic ingredients. And these include mostly generic fault geometries, 00:27:04.320 --> 00:27:07.680 generic rock properties – and, when I say rock properties, 00:27:07.680 --> 00:27:11.600 I mean Vp, Vs, density. And then also elastic versus 00:27:11.600 --> 00:27:15.096 viscoplastic or other forms of inelastic yielding. 00:27:15.120 --> 00:27:20.080 A variety of initial stress conditions and a variety of friction formulations. 00:27:20.080 --> 00:27:24.376 We wrote – our most recent – we had a group paper in 2009. 00:27:24.400 --> 00:27:28.000 And then our most recent one that summarizes the whole group effort 00:27:28.000 --> 00:27:31.680 was published in Seismological Research Letters in 2018. 00:27:31.680 --> 00:27:35.040 And we have a group website that Michael Barall has put together. 00:27:35.040 --> 00:27:38.296 And he’s also played a really important role in this whole group. 00:27:38.320 --> 00:27:40.800 And it’s now called – the SCEC website has now 00:27:40.800 --> 00:27:46.400 been renamed to – by SCEC – strike.scec.org/cvws. 00:27:46.400 --> 00:27:48.320 And, if you have any questions about any of this stuff, 00:27:48.320 --> 00:27:51.280 definitely feel encouraged to send me an email, and I’ll be happy 00:27:51.280 --> 00:27:55.747 to communicate, or we could talk some other way too. 00:27:57.120 --> 00:28:02.376 All right. So we started this because we were trying to find a code to use. 00:28:02.400 --> 00:28:07.840 That was the goal a bunch of slides ago. So here is a list – this is Table 1 of 00:28:07.840 --> 00:28:11.840 our SRL 2018 paper, and this shows a bunch of the different codes that 00:28:11.840 --> 00:28:16.160 have participated in our code comparison exercises. 00:28:16.160 --> 00:28:19.896 So we have – about 10 codes are still active. 00:28:19.920 --> 00:28:23.440 Sometimes people would have a code – they came into the group – maybe it 00:28:23.440 --> 00:28:26.560 was their code they used as a student, and maybe it was their adviser’s 00:28:26.560 --> 00:28:29.440 code or someone else’s code. And then they might say, 00:28:29.440 --> 00:28:32.320 oh, this is too hard to use. I want to go use someone else’s. 00:28:32.320 --> 00:28:34.960 Maybe they had their own code, and then they made it a little bit better, 00:28:34.960 --> 00:28:38.400 and then renamed it. But most of them, I guess, have been 00:28:38.400 --> 00:28:44.080 staying maybe with the same name and sometimes with the same main authors. 00:28:44.080 --> 00:28:47.680 A few of them are available online and have really good documentation. 00:28:47.680 --> 00:28:51.360 Some of them are available on GitHub or the Computational 00:28:51.360 --> 00:28:55.040 Infrastructure for Geophysics. And then I think one or two 00:28:55.040 --> 00:28:57.520 are available on Bitbucket too. 00:28:57.520 --> 00:29:01.680 Others are – you need to contact the authors, but all authors are very excited, 00:29:01.680 --> 00:29:05.120 unless they’ve had already to move over to industry for their employment – 00:29:05.120 --> 00:29:06.880 are very excited to have you use their code. 00:29:06.880 --> 00:29:11.200 So definitely go see our paper. And if you have problems getting 00:29:11.200 --> 00:29:16.654 access to it, just send me an email, and I’ll give you – give you a copy. 00:29:17.760 --> 00:29:21.040 Okay. So now we need to choose the code for our own work. 00:29:21.040 --> 00:29:23.896 And we used Michael Barall’s finite element code. 00:29:23.920 --> 00:29:28.160 And he wrote about it in a GJI paper in 2009. 00:29:28.160 --> 00:29:31.736 I know there’s also documentation about it on our SCEC website. 00:29:31.760 --> 00:29:35.440 And it’s been used to run all the benchmark exercises that we’ve done. 00:29:35.440 --> 00:29:38.080 And I really like to collaborate with Michael Barall, 00:29:38.080 --> 00:29:41.736 so that is the code that we use for our work. 00:29:41.760 --> 00:29:43.680 Okay, so we used FaultMod. 00:29:43.680 --> 00:29:47.176 Step 1, we’ve chosen our code. 00:29:47.200 --> 00:29:52.000 Step 2 – now we need to target this to our specific case 00:29:52.000 --> 00:29:54.160 that we’re trying to model. 00:29:54.160 --> 00:29:58.080 We need to figure out the correct ingredients for our Rodgers Creek- 00:29:58.080 --> 00:30:02.216 Hayward-Calaveras-Northern Calaveras dynamic rupture simulations. 00:30:02.240 --> 00:30:05.360 So we need to figure out our fault geometry, rock properties, 00:30:05.360 --> 00:30:08.240 initial stresses, and friction behavior. 00:30:08.240 --> 00:30:11.336 And so far, we’ve done Step 1, which is the code. 00:30:11.360 --> 00:30:12.960 Now we got to figure out all the rest of it. 00:30:12.960 --> 00:30:15.656 We got to figure out all these guys. 00:30:15.680 --> 00:30:17.600 Okay. So let’s take us back to our setting. 00:30:17.600 --> 00:30:21.440 This is a really nice figure that Luke Blair put together for me 00:30:21.440 --> 00:30:24.880 for our JGR paper. I annotated it a little bit, 00:30:24.880 --> 00:30:28.400 so anything that looks a little sketchy is because I annotated it. 00:30:28.400 --> 00:30:31.736 But Luke gave me a great figure to start with. 00:30:31.760 --> 00:30:34.800 All right. So we have our San Andreas Fault that we’re all familiar with. 00:30:34.800 --> 00:30:38.640 Then we have our Rodgers Creek, Hayward, Calaveras, 00:30:38.640 --> 00:30:41.440 and Northern Calaveras. These are – this is the fault model 00:30:41.440 --> 00:30:45.656 that we’re going to use, and you’ll see a side view of it in the next slide. 00:30:45.680 --> 00:30:48.640 Important features of it are – I should say, first of all, 00:30:48.640 --> 00:30:51.440 the Rodgers Creek Fault actually does go a little farther north, 00:30:51.440 --> 00:30:54.320 but for computational reasons, we had to truncate there. 00:30:54.320 --> 00:30:57.440 Our model got too big for our memory – computer memory. 00:30:57.440 --> 00:31:02.216 So we ended up truncating it near the city of Santa Rosa. 00:31:02.240 --> 00:31:06.536 We had this thing here called Cf, and that’s the connector fault. 00:31:06.560 --> 00:31:10.320 This is a fault that was discovered by Janet Watt, Dave Ponce, 00:31:10.320 --> 00:31:14.856 and colleagues, and it was imaged using shallow geophysical studies 00:31:14.880 --> 00:31:19.680 and imaged beneath San Pablo Bay. And this is a very important link in 00:31:19.680 --> 00:31:22.880 our modeling between the Rodgers Creek Fault and the Hayward Fault 00:31:22.880 --> 00:31:25.600 so that they’re actually a seamless fault structure. 