WEBVTT Kind: captions Language: en-US 00:00:01.120 --> 00:00:03.760 Hello, everyone. My name is Alicia Hotovec-Ellis, 00:00:03.760 --> 00:00:07.760 and I’m a research seismologist with the California Volcano Observatory. 00:00:07.760 --> 00:00:10.880 A few months ago, Sarah Minson asked me to give a talk on the physics 00:00:10.880 --> 00:00:14.560 of swarms, and I figured, sure. This will be a great excuse to 00:00:14.560 --> 00:00:18.240 refresh myself on the literature and showcase a little of my own work. 00:00:18.240 --> 00:00:20.880 Unfortunately, the format’s a bit short to do both with 00:00:20.880 --> 00:00:24.560 any amount of depth of nuance, so I’ll highlight the brief aspect 00:00:24.560 --> 00:00:28.296 of my talk’s title to keep in mind as we get rolling. 00:00:28.320 --> 00:00:31.360 Naturally, this being part review talk, I have to start with some 00:00:31.360 --> 00:00:33.920 quick definitions so we’re all on the same page. 00:00:33.920 --> 00:00:36.080 What is a swarm, and how are they different from 00:00:36.080 --> 00:00:40.240 main shock/aftershock sequences? Here I’ve two example magnitude 00:00:40.240 --> 00:00:44.000 time histories for two real earthquake sequences in Alaska. 00:00:44.000 --> 00:00:46.400 Although they are both spatiotemporally clustered, 00:00:46.400 --> 00:00:49.520 the behavior of the earthquakes with time is very different. 00:00:49.520 --> 00:00:52.800 Main shock/aftershock sequences have the largest event either first, 00:00:52.800 --> 00:00:56.000 or very near the start, in the event of a foreshock sequence. 00:00:56.000 --> 00:00:59.760 But swarms can have the largest event at any point in the sequence, 00:00:59.760 --> 00:01:02.240 with, perhaps, some preference to being in the middle. 00:01:02.240 --> 00:01:04.960 And that event tends not to be exceptionally larger than 00:01:04.960 --> 00:01:10.240 the next-largest event. This isn’t to say that swarm earthquakes can’t be large. 00:01:10.240 --> 00:01:13.920 Magnitude 5s, 6s, potentially even up to magnitude 8 earthquakes 00:01:13.920 --> 00:01:17.520 can be part of a swarm sequence. However, when there is a large 00:01:17.520 --> 00:01:20.960 regular main shock, it’s followed by aftershocks that themselves 00:01:20.960 --> 00:01:24.000 follow Omori’s law. Aftershocks become less frequent 00:01:24.000 --> 00:01:27.520 with time in a very predictable way. While I wouldn’t really call swarms 00:01:27.520 --> 00:01:31.600 truly random in time, they are certainly less predictable and often occur 00:01:31.600 --> 00:01:35.520 in bursts of events. Finally, there’s a difference in the number of small 00:01:35.520 --> 00:01:39.040 events compared to large events, usually referred to as the b value 00:01:39.040 --> 00:01:41.920 from Gutenberg-Richter. Swarms, on average, tend to have 00:01:41.920 --> 00:01:45.040 large b values, meaning that they are depleted in large events 00:01:45.040 --> 00:01:48.400 given the number of small events. Of course, these really should be 00:01:48.400 --> 00:01:53.736 ranges with some overlap and can be highly spatially heterogeneous. 00:01:53.760 --> 00:01:57.040 I’m here because swarms are among the primary modes of seismicity 00:01:57.040 --> 00:02:00.240 at volcanoes, across eruptive style and composition, 00:02:00.240 --> 00:02:02.480 and whether they’re currently erupting or not. 00:02:02.480 --> 00:02:06.640 True, there are some unique sources at volcanoes that don’t involve fault slip, 00:02:06.640 --> 00:02:10.480 but when faults do slip at volcanoes, they preferentially tend to do so 00:02:10.480 --> 00:02:14.080 in swarms. Same goes for hydrothermally and geothermally 00:02:14.080 --> 00:02:17.520 active areas, anthropogenically induced seismicity from both 00:02:17.520 --> 00:02:20.880 wastewater injection and hydro-fracturing, at the base of 00:02:20.880 --> 00:02:23.920 slipping glaciers, and in the form of low-frequency earthquakes 00:02:23.920 --> 00:02:28.400 and tremor during episodes of slow slip. Obviously, this isn’t a comprehensive 00:02:28.400 --> 00:02:33.256 list but gives a flavor of the diversity, but also some similarities, in setting. 00:02:33.280 --> 00:02:35.600 Similarities that allow us to derive a handful of 00:02:35.600 --> 00:02:37.440 common features between them. 00:02:37.440 --> 00:02:41.736 First, I’d like to highlight a few properties of the settings themselves. 