WEBVTT Kind: captions Language: en-US 00:00:01.600 --> 00:00:05.120 Hello. I’m Gareth Funning, and I’m going to present my results 00:00:05.120 --> 00:00:09.200 on the Maacama Fault from InSAR and repeating earthquakes 00:00:09.200 --> 00:00:13.692 as a means of understanding creep on that structure. 00:00:15.440 --> 00:00:20.056 So the questions I hope to answer today are a couple of things. 00:00:20.080 --> 00:00:24.080 Firstly, what can InSAR tell us about surface creep? 00:00:24.080 --> 00:00:28.160 InSAR is very good at measuring surface deformation, and so it can give 00:00:28.160 --> 00:00:33.576 us a comprehensive picture of what’s going on in the shallowest crust. 00:00:33.600 --> 00:00:36.080 And, number two, what can repeating earthquakes 00:00:36.080 --> 00:00:38.464 tell us about creep at depth? 00:00:40.000 --> 00:00:42.320 There are very useful supplementary information 00:00:42.320 --> 00:00:45.120 that we can get from earthquakes that can tell us about the 00:00:45.120 --> 00:00:48.971 creep behavior a little deeper beneath the surface. 00:00:50.000 --> 00:00:54.000 Just as a – as a little reminder, InSAR is a method that makes use 00:00:54.000 --> 00:00:58.376 of radar on satellites. The phase of the radar, 00:00:58.400 --> 00:01:01.920 which we measure periodically as the satellite flies over, 00:01:01.920 --> 00:01:05.336 is related to the distance between the satellite and the ground. 00:01:05.360 --> 00:01:10.160 If the ground moves, we can detect a phase shift, which is related to the 00:01:10.160 --> 00:01:14.160 distance that the ground has moved, and we measure that distance towards 00:01:14.160 --> 00:01:17.040 or away from the satellite. So, in this particular case, the ground 00:01:17.040 --> 00:01:21.399 has moved away, and there’s a positive phase shift in this case. 00:01:22.560 --> 00:01:26.800 We can makes maps like this of velocity covering the entire regions – 00:01:26.800 --> 00:01:32.056 the entire Bay Area and North Bay from an older data set. 00:01:32.080 --> 00:01:34.776 And you can see all kinds of things in here. 00:01:34.800 --> 00:01:37.280 Really, today, though, the only thing I really am interested 00:01:37.280 --> 00:01:40.800 in showing you is creep. And if you zoom in on the Hayward 00:01:40.800 --> 00:01:44.080 and Concord Faults, you can see these abrupt changes in surface 00:01:44.080 --> 00:01:48.320 deformation and velocity localized on fault structures indicate that those 00:01:48.320 --> 00:01:54.240 faults are moving at the surface. And, from the offset in the line-of-sight 00:01:54.240 --> 00:01:58.548 direction, we can estimate what the fault-parallel creep rate is. 00:01:59.520 --> 00:02:03.280 For this study, we’re going to use data from the Sentinel mission. 00:02:03.280 --> 00:02:09.096 Sentinel is a – is a monitoring mission that’s planned for 20 years 00:02:09.120 --> 00:02:13.440 of regular SAR acquisitions. In California, we get an acquisition 00:02:13.440 --> 00:02:16.400 every 12 days as a result of this mission, which is way better 00:02:16.400 --> 00:02:19.976 than any previous mission that we’ve had access to. 00:02:20.000 --> 00:02:22.400 And we have formed lots and lots of interferograms. 00:02:22.400 --> 00:02:27.760 Each green line on this chart, which spans the time – the timespan 00:02:27.760 --> 00:02:33.440 of the study that I’m going to show you that was conducted by my graduate 00:02:33.440 --> 00:02:37.520 student – my former graduate student Jerlyn Swiatlowski in collaboration 00:02:37.520 --> 00:02:41.440 with David Bekaert at JPL. Each of these interferograms has 00:02:41.440 --> 00:02:46.720 been combined into a time series, and we’ve solved for the best-fitting 00:02:46.720 --> 00:02:52.880 velocity from all of this data. And it’s a fantastically rich data set. 00:02:52.880 --> 00:02:56.216 And what does it look like? It looks like this. 00:02:56.240 --> 00:03:00.960 So here we have two different data sets from two different geometries. 00:03:00.960 --> 00:03:03.840 On the left, we have the so-called descending data set. 00:03:03.840 --> 00:03:09.096 This is where the satellite is flying from north to south and looking to the west. 00:03:09.120 --> 00:03:11.416 On the right is the ascending data set. 00:03:11.440 --> 00:03:15.527 Satellite, in this case, is flying to the north and looking to the east. 00:03:16.400 --> 00:03:21.256 There are all sorts of features in here that are of interest. 