00:31:25.600 --> 00:31:28.880 Without this connector fault, earthquakes starting on the Rodgers 00:31:28.880 --> 00:31:32.400 Creek Fault in our model are unable to jump across to the Hayward Fault, 00:31:32.400 --> 00:31:37.600 and vice versa. So we include this – we include this connector fault because 00:31:37.600 --> 00:31:42.240 we know at least it exists in the upper 100 meters or 00:31:42.240 --> 00:31:44.776 few hundred meters of the Earth’s crust. 00:31:44.800 --> 00:31:47.600 We have Point Pinole, which is point zero on our plots. 00:31:47.600 --> 00:31:50.480 The Hayward Fault. And, right next to the Hayward Fault, of course, 00:31:50.480 --> 00:31:53.736 goes through the city of Oakland. This is the right star. 00:31:53.760 --> 00:31:57.816 Isn’t super far away from the city of San Francisco, the left star. 00:31:57.840 --> 00:32:00.560 Traveling down the Hayward Fault, the Hayward Fault – when I first 00:32:00.560 --> 00:32:04.720 started this work many years ago, we didn’t know yet that the Hayward 00:32:04.720 --> 00:32:07.120 and Calaveras were connected. They’re not obviously connected 00:32:07.120 --> 00:32:09.920 at their surface, but at depth, they are connected. 00:32:09.920 --> 00:32:15.200 And that’s been seen by work by Bob Simpson, work by Dave Manaker 00:32:15.200 --> 00:32:21.016 and Andy Michael, and work by Dave Ponce and [audio garbled], 00:32:21.040 --> 00:32:25.280 so a bunch of people have seen this, that they actually connect. 00:32:25.280 --> 00:32:28.456 Then we have the San Jose – city of a million people. 00:32:28.480 --> 00:32:33.680 Calaveras Fault, and we truncate it at – just opposite the city of Gilroy. 00:32:33.680 --> 00:32:36.720 It does go also a little farther and merge into the San Andreas Fault, 00:32:36.720 --> 00:32:39.976 but for computational reasons, we needed to end there. 00:32:40.000 --> 00:32:42.320 Then we have the Northern Calaveras Fault over here, 00:32:42.320 --> 00:32:47.256 which is a branch off of the main fault surface. 00:32:47.280 --> 00:32:49.200 So what we’re going to do is we’re going to be nucleating 00:32:49.200 --> 00:32:53.896 earthquakes on the Rodgers Creek Fault, the Hayward Fault, 00:32:53.920 --> 00:32:56.720 Calaveras Fault, and the Northern Calaveras Fault. 00:32:56.720 --> 00:32:59.736 We’re going to start them there and then see what happens. 00:32:59.760 --> 00:33:02.640 Okay, so here’s a side view of our fault geometry. 00:33:02.640 --> 00:33:07.680 It’s a 3D fault geometry, and it’s set in a 3D mesh. 00:33:07.680 --> 00:33:12.000 This is our schematic of our – not quite a schematic, but a coarse view 00:33:12.000 --> 00:33:16.080 of our 3D finite element mesh. The top view, just showing 00:33:16.080 --> 00:33:19.600 the fault surface itself. And it’s dark blue around it because 00:33:19.600 --> 00:33:22.880 we’re using smaller elements closer to the fault, bigger elements 00:33:22.880 --> 00:33:26.056 farther away from the faults. And then here’s a side view 00:33:26.080 --> 00:33:30.216 going from southeast to north-35-west. 00:33:30.240 --> 00:33:34.640 Central Calaveras, Hayward, connector faults imaged by 00:33:34.640 --> 00:33:36.536 Janet Watt and colleagues. 00:33:36.560 --> 00:33:38.960 And then the Rodgers Creek Fault, and then we have this kind of 00:33:38.960 --> 00:33:43.656 fancy branch off the Northern Calaveras Fault there. 00:33:43.680 --> 00:33:46.640 We’re going to nucleate earthquakes on each of these – 00:33:46.640 --> 00:33:51.280 on the Rodgers Creek Fault here, Hayward there, Central Calaveras 00:33:51.280 --> 00:33:54.880 there, and Northern Calaveras there. And, of these, three of them are 00:33:54.880 --> 00:33:58.080 a little bit arbitrary locations for nucleation. 00:33:58.080 --> 00:34:01.280 The Hayward Fault is the one that seems to have a physical basis for 00:34:01.280 --> 00:34:05.120 starting there, and that’s based on work by Dave Ponce and colleagues, 00:34:05.120 --> 00:34:08.400 who noted that the San Leandro Gabbro geologic unit that you’ll see 00:34:08.400 --> 00:34:12.080 in the next slide – the next two slides appears to concentrate 00:34:12.080 --> 00:34:16.160 the stress in that region. So they propose that this could be 00:34:16.160 --> 00:34:19.840 a likely nucleation site for large Hayward Fault earthquakes. 00:34:21.760 --> 00:34:25.840 Oops. Go back. All right. Now let’s go to our rock properties. 00:34:25.840 --> 00:34:28.080 We’ve got our fault geometry. Now we have our rock properties. 00:34:28.080 --> 00:34:30.296 So here are our rock units. 00:34:30.320 --> 00:34:36.960 And they – this is a really nice model. I don’t think we have this in 00:34:36.960 --> 00:34:41.416 too many other places, even in the world – the 3D geologic model. 00:34:41.440 --> 00:34:45.440 And it starts with the Big Bay Model. So this is – and I should say these are 00:34:45.440 --> 00:34:48.400 mapped – these are horizontal slices through the 3D model at 00:34:48.400 --> 00:34:51.840 140 meters’ depth. And, at 7 kilometers’ depth, 00:34:51.840 --> 00:34:55.496 we have a whole 3D model, but these are horizontal slices through it. 00:34:55.520 --> 00:34:59.680 And the Big Bay Model is, I think, probably still in publication too – 00:34:59.680 --> 00:35:04.480 even though it’s got a ton of work in it, is by Bob Jachens and co-authors. 00:35:04.480 --> 00:35:09.840 So that’s the bigger view – bigger view. And then the smaller view, which is 00:35:09.840 --> 00:35:16.000 in this little light gray rectangle that’s a little hard to see on this – on this figure, 00:35:16.000 --> 00:35:22.880 is the finer-scale geologic unit structure published in – presented in 00:35:22.880 --> 00:35:27.440 Russ Graymer’s paper – and co-authors paper in 2005 and Jeff Phelps and 00:35:27.440 --> 00:35:30.560 co-authors papers in 2008. And I should also note that there’s 00:35:30.560 --> 00:35:35.920 a lot of overlap among the authors. So this is a really impressive geological 00:35:35.920 --> 00:35:38.960 study that has a nice 3D geologic model. 00:35:38.960 --> 00:35:43.576 And we used this geologic model to then convert that information 00:35:43.600 --> 00:35:47.840 into physical properties. So what are the P wave, S wave – 00:35:47.840 --> 00:35:52.325 S wave velocities of the rocks, and then what are their densities? 00:35:54.000 --> 00:35:56.320 Okay, so now let’s take a side view of this. 00:35:56.320 --> 00:36:00.400 This is looking along the fault surfaces. Remember, before we had our central – 00:36:00.400 --> 00:36:05.496 this is going from southeast to northwest, looking along strike. 