00:02:41.760 --> 00:02:45.200 For one, swarms tend to happen in areas of high heat flux. 00:02:45.200 --> 00:02:48.800 For volcanoes in geothermal areas, this is due to the presence of magma. 00:02:48.800 --> 00:02:51.840 But, as you go deeper – for example, in a subduction zone – 00:02:51.840 --> 00:02:54.880 the hotter things get as well. The implication here is that 00:02:54.880 --> 00:02:58.400 the heat lowers the viscosity. And we transition from brittle to 00:02:58.400 --> 00:03:02.560 more ductile rheology and behavior. This is also where, in part, 00:03:02.560 --> 00:03:05.840 glaciers fit in, as ice can be ductile enough to flow 00:03:05.840 --> 00:03:10.000 yet still be brittle enough to fracture. Heat also means hot fluids, which 00:03:10.000 --> 00:03:13.920 we’ll get to talk more about in a bit. But those bring in dissolved minerals 00:03:13.920 --> 00:03:17.576 that can precipitate and alter the frictional properties of the faults. 00:03:17.600 --> 00:03:20.720 Heterogeneities in these properties may also be a factor. 00:03:20.720 --> 00:03:24.376 And, of course, having faults available to slip is important as well. 00:03:24.400 --> 00:03:28.480 I have host rock here to encompass other bulk properties of the rock that 00:03:28.480 --> 00:03:32.800 may be important, like permeability or porosity, in addition to their viscosity, 00:03:32.800 --> 00:03:37.120 but also as a nod to induced seismicity, where proximity to crystalline basement 00:03:37.120 --> 00:03:39.840 and the old faults there being a good predictor of whether 00:03:39.840 --> 00:03:43.656 unintended seismicity will be induced. 00:03:43.680 --> 00:03:47.600 Next, I’d like to stress – ha ha – the S in faults. 00:03:47.600 --> 00:03:51.440 As we continue to improve the tools available to better observe swarms, 00:03:51.440 --> 00:03:54.560 the more frequently we see that swarms involved multiple faults, 00:03:54.560 --> 00:03:58.320 sometimes with different orientations and slip directions, not unlike the 00:03:58.344 --> 00:04:03.656 interconnected fracture mesh that Dave Hill conceived of back in the ’70s. 00:04:03.680 --> 00:04:06.640 In his model, the fracture mesh accommodated mode I opening 00:04:06.640 --> 00:04:09.760 of dikes, but these cracks don’t have to be filled with magma. 00:04:09.760 --> 00:04:13.840 Which leads us to the next major player. You called it. Fluids. 00:04:13.840 --> 00:04:17.336 The geophysical go-to when you need a hand-waving explanation. 00:04:17.360 --> 00:04:20.960 In all seriousness, fluids of all kinds are ubiquitous in the subsurface. 00:04:20.960 --> 00:04:25.120 The Earth is not a dry place. Whether it be water, supercritical gas, 00:04:25.120 --> 00:04:29.040 or a multiphase combination, fluids are a great way to lower the effective stress 00:04:29.040 --> 00:04:33.336 on a fault and promote it to slip or to change the local stress state. 00:04:33.360 --> 00:04:37.040 One of the features of swarms linked to the intrusion of high-pressure fluids 00:04:37.040 --> 00:04:40.720 is that they migrate with time. In some cases, like this one, 00:04:40.720 --> 00:04:43.680 the rate of that movement is consistent with the diffusion of fluid 00:04:43.680 --> 00:04:47.520 through permeable host rock. Indeed, if we watch the propagation 00:04:47.520 --> 00:04:51.760 of these pulses of seismicity, it’s hard not to imagine fluids or 00:04:51.760 --> 00:04:55.200 something else moving up from depth, hopping from fault to fault. 00:04:55.200 --> 00:04:59.040 And, in many cases, the most logical culprit is hydrothermal fluids. 00:04:59.040 --> 00:05:02.960 Other times, such as for magmatic intrusions, seismicity follows closely 00:05:02.960 --> 00:05:06.240 the tip of a propagating dike, which certainly isn’t a diffusive 00:05:06.240 --> 00:05:12.000 process, nor are pulses of slow slip. These possibilities are all examples 00:05:12.000 --> 00:05:16.480 of a transient perturbation to the system – something, whether it 00:05:16.480 --> 00:05:19.800 be fluids, magma, slow slip, or a combination of any of them – 00:05:19.800 --> 00:05:23.040 alters the state of the system. It can be highly localized, 00:05:23.040 --> 00:05:25.920 such as a pulse of fluids traveling directly through a fault mesh, 00:05:25.920 --> 00:05:29.040 or creep along a fault. Or it can be the elastic response 00:05:29.040 --> 00:05:32.800 to a nearby intrusion at depth. In fact, the evacuation of mass can 00:05:32.800 --> 00:05:36.800 also trigger swarms, such as during the summit eruption at Kilauea in 2018 00:05:36.800 --> 00:05:39.736 and the partial draining of the magma chamber there. 