00:03:21.280 --> 00:03:26.000 There are some obvious big blobs, which I should explain what they are. 00:03:26.000 --> 00:03:31.040 This dark blue thing that you see in both data sets is the 00:03:31.040 --> 00:03:35.840 Geysers geothermal field. That’s the – one of the most prominent 00:03:35.840 --> 00:03:41.440 deformation signals I see in my data. It’s subsiding. That’s what blue colors 00:03:41.440 --> 00:03:46.080 mean in this – in this data set, is moving away from the satellite. 00:03:46.080 --> 00:03:50.057 And we can see, because it’s moving away from a satellite that’s pointing – 00:03:50.057 --> 00:03:52.800 there are two different geometries that are pointing in different horizontal 00:03:52.800 --> 00:03:57.680 directions, what they have in common is they’re both moving away 00:03:57.680 --> 00:04:01.029 from the satellite. This means the ground is moving down. 00:04:02.080 --> 00:04:08.000 But, ignoring that, there are features that are much more similar to 00:04:08.000 --> 00:04:10.640 the Hayward Fault that you can pick out in the data. 00:04:10.640 --> 00:04:15.840 So, for example, in the northwest, there’s a section of the Maacama Fault, 00:04:15.840 --> 00:04:20.000 and you can see it here, where there is an abrupt transition 00:04:20.000 --> 00:04:23.440 in velocity across the fault. In general, you have cooler colors 00:04:23.440 --> 00:04:28.640 on the west side and warmer colors on the east side, either from green to – 00:04:28.640 --> 00:04:33.016 or yellow to red, or blue to green. 00:04:33.040 --> 00:04:35.416 This is consistent with creep. 00:04:35.440 --> 00:04:38.240 There are a couple of other places where you might pick that out as well. 00:04:38.240 --> 00:04:41.336 Another is around here. 00:04:41.360 --> 00:04:44.880 This is Cloverdale – CL. It’s just southeast of there. 00:04:44.880 --> 00:04:48.880 You can see there’s an abrupt transition in the velocity localized 00:04:48.880 --> 00:04:51.840 on a fault structure there. And actually, you can see the 00:04:51.840 --> 00:04:56.160 reverse sense of color offset in the ascending data, 00:04:56.160 --> 00:05:01.336 which is consistent also with horizontal movement of that fault. 00:05:01.360 --> 00:05:03.976 So there are some other things. 00:05:04.000 --> 00:05:09.840 One feature that remains to be explained is the color offset 00:05:09.840 --> 00:05:11.416 that you see here. 00:05:11.440 --> 00:05:16.960 In some sense, having a warm color to a cooler color from west to east is 00:05:16.960 --> 00:05:22.720 consistent with the viewing geometry and there being an amount of creep. 00:05:22.720 --> 00:05:26.696 But it also suggests that there’s likely to be vertical motion 00:05:26.720 --> 00:05:30.240 of the ground on the east – on the west side of this fault structure. 00:05:30.240 --> 00:05:32.720 And we’re not sure that that would be tectonic. 00:05:32.720 --> 00:05:35.680 That could – that could be aquifer-related. 00:05:35.680 --> 00:05:38.536 That’s something we’re looking into. 00:05:38.560 --> 00:05:43.840 Anyway, we analyzed these velocity offsets by taking profiles through 00:05:43.840 --> 00:05:47.680 them and fitting lines to them. So this is a procedure that was 00:05:47.680 --> 00:05:52.240 documented in an earlier paper where we explain how we 00:05:52.240 --> 00:05:54.960 select windows of data on either side of the fault. 00:05:54.960 --> 00:05:58.698 We blank out the data in the very near field of the fault zone. 00:05:59.440 --> 00:06:04.080 And that’s because there’s usually some complexity there. 00:06:04.080 --> 00:06:06.720 And we also simplify the fault trace somewhat. 00:06:06.720 --> 00:06:12.856 So we don’t have an exact location of the – of the fault in our profiles. 