00:36:05.520 --> 00:36:09.680 And then this is depth. The vertical axis is depth in kilometers. 00:36:09.680 --> 00:36:13.680 So we’re going Central Calaveras, Hayward, connector fault, 00:36:13.680 --> 00:36:16.880 Rodgers Creek Fault, and now we’re blowing apart the two sides 00:36:16.880 --> 00:36:20.320 of the fault – not literally blowing them up, but just separating them out. 00:36:20.320 --> 00:36:25.120 So we have the east face of that fault surface, the west face of that 00:36:25.120 --> 00:36:29.520 fault surface, and now we’re seeing which rock types – which rock units 00:36:29.520 --> 00:36:32.560 touch each side of the fault. And we do the same thing for 00:36:32.560 --> 00:36:35.840 the Northern Calaveras Fault. The east face of the Northern Calaveras 00:36:35.840 --> 00:36:40.782 Fault surface and the west face of the Northern Calaveras Fault surface. 00:36:40.782 --> 00:36:44.160 The main thing to notice here is that San Leandro Gabbro, which 00:36:44.160 --> 00:36:50.456 I mentioned briefly a little bit ago, and that’s this orange rectangle here. 00:36:50.480 --> 00:36:55.840 So it is a strong, dense rock that’s – the fact that it’s strong will come up 00:36:55.840 --> 00:36:58.240 later when I talk about friction, but it’s also a dense rock – 00:36:58.240 --> 00:37:03.120 has higher velocities – wave propagation velocities. 00:37:03.120 --> 00:37:06.000 And then, right next to it, we have the serpentinite, and that’s 00:37:06.000 --> 00:37:10.800 a weaker kind of slower rock. That’s on the east face. 00:37:10.800 --> 00:37:14.080 And when you go to the west face, you also see the San Leandro Gabbro 00:37:14.080 --> 00:37:19.600 again touching the west face of the fault, but it only goes down this far. 00:37:19.600 --> 00:37:22.000 It only goes down a few kilometers on the west face. 00:37:22.000 --> 00:37:25.200 It doesn’t extend as deep as it does on the other face. 00:37:25.200 --> 00:37:28.640 So we’ll need to deal with that later on when we’re doing friction – 00:37:28.640 --> 00:37:31.520 the fact that you have San Leandro Gabbro on one side at depth, 00:37:31.520 --> 00:37:36.080 but something else on the other side at the same depth. 00:37:36.080 --> 00:37:39.280 But, for right now, we’re looking at the velocities – especially the shear 00:37:39.280 --> 00:37:44.080 wave velocities, and the densities. Okay, so we have our geologic model, 00:37:44.080 --> 00:37:47.416 and we’re going to see this again just a little bit smaller. 00:37:47.440 --> 00:37:50.960 Here, that same figure I just showed you – east face and west face of 00:37:50.960 --> 00:37:54.000 the main fault surface – Central Calaveras, Hayward, 00:37:54.000 --> 00:37:56.880 connector, Rodgers Creek, and then Northern Calaveras. 00:37:56.880 --> 00:37:59.280 And now we’re going to turn that into a shear wave velocity. 00:37:59.280 --> 00:38:02.400 So, for our 3D simulations, we need to know what the shear 00:38:02.400 --> 00:38:06.160 wave velocities are throughout our whole big, giant structure. 00:38:06.160 --> 00:38:10.880 We need to know what the densities are so we can do our wave propagation, 00:38:10.880 --> 00:38:13.280 and the shear modulus tells us how much slip we’ll get. 00:38:13.280 --> 00:38:15.016 We need all that stuff. 00:38:15.040 --> 00:38:20.616 So we convert from the geologic model – 3D geologic model 00:38:20.640 --> 00:38:24.960 to a 3D model of shear wave velocity and density. 00:38:24.960 --> 00:38:28.856 And the way we do that is using the equations of Tom Brocher. 00:38:28.880 --> 00:38:32.240 So he put together equations – I think it’s published in 2005 – 00:38:32.240 --> 00:38:38.376 relating the rock units to Vs, Vp/Vs, and density. 00:38:38.400 --> 00:38:39.920 All right. So we have our rock properties. 00:38:39.920 --> 00:38:43.656 Now we need to know about friction. How does friction work? 00:38:43.680 --> 00:38:49.016 So some people like to use very fancy friction when they do their simulations. 00:38:49.040 --> 00:38:51.920 And this is a good idea of you’re doing, say, earthquake cycle models. 00:38:51.920 --> 00:38:54.720 You do need to do that so you’re not only doing the earthquake itself, 00:38:54.720 --> 00:38:58.456 but you’re also doing what happens during the postseismic time. 00:38:58.480 --> 00:39:00.480 Then you need to do the interseismic time. 00:39:00.480 --> 00:39:04.056 So that – you need to do that. You need to do fancier friction. 00:39:04.080 --> 00:39:06.640 But we are just doing the earthquake itself. 00:39:06.640 --> 00:39:11.200 We’re just doing the coseismic rupture. And, for that, as far as I understand 00:39:11.200 --> 00:39:15.680 from talking to people who do the lab experiments, that that is fine to do that. 00:39:15.680 --> 00:39:20.960 Slip-weakening is the same as rate-state friction if you have – 00:39:20.960 --> 00:39:23.920 if the parameters are similar. And this – and Kenny Ryan and 00:39:23.920 --> 00:39:27.360 David Oglesby showed this a while ago – maybe seven years ago 00:39:27.360 --> 00:39:31.336 in a paper they published. And others have also shown this to be true. 00:39:31.360 --> 00:39:33.360 So I’m going to use slip-weakening. And the thing I like about 00:39:33.360 --> 00:39:37.416 slip-weakening is it doesn’t require as many parameters to be assumed. 00:39:37.440 --> 00:39:40.000 So, for slip-weakening, behaves kind of like Coulomb. 00:39:40.000 --> 00:39:43.976 It’s kind of like a Coulomb friction with some extra features. 00:39:44.000 --> 00:39:47.440 So you start out – the strength of the two sides of the faults that – 00:39:47.440 --> 00:39:51.600 what lets them slip past each other or not starts out as a static coefficient 00:39:51.600 --> 00:39:55.497 of friction times your normal stress, which is time-dependent. 00:39:56.720 --> 00:40:01.360 And then, as the fault gradually slips up to this critical slip-weakening distance, 00:40:01.360 --> 00:40:05.840 the strength linearly decreases down to a dynamic coefficient of friction times 00:40:05.840 --> 00:40:09.680 the time-dependent normal stress. So you don’t instantly drop as soon as – 00:40:09.680 --> 00:40:13.096 as soon as the fault starts slipping. Instead, there’s a gradual drop. 00:40:13.120 --> 00:40:16.936 And this actually uses up some energy in the process. 