00:05:39.760 --> 00:05:42.800 The perturbation doesn’t even need to be local. 00:05:42.800 --> 00:05:46.720 Areas prone to swarms and tremor are also often dynamically triggered by 00:05:46.720 --> 00:05:50.640 large remote earthquakes or are modulated by moon or Earth tides, 00:05:50.640 --> 00:05:53.280 suggesting that they are areas near failure already, 00:05:53.280 --> 00:05:57.576 requiring very little additional stimulus to become active. 00:05:57.600 --> 00:06:01.040 All this to say there are many factors that are likely shared between many 00:06:01.040 --> 00:06:05.520 swarms in general, however identifying the unique factors contributing to a 00:06:05.520 --> 00:06:10.000 specific swarm can still be a challenge. By way of example, I’ll briefly 00:06:10.000 --> 00:06:12.400 show you the highlights from a series of swarms at depth 00:06:12.400 --> 00:06:14.696 under Mammoth Mountain. 00:06:14.720 --> 00:06:17.360 Mammoth Mountain sits atop the southwest rim of Long Valley 00:06:17.360 --> 00:06:21.360 Caldera – home to skiing and some excitement in the late 1980s 00:06:21.360 --> 00:06:23.920 due to a vigorous shallow earthquake swarm thought to be 00:06:23.920 --> 00:06:26.536 due to the intrusion of a magmatic dike. 00:06:26.560 --> 00:06:29.760 Several years later, degassing of carbon dioxide killed patches 00:06:29.760 --> 00:06:33.280 of trees near Horseshoe Lake. The gas probably escaped because 00:06:33.280 --> 00:06:36.960 the intrusion of the aforementioned dike breached a cap on a reservoir 00:06:36.960 --> 00:06:42.000 of exsolved CO2. Flux of CO2 and other hydrothermal fluids are the likely 00:06:42.000 --> 00:06:45.520 cause of shallow swarms in the area. But Mammoth also has a decent amount 00:06:45.520 --> 00:06:50.240 of significantly deeper activity. Here I show earthquakes with time – 00:06:50.240 --> 00:06:52.880 gray dots for shallow and black dots – scaled to magnitude – 00:06:52.880 --> 00:06:56.400 for the deeper events. Don’t let the amount of ink here fool you. 00:06:56.400 --> 00:06:59.440 There are about 100 times more earthquakes above the dashed line 00:06:59.440 --> 00:07:02.080 than below it, but they aren’t the focus of this study. 00:07:02.080 --> 00:07:05.680 The dashed line represents the approximate brittle-ductile transition 00:07:05.680 --> 00:07:08.160 below which earthquakes less frequently occur. 00:07:08.160 --> 00:07:11.816 And, yet, we still have plenty here, just episodically. 00:07:11.840 --> 00:07:15.040 Another detail is that, at these depths, there are two different flavors 00:07:15.040 --> 00:07:18.960 of earthquakes. On the left, the classic brittle failure – what we call 00:07:18.960 --> 00:07:23.176 volcano tectonic earthquake that looks pretty much normal, though deep. 00:07:23.200 --> 00:07:26.160 Then, on the right, an earthquake in nearly the same location 00:07:26.160 --> 00:07:28.800 but with waveforms enriched in lower frequencies, 00:07:28.800 --> 00:07:32.696 called deep long period, or DLP, earthquakes. 00:07:32.720 --> 00:07:35.680 These are often thought to be due directly to the movement of magma 00:07:35.680 --> 00:07:39.496 at depth, but their exact mechanism is still unknown. 00:07:39.520 --> 00:07:43.440 Going back to the depth-time plot, I colored DLPs as open red circles, 00:07:43.440 --> 00:07:45.920 and we can begin to see some interesting patterns in the occurrence 00:07:45.920 --> 00:07:49.656 of these two types of earthquakes, especially if we zoom in. 00:07:49.680 --> 00:07:52.880 Certainly there are differences in the long-term behavior of the earthquakes, 00:07:52.880 --> 00:07:57.336 but the temporal proximity suggests a relationship between them. 00:07:57.360 --> 00:08:00.160 Using high-precision relocation techniques, we can also clean up 00:08:00.160 --> 00:08:03.040 the catalog and begin to see more clearly the spatial relationship 00:08:03.040 --> 00:08:06.160 between the two as well. On the left, I’ve got a cross-sectional 00:08:06.160 --> 00:08:10.080 view of the original locations with depth. Same color scheme as before. 