00:06:12.880 --> 00:06:18.376 Anyway, we take this window data, and we fit a single – a line with 00:06:18.400 --> 00:06:22.056 a common gradient but different offsets on either side. 00:06:22.080 --> 00:06:25.680 The offset is related to the line-of-sight velocity offset 00:06:25.680 --> 00:06:29.680 of the fault at the surface. We actually use that gradient 00:06:29.680 --> 00:06:34.320 to de-trend the profiles when we plot them, for neatness’ sake. 00:06:34.320 --> 00:06:38.080 And what you should see, at least in the descending data, 00:06:38.080 --> 00:06:44.376 is a flat profile on one side of the fault with a step in it. 00:06:44.400 --> 00:06:47.760 That shows that we have estimated what that step is 00:06:47.760 --> 00:06:52.080 and what that gradient is. So here is Jerlyn’s profiles. 00:06:52.080 --> 00:06:55.600 We’re going to focus on a subset of them in and around Cloverdale 00:06:55.600 --> 00:06:59.256 because these are the ones that showed the largest offsets. 00:06:59.280 --> 00:07:02.480 In the descending, you can see, by and large, most of these profiles 00:07:02.480 --> 00:07:05.423 show a step to the right – a step up to the right. 00:07:05.423 --> 00:07:09.760 And you can see here a number of them do that. 00:07:09.760 --> 00:07:12.720 And you can also see, actually, in the ascending data, which is 00:07:12.720 --> 00:07:18.080 looking the other way, the horizontal sense of offset should be reversed 00:07:18.080 --> 00:07:23.200 in that case, and you actually see steps down in the profiles on the right, 00:07:23.200 --> 00:07:28.943 which is consistent with fault-parallel right-lateral motion. 00:07:30.080 --> 00:07:35.840 We can convert these offset rates into line-of-sight – into fault-parallel rates 00:07:35.840 --> 00:07:40.696 using the fault geometry and the acquisition geometry of the radar. 00:07:40.720 --> 00:07:43.600 And what do we see? Well, plotted at the top here 00:07:43.600 --> 00:07:48.605 are the creep rates – the fault-parallel surface offset rates. 00:07:49.440 --> 00:07:52.720 Some portions of some of the fault systems – I’ve already talked about 00:07:52.720 --> 00:07:58.160 the Rodgers Creek Fault, but that shows consistent but low creep rates 00:07:58.160 --> 00:08:00.827 on the order of 1 to 2 millimeters a year. 00:08:02.080 --> 00:08:07.760 Northern Maacama Fault around Ukiah and further north, 00:08:07.760 --> 00:08:11.976 that shows rates in the range of 4 to 6 millimeters a year. 00:08:12.000 --> 00:08:17.680 And the area southeast of Cloverdale, highlighted before, has slip – 00:08:17.680 --> 00:08:20.480 creep rates of the order of 7 to 11 millimeters per year. 00:08:20.480 --> 00:08:24.480 And one intriguing feature is that they – the highest creep rates – 00:08:24.480 --> 00:08:28.480 the ones around 10 or a little bit higher – are very close in proximity 00:08:28.480 --> 00:08:32.320 to the Geysers geothermal field, and it’s making me wonder, 00:08:32.320 --> 00:08:36.560 and it’s something we’re looking into, whether there is a causal relationship 00:08:36.560 --> 00:08:41.910 between the presence of the Geysers field and the high creep rates. 00:08:43.360 --> 00:08:47.680 Just a little aside, then in the second part of the talk, 00:08:47.680 --> 00:08:49.840 to talk about repeating earthquakes. 00:08:49.840 --> 00:08:51.918 These are seismic evidence for creep. 00:08:51.918 --> 00:08:55.496 A little – or, at least they’re taken to be so. 00:08:55.520 --> 00:08:59.040 These are little asperities, we think, that have been dragged along 00:08:59.040 --> 00:09:06.540 by aseismic slip and continually reloaded to failure. They periodically fail. 00:09:07.360 --> 00:09:10.856 And they fail in the same places with the same mechanisms, 00:09:10.880 --> 00:09:14.456 and they produce identical waveforms. 00:09:14.480 --> 00:09:17.440 Here’s an example of a set of identical waveforms from 00:09:17.440 --> 00:09:21.200 a repeating earthquake source. This is all part of Nader Shakibay 00:09:21.200 --> 00:09:26.420 Senobari’s Ph.D. that he completed a couple of years ago at UCR. 00:09:27.520 --> 00:09:32.480 And the process of identifying repeating earthquakes is the process 00:09:32.480 --> 00:09:35.440 of taking earthquake waveforms and cross-correlating them and 00:09:35.440 --> 00:09:40.400 finding out how similar they are. Repeating earthquakes that we – 00:09:40.400 --> 00:09:44.880 that we show in our – in our publications generally 00:09:44.880 --> 00:09:48.800 have cross-correlation coefficients of 0.95 or higher. 