00:40:16.960 --> 00:40:21.760 Slip-weakening critical distance has been inferred from inversions 00:40:21.760 --> 00:40:27.497 of seismological source data to be anywhere from, like, 0.1 meters – 00:40:27.497 --> 00:40:31.096 or I should – or even less than that, all the way up to 5 meters. 00:40:31.120 --> 00:40:35.920 So, for our study, we used 30 centimeters – 0.3 meters – 00:40:35.920 --> 00:40:39.920 as our slip-weakening critical distance. Some people do a fancier model, and 00:40:39.920 --> 00:40:44.616 they might say, oh, it should vary over the fault with depth or along strike. 00:40:44.640 --> 00:40:49.600 There’s very little, maybe no, direct evidence of how we should vary that 00:40:49.600 --> 00:40:55.416 parameter. So, for ours, we just keep it constant at 0.3 meters. 00:40:55.440 --> 00:40:59.280 Then we need to figure out our static and dynamic coefficients of friction. 00:40:59.280 --> 00:41:04.480 So what we do is go back to our geology and our lab studies to look at 00:41:04.480 --> 00:41:08.936 sliding coefficients of friction that have been measured in the lab. 00:41:08.960 --> 00:41:12.696 So we look at the geology for this in lab experiments. 00:41:12.720 --> 00:41:17.840 And then the static coefficient of friction has been related to dynamic 00:41:17.840 --> 00:41:21.920 coefficient of friction as being about 20% bigger, and that’s work 00:41:21.920 --> 00:41:25.716 by Teng‐Fong Wong that he published a wile ago. 00:41:27.040 --> 00:41:33.360 All right. So let’s go back. For our geology, our main rocks that 00:41:33.360 --> 00:41:36.160 are pretty different along our fault surfaces – we’re back looking at 00:41:36.160 --> 00:41:39.336 our fault surfaces again – east face and west face. 00:41:39.360 --> 00:41:42.240 So we noted that San Leandro Gabbro earlier. 00:41:42.240 --> 00:41:45.576 It’s a strong rock – much stronger than everything else. 00:41:45.600 --> 00:41:48.240 We have our San Leandro Gabbro. Right next to it, we have our 00:41:48.240 --> 00:41:50.696 serpentinite, which is a much weaker rock. 00:41:50.720 --> 00:41:54.080 And there were field samples collected on the Hayward Fault, and they were 00:41:54.080 --> 00:41:58.720 studied carefully, and the results were published in a California Geological 00:41:58.720 --> 00:42:02.456 Survey report by Carolyn Morrow and colleagues. 00:42:02.480 --> 00:42:05.751 In addition, there’s been a bunch of work, of course, for the San Andreas 00:42:05.751 --> 00:42:08.800 Fault at depth – San Andreas Fault Observatory at Depth. 00:42:08.800 --> 00:42:12.080 They had a lot of lab work done on those samples from there. 00:42:12.080 --> 00:42:16.560 And those are discussed a lot in many papers, including one by Diane Moore 00:42:16.560 --> 00:42:21.496 and colleagues that’s published in Journal of Structural Geology in 2016. 00:42:21.520 --> 00:42:25.440 So the San Leandro Gabbro from the Morrow et al. results has 00:42:25.440 --> 00:42:27.920 a dynamic sliding coefficient of friction of about 0.8, 00:42:27.920 --> 00:42:32.080 but due to the – that’s at the surface. But, as you go to depth, it makes sense 00:42:32.080 --> 00:42:35.920 for it to be a little lower, so we choose a dynamic sliding coefficient 00:42:35.920 --> 00:42:40.160 of friction of 0.65. Then work on serpentinite – 00:42:40.160 --> 00:42:43.520 lab experiments on serpentinite show a dynamic sliding coefficient 00:42:43.520 --> 00:42:46.376 of friction of 0.3 to be appropriate. 00:42:46.400 --> 00:42:49.760 The rest of the rocks don’t behave that differently from each other 00:42:49.760 --> 00:42:54.160 in lab experiments. So we assign most of the rest of 00:42:54.160 --> 00:42:57.360 the rocks, many of which are Franciscan mélange, 00:42:57.360 --> 00:43:01.496 a dynamic sliding coefficient of friction of 0.5. 00:43:01.520 --> 00:43:04.720 All right. So now we are back to the situation where we have the 00:43:04.720 --> 00:43:08.880 San Leandro Gabbro – that really strong rock on one side at depth, 00:43:08.880 --> 00:43:12.480 but not on the other side. Also have the serpentinite on one side. 00:43:12.480 --> 00:43:16.056 It doesn’t match up with the serpentinite on the other side. 00:43:16.080 --> 00:43:18.480 So what do we do when we have two different strengths of rocks 00:43:18.480 --> 00:43:21.840 on each side of the fault? So what we do is we end up using 00:43:21.840 --> 00:43:27.496 the weaker value between the two. And this is work that Collettini et al. 00:43:27.520 --> 00:43:31.200 published in 2009 that we should do that approach, and then also work by 00:43:31.200 --> 00:43:34.376 Dave Lockner and Diane Moore and co-authors. 00:43:34.400 --> 00:43:37.680 So this is what we end up for our final dynamic sliding coefficient 00:43:37.680 --> 00:43:41.840 of friction along this fault surface for Central Calaveras, Hayward, 00:43:41.840 --> 00:43:45.200 connector, and Rodgers Creek. Now, the Northern Calaveras, 00:43:45.200 --> 00:43:49.280 it has lots of cool rocks abutting it, but they’re – we don’t have – at least, 00:43:49.280 --> 00:43:52.560 from the views that we have, or the information that we have, 00:43:52.560 --> 00:43:57.280 there’s no serpentinite or San Leandro Gabbro abutting it, so we just give it 00:43:57.280 --> 00:44:01.976 a coefficient of friction – sliding coefficient of friction of 0.5. 00:44:02.000 --> 00:44:06.216 All right. We’ve got our friction – our fault geometry, got our friction, 00:44:06.240 --> 00:44:08.960 and now we’ve got to figure out our initial shear stresses. 00:44:08.960 --> 00:44:12.376 So this is when we get kind of creative. 00:44:12.400 --> 00:44:17.416 We don’t know what they actually are, but we can assume them. 00:44:17.440 --> 00:44:21.840 And the technique I did is to say that, when you have a creeping fault, 00:44:21.840 --> 00:44:25.840 it’s actually relieving some of the accumulating tectonic strain that 00:44:25.840 --> 00:44:31.896 might be building up on a locked fault. So first we had to go and construct 00:44:31.920 --> 00:44:38.056 a model for the creep rate on this – on these fault surfaces that we have here. 00:44:38.080 --> 00:44:43.587 So Estelle Chaussard had already presented a creep rate figure – 00:44:43.587 --> 00:44:47.416 or, picture for Central Calaveras and part of the Hayward Fault. 00:44:47.440 --> 00:44:50.640 And then Gareth Funning, in his work, had a model for 00:44:50.640 --> 00:44:55.336 the Hayward Fault, which he will hopefully publish really soon. 00:44:55.360 --> 00:44:58.480 So we have the Hayward Fault, but we don’t have one big model 00:44:58.480 --> 00:45:00.880 for the whole thing. So we merged those two. 00:45:00.880 --> 00:45:03.760 And then, for the Rodgers Creek Fault, we used the wisdom of Gareth. 