00:08:10.080 --> 00:08:13.840 Middle is the updated relocations. And right is the map view. 00:08:13.840 --> 00:08:17.360 I’ll set this to rotate so that you can appreciate the shape of the volume. 00:08:17.360 --> 00:08:20.856 Basically, we ended up with a northeast-southwest-trending, 00:08:20.880 --> 00:08:23.600 near-vertical plane, where the DLPs that I was 00:08:23.600 --> 00:08:28.160 able to relocate preferentially tend to occur in the center between two arms 00:08:28.160 --> 00:08:31.656 surrounding an otherwise aseismic volume. 00:08:31.680 --> 00:08:34.480 Of course, there are also some interesting behaviors that pop out 00:08:34.480 --> 00:08:37.040 when we show the occurrence of these earthquakes with time. 00:08:37.040 --> 00:08:40.480 I’ve got a cross-sectional view here, which will march forward in 30-minute 00:08:40.505 --> 00:08:46.530 increments. DLPs are denoted as open circles. Otherwise, they’ll be filled. 00:08:47.200 --> 00:08:50.320 Earthquakes will fade from yellow to black with age, and we’ll skip 00:08:50.320 --> 00:08:52.720 any frames that don’t have any new offense to show. 00:08:52.720 --> 00:08:56.800 The date up at the top can skip several weeks at a time, but, for most of 00:08:56.800 --> 00:08:59.120 the swarms, we’ll follow without any breaks. 00:08:59.120 --> 00:09:03.576 And I will go through and narrate as it loops through the second time. 00:09:03.600 --> 00:09:07.440 All right. So the first major activity starts the deepest, then intermittently 00:09:07.440 --> 00:09:11.176 works its way to the – way up the southwest arm. 00:09:11.200 --> 00:09:15.360 Activity then moves deeper again and begins to fill out the bottom 00:09:15.360 --> 00:09:19.120 of the northeast arm in this slowly expanding and the largest of the 00:09:19.120 --> 00:09:23.040 swarms that I was able to relocate. Which then lights up in sort of 00:09:23.040 --> 00:09:26.880 these smaller swarms. A few months later, activity jumps 00:09:26.880 --> 00:09:30.960 to the top of the northeast arm and migrates down, then jumps over 00:09:30.960 --> 00:09:35.440 to the other arm and migrates both up and down simultaneously. 00:09:35.440 --> 00:09:38.240 It pops back over to the other arm briefly, and then most of 00:09:38.240 --> 00:09:41.280 the longer-lived DLP activity that just kind of flew by 00:09:41.280 --> 00:09:44.936 in this view, fills in the rest of the middle. 00:09:44.960 --> 00:09:47.920 Another curiosity with the swarms was the presence of earthquakes 00:09:47.920 --> 00:09:52.080 that I had called flipped polarities. Here, I have the waveforms for 00:09:52.080 --> 00:09:54.560 two earthquakes that occurred closely in both time – 00:09:54.560 --> 00:09:57.600 about 10 minutes apart – and space – of order of, perhaps, 00:09:57.600 --> 00:10:01.040 50 to 100 meters apart. But I flipped the waveform of the 00:10:01.040 --> 00:10:04.856 red earthquake upside-down to show how similar it is to the black one. 00:10:04.880 --> 00:10:07.920 This implies that the earthquakes occurred on faults of very similar 00:10:07.920 --> 00:10:11.016 orientation but with opposite senses of slip. 00:10:11.040 --> 00:10:14.400 While this observation is unusual, it’s certainly not unprecedented. 00:10:14.400 --> 00:10:17.280 For example, White and others observe a similar flipping of focal 00:10:17.280 --> 00:10:20.640 mechanisms during a dike intrusion in Iceland, which they interpreted 00:10:20.640 --> 00:10:25.040 as either fault slipping with opposite slip on either side of the dike tip, 00:10:25.040 --> 00:10:29.200 or as slip on either side of a plug of solidified magma within the dike 00:10:29.200 --> 00:10:31.840 and the country rock surrounding it. 00:10:31.840 --> 00:10:35.440 An alternative possibility is that the orientations of the two faults are similar 00:10:35.440 --> 00:10:39.120 but just ever so slightly different. And that their orientation relative to 00:10:39.120 --> 00:10:42.400 the regional stress field causes the difference in slip factor. 