00:45:03.760 --> 00:45:06.616 The Rodgers Creek Fault does creep farther north. 00:45:06.640 --> 00:45:09.920 But it does not have much of a creep rate, if any creep rate, 00:45:09.920 --> 00:45:12.240 as you go into the section that we are looking at. 00:45:12.240 --> 00:45:16.720 So we put together this great big model. Northern Calaveras, we used – 00:45:16.720 --> 00:45:21.920 Estelle Chaussard’s model does cover our entire Northern Calaveras Fault. 00:45:21.920 --> 00:45:24.560 So we use her model, and that’s published in her 00:45:24.560 --> 00:45:28.376 co-author paper in JGR 2015. 00:45:28.400 --> 00:45:31.520 So now what we do is we look at where the creep rate is 00:45:31.520 --> 00:45:35.760 1 millimeter per year or faster or 3 millimeters per year or faster. 00:45:35.760 --> 00:45:38.616 So that’s the interseismic creep rate. 00:45:38.640 --> 00:45:42.160 And anywhere where it’s either one or the other – and we look at them 00:45:42.160 --> 00:45:45.440 individually, so then we have two different images of what the creeping 00:45:45.440 --> 00:45:49.680 pattern looks like on the fault. So, say our 3 millimeters per year 00:45:49.680 --> 00:45:53.440 slip rate, we use that or faster. Then, anywhere where it’s slipping 00:45:53.440 --> 00:46:00.080 faster than that, we assign it a lower initial sheer stress. 00:46:00.080 --> 00:46:03.336 Okay, so let’s go [inaudible] with that. 00:46:03.360 --> 00:46:06.560 So we have our fault geometry, our rock properties, our initial – 00:46:06.560 --> 00:46:09.600 our friction, and our initial stresses. We have our code. 00:46:09.600 --> 00:46:12.056 And let’s go see some results. 00:46:12.080 --> 00:46:15.256 Scenario large earthquakes. 00:46:15.280 --> 00:46:20.080 Okay, so remind us again, we’re going to start scenarios 00:46:20.080 --> 00:46:23.840 that start on the Rodgers Creek Fault, scenario earthquakes that start on the 00:46:23.840 --> 00:46:27.680 Hayward Fault, scenarios that start on the Central Calaveras, and then 00:46:27.680 --> 00:46:30.400 scenarios that start on the Northern Calaveras. 00:46:30.400 --> 00:46:32.880 We start an earthquake. We don’t know if it’s going to go anywhere. 00:46:32.880 --> 00:46:35.680 We don’t know how big it’s going to get. That’s the part of the 00:46:35.680 --> 00:46:40.216 spontaneous rupture propagation. We find out what happens. 00:46:40.240 --> 00:46:43.040 All right. So let’s start here. Pretend the right side 00:46:43.040 --> 00:46:45.416 of the slide is not there. Just look at the left side. 00:46:45.440 --> 00:46:46.880 So we’re going to do a locked scenario. 00:46:46.880 --> 00:46:49.840 So, in this case, we pretend that the faults do not know 00:46:49.840 --> 00:46:52.160 that they’re creeping. They just behave like regular 00:46:52.160 --> 00:46:55.440 garden-variety faults. We’re going to do a nucleation 00:46:55.440 --> 00:46:58.320 on the Rodgers Creek Fault, and then the contours that you see – 00:46:58.320 --> 00:47:02.000 so, sorry, I should say, this is depth. This is distance along strike. 00:47:02.000 --> 00:47:05.280 So we’re looking at the fault surface. We’re nucleating here on the 00:47:05.280 --> 00:47:08.000 Rodgers Creek Fault, and then the contours that you see are the 00:47:08.000 --> 00:47:11.120 progress of the rupture at half-second intervals as it’s propagating outward. 00:47:11.120 --> 00:47:14.960 So it’s going out, out, out, out. See it all the way – and it gets all the 00:47:14.960 --> 00:47:18.880 way to the southern end of the fault. All the way down the Central 00:47:18.880 --> 00:47:23.256 Calaveras, all the way to the end, and then all the way to its northern end. 00:47:23.280 --> 00:47:26.640 We do the same thing. We start a nucleation – 00:47:26.640 --> 00:47:29.096 nucleate a rupture on the Hayward Fault. 00:47:29.120 --> 00:47:31.920 Propagates, propagates, gets all the way to that end. 00:47:31.920 --> 00:47:34.776 Propagates outwards. Gets all the way to the other end. 00:47:34.800 --> 00:47:38.000 Calaveras nucleation – same thing. Sometimes they get slowed up a bit, 00:47:38.000 --> 00:47:40.320 and that’s where you see these bunched-up contours, 00:47:40.320 --> 00:47:43.120 but still makes it all the way. And the Calaveras nucleation 00:47:43.120 --> 00:47:46.296 actually triggers a bit on the Northern Calaveras Fault. 00:47:46.320 --> 00:47:49.600 Northern Calaveras nucleation propagates all over its whole fault, 00:47:49.600 --> 00:47:52.320 and it always seems to do this in all our simulations. 00:47:52.320 --> 00:47:57.200 This one actually manages to jump across, trigger Calaveras-Hayward and 00:47:57.200 --> 00:48:01.736 then propagate the main surface too. Then, on the right side of the slide, 00:48:01.760 --> 00:48:04.800 we see the final slip patterns for each of these situations. 00:48:04.800 --> 00:48:10.080 And this is final slip and – contoured up to a little more than 6 meters. 00:48:10.080 --> 00:48:12.320 So it’s hearty slip. A lot of slip. 00:48:12.320 --> 00:48:14.876 And these are the locked fault cases. 00:48:15.760 --> 00:48:18.080 All right. And then snapshots of the ground shaking at 00:48:18.080 --> 00:48:22.880 15 seconds after nucleation, 30 seconds after nucleation – 00:48:22.880 --> 00:48:27.416 this is the ground shaking at the Earth’s surface. So this is a map view. 00:48:27.440 --> 00:48:30.000 And, for all these different things. And the main purpose of this figure 00:48:30.000 --> 00:48:32.480 is just to show you that, yes, indeed, we are calculating 00:48:32.480 --> 00:48:35.115 ground motion in these models. 00:48:36.240 --> 00:48:39.680 All right. Now let’s go say that we do care about the fact that that faults 00:48:39.680 --> 00:48:42.696 are creeping, that does make a difference to these earthquakes. 00:48:42.720 --> 00:48:46.960 And now we’re going to look at where the creeping rate – the creep rate 00:48:46.960 --> 00:48:51.256 was 3 millimeters per year or higher. So these parts. 00:48:51.280 --> 00:48:57.416 And then assign those lower initial sheer stresses with the idea that 00:48:57.440 --> 00:49:00.480 some of the accumulating tectonic strain has already been relieved 00:49:00.480 --> 00:49:04.208 through the interseismic creep. Oops, sorry. 00:49:05.520 --> 00:49:08.720 Okay, nope. Went too far. All right. So let’s look at creeping 00:49:08.720 --> 00:49:12.480 patches – left side of the slide. Creeping patches are highlighted 00:49:12.480 --> 00:49:16.536 by the shading here. So these have lower initial sheer stresses. 00:49:16.560 --> 00:49:21.120 Rodgers Creek nucleation propagates to the north, propagates to the south, 00:49:21.120 --> 00:49:26.800 slows down a bit, slows down a bit, slows down a bit, and then stops – 00:49:26.800 --> 00:49:31.256 almost makes it all the way to the end of the Calaveras, but not quite. 00:49:31.280 --> 00:49:38.776 Hayward nucleation – slows down but makes it through going to the south. 00:49:38.800 --> 00:49:41.976 Gets slowed down and then stops early. 00:49:42.000 --> 00:49:45.040 Northern Calaveras nucleation always goes. Has no problem 00:49:45.040 --> 00:49:48.560 ever in our model. Makes it all the way through 00:49:48.560 --> 00:49:52.115 and then jumps a little bit but then stops. 00:49:54.800 --> 00:49:59.256 And then just samples of what the ground shaking looks like. 00:49:59.280 --> 00:50:02.720 All right. Now let’s use the 1 millimeter per year contour to find where the 00:50:02.720 --> 00:50:06.320 fault is creeping significantly and thereby has lower initial sheer stress. 00:50:06.320 --> 00:50:10.560 So this is our biggest creeping section. So this whole part is, 00:50:10.560 --> 00:50:13.760 with a few exceptions. And then Northern Calaveras, 00:50:13.760 --> 00:50:15.982 it’s a bigger patch also. 00:50:17.360 --> 00:50:20.640 So, for here – oh, and I should say, for the last one, the Central Calaveras 00:50:20.640 --> 00:50:23.176 one nucleated but didn’t go anywhere. 00:50:23.200 --> 00:50:26.720 Here, we have – for the 1 millimeter per year contour, so this whole 00:50:26.720 --> 00:50:31.520 shaded area is all creeping. We have Rodgers Creek nucleation 00:50:31.520 --> 00:50:35.200 propagates down, gets slower, and then stops. 00:50:35.200 --> 00:50:39.176 So it makes it a little ways down the Hayward Fault and then stops. 00:50:39.200 --> 00:50:42.960 Hayward nucleation starts, goes nowhere. 00:50:42.960 --> 00:50:46.160 Central Calaveras nucleation starts, goes nowhere. 00:50:46.160 --> 00:50:52.888 Northern Calaveras nucleation ruptures the entire Northern Calaveras Fault. 00:50:54.320 --> 00:50:56.950 And snapshots of ground shaking. 00:50:59.680 --> 00:51:03.600 All right. So, to summarize, we’ve used the spontaneous, 00:51:03.600 --> 00:51:07.520 or dynamic, rupture method to simulate large earthquakes on the Rodgers 00:51:07.520 --> 00:51:11.656 Creek, Hayward, Calaveras, and Northern Calaveras Fault system. 00:51:11.680 --> 00:51:14.880 We produced large simulated earthquakes, some of which are 00:51:14.880 --> 00:51:19.411 stopped by the creeping sections, and some of which aren’t stopped. 00:51:20.400 --> 00:51:26.456 The most important factors were the nucleation site, the fault geometry, 00:51:26.480 --> 00:51:29.096 and the pattern of fault creep. 00:51:29.120 --> 00:51:31.600 And, in the future, we have a bunch more things that we want to do, 00:51:31.600 --> 00:51:35.920 and included in this list is modeling a lengthier fault system. 00:51:35.920 --> 00:51:38.400 I think we have more computer memory, so we could make the faults 00:51:38.400 --> 00:51:40.696 a little bit longer, as they should be. 00:51:40.720 --> 00:51:44.640 And then also looking at things such as the effects of inelastic yielding 00:51:44.640 --> 00:51:48.376 and using different distributions of the initial stresses. 00:51:48.400 --> 00:51:50.160 We have our new published paper. 00:51:50.160 --> 00:51:53.310 And there’s just another advertisement for it. 00:51:57.120 --> 00:52:00.320 Okay, so now we’re going to look at movies from a case that’s a little 00:52:00.320 --> 00:52:03.920 variation on what you just saw. And, for this one, we’re going to 00:52:03.920 --> 00:52:08.456 use the 3 millimeters per year with 3 kilometers of cohesion. 00:52:08.480 --> 00:52:11.760 We’re going to look at a Hayward nucleation, so it’s this case here – 00:52:11.760 --> 00:52:15.976 this second-line case. Hayward nucleation. 00:52:16.000 --> 00:52:18.720 And the previous ones didn’t have cohesion at all, so we have our 00:52:18.720 --> 00:52:21.920 creeping patch here, but then we also add a little factor of cohesion 00:52:21.920 --> 00:52:25.920 in the upper parts of the fault surface of the faults, and that’s to make it 00:52:25.920 --> 00:52:28.400 so that you don’t have quite as strong ground shaking. 00:52:28.400 --> 00:52:31.040 And, in simulations, sometimes interactions with the free surface 00:52:31.040 --> 00:52:33.760 causes a huge amount of ground shaking that probably doesn’t happen 00:52:33.760 --> 00:52:37.680 in nature, and there’s probably damage and stuff that prevents that in reality. 00:52:37.680 --> 00:52:40.960 So we do that – we accommodate that by including cohesion. 00:52:40.960 --> 00:52:43.875 So let’s go look at a simulation. 00:52:47.680 --> 00:52:51.708 And this is the one time the laser pointer does not work. 00:52:52.021 --> 00:52:54.223 All right. 00:52:54.895 --> 00:52:56.341 Huh. 00:52:58.060 --> 00:53:00.696 I wonder if I could do that. 00:53:00.720 --> 00:53:03.328 All right. Give me 10 seconds, and if this doesn’t work … 00:53:03.328 --> 00:53:09.096 - Ruth, we’re seeing the laser pointer. - Yeah. But it doesn’t start the movie. 00:53:09.120 --> 00:53:11.680 The laser pointer and the movie seem incompatible. 00:53:11.680 --> 00:53:14.856 Oh, I was so close. 00:53:14.880 --> 00:53:18.800 All right. Anyway, so there’s a fancy movie that you might 00:53:18.800 --> 00:53:21.896 or might not be able to see – that you probably can’t see. 00:53:21.920 --> 00:53:26.376 So there are movies also in the supplement of our paper. 00:53:26.400 --> 00:53:28.640 And now I’m going to have to figure out how to do that in the future – 00:53:28.640 --> 00:53:30.640 how to get out – I don’t want to escape the whole slide show 00:53:30.640 --> 00:53:31.680 because that’ll mess it up. 00:53:31.680 --> 00:53:36.960 - [inaudible] - My spouse is giving me a clue, 00:53:36.960 --> 00:53:41.120 but it is not working. Okay, anyway, so there’s a cool movie. 00:53:41.120 --> 00:53:44.560 And maybe I’ll be able to show it in the Q-and-A. Maybe I’ll figure it out. 00:53:44.560 --> 00:53:49.600 But, in the meantime, I want to say thank you very much for listening, 00:53:49.600 --> 00:53:52.656 and I welcome all questions. 