00:10:42.400 --> 00:10:46.000 In this example in Texas from Rutledge and others, they can see the slight 00:10:46.000 --> 00:10:49.360 difference in fault orientations from the relocated hypocenters. 00:10:49.360 --> 00:10:53.120 In our work, we’re looking at sort of a similar elongated system of faults 00:10:53.120 --> 00:10:57.120 but rotated 90 degrees and at 15 to 20 kilometers’ depth. 00:10:57.120 --> 00:11:00.080 So we don’t have the resolution to see these small deviations 00:11:00.080 --> 00:11:03.336 in orientation if they exist. 00:11:03.360 --> 00:11:06.080 Tying it all together, we have the earthquakes and, by proxy, 00:11:06.080 --> 00:11:09.680 the fault system they occur on situated in the mid- to lower crust with 00:11:09.680 --> 00:11:12.560 the Moho here at the bottom and the surface at the top, 00:11:12.560 --> 00:11:16.296 and then the usual brittle-ductile transition closer to the surface. 00:11:16.320 --> 00:11:20.720 Hot fluids, whether they be magma itself or exsolved supercritical CO2 00:11:20.720 --> 00:11:24.080 and/or water, are sourced from depth. And they work their way through the 00:11:24.080 --> 00:11:28.160 system in discrete bursts of activity. During these swarms, we see flipped 00:11:28.160 --> 00:11:31.120 mechanisms with two possible explanations. 00:11:31.120 --> 00:11:34.480 Either the response to the opening of a tip of an intruding dike 00:11:34.480 --> 00:11:38.296 or the opening of a fracture mesh with slightly different orientations. 00:11:38.320 --> 00:11:41.520 Both of these possibilities infer mode I opening of fractures 00:11:41.520 --> 00:11:44.936 to accommodate the intruded volume of fluids. 00:11:44.960 --> 00:11:48.160 Given the amount of bouncing around that the activity does, there is a pretty 00:11:48.160 --> 00:11:51.840 good possibility that there must be some amount of aseismic slip – 00:11:51.840 --> 00:11:56.000 sorry, aseismic flow going on, either through this central portion 00:11:56.000 --> 00:11:59.176 or up the arms in pulses that have already slipped. 00:11:59.200 --> 00:12:02.880 Speaking of the central portion, its aseismic nature and the presence of 00:12:02.880 --> 00:12:07.280 DLPs suggest that it likely has a different rheology than the outer arms, 00:12:07.280 --> 00:12:14.136 which we suspect may be due to the – sorry, due to an area of partial melt. 00:12:14.160 --> 00:12:17.760 Additionally, horizontal feature in the seismicity being distinct breaks 00:12:17.760 --> 00:12:21.360 in the start or endpoint of a swarm, suggest that there may also be sills 00:12:21.360 --> 00:12:24.456 or some other horizontal barriers to vertical flow. 00:12:24.480 --> 00:12:27.840 Magma occasionally intrudes more shallowly than this body, 00:12:27.840 --> 00:12:31.520 such as it likely did in 1989. Shallow swarms are then driven 00:12:31.520 --> 00:12:33.840 by volatiles exsolved from the magma at depth, 00:12:33.840 --> 00:12:36.536 which then finally escaped the surface. 00:12:36.560 --> 00:12:39.520 One of the primary conclusions from this work is that the speed at which 00:12:39.520 --> 00:12:43.200 these swarms migrated highlighted the permeability of this structure, 00:12:43.200 --> 00:12:47.040 which acted like something of a geologically high-speed chimney, 00:12:47.040 --> 00:12:49.600 connecting the root of the volcanic system to the surface 00:12:49.600 --> 00:12:53.176 with implications for how quickly fluids can ascend. 00:12:53.200 --> 00:12:56.880 That said, this ties back to my point from prior to this section. 00:12:56.880 --> 00:13:00.240 Even with a great set of observations for this set of swarms, 00:13:00.240 --> 00:13:03.760 we still aren’t able to definitively pin the blame on these – for these 00:13:03.760 --> 00:13:08.160 swarms on intrusions of new basaltic magma or if they were other 00:13:08.160 --> 00:13:11.200 magmatically derived fluids. Though, to be fair, much of 00:13:11.200 --> 00:13:13.760 the challenge with this work was that everything was so deep. 00:13:13.760 --> 00:13:16.960 Shallower swarms certainly have their own challenges as well, though. 00:13:16.960 --> 00:13:19.040 And, with that, I’d like to thank you for your attention, 00:13:19.040 --> 00:13:21.360 and I look forward to your questions.