00:53:53.732 --> 00:53:58.880 - All right. Thanks, Ruth. I didn’t catch it all because I got 00:53:58.880 --> 00:54:02.240 booted out a couple times. [laughs] But I’m sure it was great for 00:54:02.240 --> 00:54:04.497 those who were able to hear it all. 00:54:06.000 --> 00:54:07.920 And we’re going to open up to questions now. 00:54:07.920 --> 00:54:10.480 We’ve had a few in the chat, so we’ll get started there. 00:54:10.480 --> 00:54:15.040 And, of course, feel free to type your questions in the chat or raise your hand, 00:54:15.040 --> 00:54:18.960 and we’ll help you turn on your videos and ask your question yourself. 00:54:18.960 --> 00:54:22.880 So the first question was from Joan, and she asks, does your compilation 00:54:22.880 --> 00:54:27.656 of creeping crustal faults include those that exhibit transient creep, 00:54:27.680 --> 00:54:32.800 short-load aseismic slip events, and what is the fraction of these faults 00:54:32.800 --> 00:54:36.960 for which their aseismic and slow-slip behavior is simply unknown due to 00:54:36.960 --> 00:54:41.075 insufficient observations? So a couple questions there. 00:54:41.099 --> 00:54:44.960 - Okay, so this is a question about the Reviews of Geophysics article – 00:54:44.960 --> 00:54:46.747 the compilation? - Yeah. 00:54:46.747 --> 00:54:50.480 - That I did there? So, for those, the creeping faults 00:54:50.480 --> 00:54:56.776 that were included in that list were ones where we have modern geodetic 00:54:56.800 --> 00:55:00.240 observations that the fault creep is occurring. 00:55:00.240 --> 00:55:03.760 And I would say that the list in that paper – there was a big table 00:55:03.760 --> 00:55:09.200 that I put together in that paper – 2017 paper is a subset. 00:55:09.200 --> 00:55:13.600 So I went and did a literature search and looked for everywhere that I could 00:55:13.600 --> 00:55:17.816 find things, but I didn’t include – so, yeah, subduction zones. 00:55:17.840 --> 00:55:23.736 And then, lots of times, there were – there was evidence of postseismic slip, 00:55:23.760 --> 00:55:30.080 but I only included cases where there was a good observation of interseismic 00:55:30.080 --> 00:55:33.920 slip and generally not just at the Earth’s surface but a little bit deeper. 00:55:33.920 --> 00:55:37.536 In that table, too, it also says what the depth extent was. 00:55:38.505 --> 00:55:44.800 - Okay. And then a question from Brad about your models. 00:55:44.800 --> 00:55:49.982 It says, what are you using for the initial normal tractions on the fault? 00:55:50.022 --> 00:55:52.696 - Yeah, so the normal stress is constant. 00:55:52.720 --> 00:55:58.640 I think it’s – going back – it’s 75 MPa, is that right? I have to look again. 00:55:58.640 --> 00:56:01.680 But anyway, it’s a constant value. Yeah, I don’t – I don’t use 00:56:01.680 --> 00:56:04.480 a depth-dependent value. We don’t use a depth-dependent value for 00:56:04.480 --> 00:56:09.416 the normal stress, so we just – we just use a constant value for it. 00:56:09.440 --> 00:56:14.080 And that’s something that – to look at in the future when we look to vary the 00:56:14.080 --> 00:56:18.880 initial stress conditions a little bit more. So, for these dynamic rupture models – 00:56:18.880 --> 00:56:22.640 well, Brad, of course, knows this, but in case other people that don’t 00:56:22.640 --> 00:56:26.000 know, everything is time-dependent. So you’re starting with your initial 00:56:26.000 --> 00:56:29.280 sheer and normal stresses, but then they vary with time as you have 00:56:29.280 --> 00:56:32.400 all the interactions of all the effects, especially the propagating waves 00:56:32.400 --> 00:56:36.720 and the stress waves. So, for this, we have constant 00:56:36.720 --> 00:56:42.320 normal stress at depth because I didn’t know from observations – 00:56:42.320 --> 00:56:45.520 it didn’t seem to indicate how we should vary it with depth. 00:56:45.520 --> 00:56:48.560 But some people instead do – might do a depth-dependent normal stress. 00:56:48.560 --> 00:56:52.809 And I know, Brad, that you have done this for your work. 00:56:52.833 --> 00:56:56.400 - Okay. And then Brad also asked if you’ve compared your 00:56:56.400 --> 00:56:59.990 ground motions to empirical ground motion models. 00:56:59.990 --> 00:57:02.400 - So, I’m a little scared to do that right now. [laughs] 00:57:02.400 --> 00:57:08.800 Our finite element mesh is 250 meters, so we don’t really have great resolution 00:57:08.800 --> 00:57:10.880 for the ground – for the ground shaking. 00:57:10.880 --> 00:57:15.120 It’s kind of more of a qualitative – a qualitative look at the ground motion – 00:57:15.120 --> 00:57:17.360 at the ground motions. We have a really long – 00:57:17.360 --> 00:57:20.160 I guess, maybe we’d be okay in our long periods. 00:57:20.160 --> 00:57:23.600 Another thing we had to do is we had to truncate the shear wave velocities. 00:57:23.600 --> 00:57:28.936 And I wrote down what we truncated it to – down to 1,950 meters per second. 00:57:28.960 --> 00:57:32.640 So we not only have kind of big-ish elements, but we also had to 00:57:32.640 --> 00:57:36.400 truncate the ground motions. So we’re kind of hesitant to 00:57:36.400 --> 00:57:38.800 do that comparison, but I definitely did think about it. 00:57:38.800 --> 00:57:42.960 It would be better to do if we had smaller element sizes and could 00:57:42.960 --> 00:57:47.840 do more of the low velocities. But it’s a good – it’s a good – 00:57:47.840 --> 00:57:50.810 definitely a good thought and something we should do. 00:57:50.810 --> 00:57:52.856 - Okay. Thank you. 00:57:52.880 --> 00:57:56.640 And a question from Janet. Might there be a greater degree of 00:57:56.640 --> 00:58:02.559 cohesion need fault asperities – step-overs, bends, or intersections? 00:58:02.583 --> 00:58:07.520 - So that’s a good question. So we could put the initial stresses 00:58:07.520 --> 00:58:12.880 into – we could either put them into the normal stress or we 00:58:12.880 --> 00:58:17.600 could put them into cohesion. So, yeah. Yeah, that’s a good idea 00:58:17.600 --> 00:58:21.760 to just say that they’re – there fault’s more stuck where you have these 00:58:21.760 --> 00:58:26.320 other – so we have the geometry pinning some parts of the fault. 00:58:26.320 --> 00:58:29.120 And if you could put in – if we had model fault roughness, 00:58:29.120 --> 00:58:32.080 that would also take into account – this into account, where we might 00:58:32.080 --> 00:58:34.800 not need to put into cohesion, but instead, we might end up 00:58:34.800 --> 00:58:38.971 in the normal stress. But that’s a really good idea. 00:58:42.323 --> 00:58:45.920 - Okay, another question from Joan. What does the paleoseismic evidence 00:58:45.920 --> 00:58:50.640 tell us about the persistence of creep on crustal – sorry, on crustal faults 00:58:50.640 --> 00:58:54.296 over time scales of hundreds to thousands of years? 00:58:54.320 --> 00:59:00.240 So that is a big topic, especially for the creeping section of the 00:59:00.240 --> 00:59:04.640 San Andreas Fault. And we want to know. 00:59:04.640 --> 00:59:10.000 And that’s the part of the fault that has – appears to have the fastest 00:59:10.000 --> 00:59:15.120 creep rate relative to, say, deep slip rate of any fault, I think, in the world. 00:59:15.120 --> 00:59:17.600 Although maybe there – maybe there were one or two others in that 00:59:17.600 --> 00:59:21.360 compilation that were also inferred to be creeping really fast or to be 00:59:21.360 --> 00:59:27.120 relieving everything aseismically. So, so far, paleoseismic studies by 00:59:27.120 --> 00:59:33.360 Nate Toké and Ramon Arrowsmith and others have not been able to find 00:59:33.360 --> 00:59:37.816 evidence of large events – really large events in the creeping sections. 00:59:37.840 --> 00:59:42.400 I think this is still a topic of discussion and maybe work that Josie Nevitt 00:59:42.400 --> 00:59:47.176 is doing now for Mee Ranch, maybe that will be able to help 00:59:47.200 --> 00:59:49.920 find some more evidence. But that’s a big question. 00:59:49.920 --> 00:59:53.280 Like, how long – when you have fault creep, how long does that 00:59:53.280 --> 00:59:58.146 last over geologic time? I don’t think that we know that. 00:59:58.146 --> 01:00:00.240 - Yeah, you’re right there. That’s still something 01:00:00.240 --> 01:00:02.456 we are working on figuring out. 01:00:02.480 --> 01:00:04.056 Okay. 01:00:04.080 --> 01:00:07.760 From Jeff. As you go towards more complex models, how does – 01:00:07.760 --> 01:00:13.040 how uncertain is the energy sink aspect of the creeping segment’s rheology 01:00:13.040 --> 01:00:18.960 that you were trying to implement? - Yeah. So, right now, the energy 01:00:18.960 --> 01:00:23.040 sink we – while we used a lower initial sheer stress – 01:00:23.040 --> 01:00:25.336 that’s probably my fault we did that. 01:00:25.360 --> 01:00:28.536 My co-authors went along with it, probably hesitantly. 01:00:28.560 --> 01:00:31.680 So that’s, like, an energy sink – a lower initial sheer stress. 01:00:31.680 --> 01:00:33.760 But there are probably other ways of doing it. 01:00:33.760 --> 01:00:36.400 You could also model it with rate strength and friction, 01:00:36.400 --> 01:00:39.200 but then you need to assume even more parameters. 01:00:39.200 --> 01:00:41.600 We also have our slip-weakening critical distance, although that 01:00:41.600 --> 01:00:45.040 applies for – that’s, like, an energy sink, although that applies for both 01:00:45.040 --> 01:00:50.320 the creeping and the locked sections. So probably the transform fault setting 01:00:50.320 --> 01:00:54.480 is the best way – the best place to look at – look at these sorts of things, 01:00:54.480 --> 01:00:55.760 playing around with the models. 01:00:55.760 --> 01:00:59.760 Because you have so many events and, you know, of course, Jeff, because 01:00:59.760 --> 01:01:03.760 you’ve worked a lot on the oceanic transform faults, but trying to figure out 01:01:03.760 --> 01:01:07.576 what limits – what limits the rupture extent in those settings 01:01:07.601 --> 01:01:11.419 and then just playing around with the models for that. 01:01:11.450 --> 01:01:15.896 - Okay. And another question from Ian. 01:01:15.920 --> 01:01:20.240 In the plots showing slip contours for your various scenarios, it looks like 01:01:20.240 --> 01:01:23.760 the slip consistently slows down shortly before initiating 01:01:23.760 --> 01:01:26.960 before speeding up again. Is this an actual effect? 01:01:26.960 --> 01:01:30.708 And, if so, do you have any idea what may cause it? 01:01:30.708 --> 01:01:34.080 - Yeah, so sometimes that’s slowing down just due to the 01:01:34.080 --> 01:01:36.640 change in fault geometry. Sometimes it’s slowing down 01:01:36.640 --> 01:01:41.680 because they’re encountering other shear wave velocities of the rocks. 01:01:41.680 --> 01:01:46.480 So we have our 3D geologic model, and each of those rocks have 01:01:46.480 --> 01:01:49.680 different shear wave speeds. So there’s a whole combination of stuff. 01:01:49.680 --> 01:01:54.160 We have the initial stresses. And they’re evolving with time. 01:01:54.160 --> 01:01:57.520 The stresses are evolving with time, particularly the sheer stress, 01:01:57.520 --> 01:02:01.016 but also the normal stress, because we don’t have a planar fault. 01:02:01.040 --> 01:02:06.000 We have the fault – the rocks – the different shear wave velocities 01:02:06.000 --> 01:02:10.080 for the rocks. And then we have our geometry. 01:02:10.080 --> 01:02:12.880 So we have all these different things. So we did – we do have a figure 01:02:12.880 --> 01:02:17.040 in our paper – Figure 12 of our paper, where we looked at the relative effects 01:02:17.040 --> 01:02:20.320 of the fault friction and the velocity structure. 01:02:20.320 --> 01:02:25.840 And we did 1D models and our usual 3D models to try to 01:02:25.840 --> 01:02:28.800 see if we could pinpoint stuff. But it’s kind of hard because you 01:02:28.800 --> 01:02:32.376 also have the effects of the free surface coming in there too. 01:02:32.400 --> 01:02:34.960 But, yeah, so it’s a real effect that you really are slowing down 01:02:34.960 --> 01:02:37.896 and then kicking up again and then starting up again. 01:02:37.920 --> 01:02:42.480 But why it does that in some locations is easily understandable, but then, 01:02:42.480 --> 01:02:45.280 in others, is a little more complicated just because you have the interactions 01:02:45.280 --> 01:02:50.083 of all these different factors going on. So that’s a good question. 01:02:50.107 --> 01:02:54.960 - Great. Thanks, everybody. A lot of gratitude for your talk and 01:02:54.960 --> 01:02:58.960 compliments on a great talk, Ruth. We have reached the end of our time, 01:02:58.960 --> 01:03:02.400 but we do invite everyone to stick around a little bit to continue 01:03:02.400 --> 01:03:04.776 on the conversation a little bit more informally. 01:03:04.800 --> 01:03:07.600 Feel free to turn on your videos and chat a little bit. 01:03:07.600 --> 01:03:09.280 That would be great. And thanks, everyone, for coming. 01:03:09.280 --> 01:03:13.056 We’ll see you next week for Ian’s talk. 01:03:14.054 --> 01:03:15.840 - Thank you.