00:09:48.800 --> 00:09:51.544 They are very, very similar. 00:09:53.200 --> 00:09:58.936 This is a data-mining process. We look at lots of data. 00:09:58.960 --> 00:10:03.840 There were 350 stations recording 75,000 earthquakes in the – 00:10:03.840 --> 00:10:09.016 in the North Bay – the extended North Bay that we show here. 00:10:09.040 --> 00:10:13.200 670,000 waveforms. We didn’t compare all of them 00:10:13.200 --> 00:10:18.536 to all of them, but we did run a lot of cross-correlations. 00:10:18.560 --> 00:10:23.200 And part of the issue is that the network density is heterogeneous. 00:10:23.200 --> 00:10:26.320 There’s a lot of dense station coverage in the south. 00:10:26.320 --> 00:10:29.976 There’s much less station coverage in the north. 00:10:30.000 --> 00:10:33.649 Stations are up for different amounts of time. 00:10:34.800 --> 00:10:40.800 And a strategy that is often used for sort of detecting repeating earthquakes 00:10:40.800 --> 00:10:46.800 is just to look at a single station and look at all the events that are recorded on it. 00:10:46.800 --> 00:10:50.800 And, when you have a network that’s so varied like this, it’s quite 00:10:50.800 --> 00:10:54.616 difficult to do that. So we had to come up with something else. 00:10:54.640 --> 00:11:00.160 Our strategy is actually to look at pair-wise comparisons of earthquakes. 00:11:00.160 --> 00:11:04.720 For example, these three colored events here – the red number 1, 00:11:04.720 --> 00:11:07.760 magenta number 2, and blue number 3 are, 00:11:07.760 --> 00:11:10.376 we think, a repeating earthquake family. 00:11:10.400 --> 00:11:14.776 But they were not all recorded on common stations. 00:11:14.800 --> 00:11:20.456 A different sets of earthquakes recorded 1 and 2 as recorded 2 and 3. 00:11:20.480 --> 00:11:24.720 But, since the cross-correlations in each case were very high, we were able to 00:11:24.720 --> 00:11:29.120 assemble them into a single cluster – a single repeating earthquake family. 00:11:29.120 --> 00:11:32.880 In fact, only – of all of the waveforms shown here – and you can see that the 00:11:32.880 --> 00:11:36.320 red waveforms, the blue waveforms, and the – and the magenta waveforms 00:11:36.320 --> 00:11:40.320 are all very similar to each other in pairs, there was only one station, 00:11:40.320 --> 00:11:45.088 NMC, that recorded all three of these events. 00:11:46.080 --> 00:11:48.080 But they’re very, very highly cross-correlated. 00:11:48.080 --> 00:11:53.336 If you looked on a cross-correlation dendrogram, these three events 00:11:53.360 --> 00:11:57.520 had a similarity of 0.95 or higher. And the next-nearest most similar 00:11:57.520 --> 00:12:03.261 event has a cross-correlation of 0.65, which is much lower. 00:12:04.320 --> 00:12:08.000 In the end, we found 59 repeating earthquake families. 00:12:08.000 --> 00:12:12.216 These are actually confirmed by doing relative relocations. 00:12:12.240 --> 00:12:15.760 We weren’t able to do that for all of our highly cross-correlated events 00:12:15.760 --> 00:12:20.720 because of station azimuthal coverage. The ones that are very highly – 00:12:20.720 --> 00:12:23.840 have very high cross-correlations but we were not able to confirm 00:12:23.840 --> 00:12:27.814 are shown as triangles, and there’s quite a lot of those as well. 00:12:29.040 --> 00:12:33.280 You’ll see that, along much of the Maacama Fault, there are repeating 00:12:33.280 --> 00:12:37.520 earthquakes. For some of these, we’re able to estimate creep rates 00:12:37.520 --> 00:12:40.240 based on comparisons to shallow InSAR offsets. 00:12:40.240 --> 00:12:44.320 And you can see that creep rates are somewhere between 2 – which are 00:12:44.320 --> 00:12:48.891 the green colors – and 5 millimeters a year on the Maacama Fault. 00:12:50.240 --> 00:12:52.960 Looking at the fault surface, you can see that most of these 00:12:52.960 --> 00:12:57.440 repeaters follow a streak in seismicity that starts off 00:12:57.440 --> 00:13:00.444 deep in the north and shallows to the south. 00:13:01.440 --> 00:13:05.040 It seems that that may be related to creep behavior because a lot of 00:13:05.040 --> 00:13:09.520 repeaters are found within it. Looking in detail at some of the 00:13:09.520 --> 00:13:15.840 features of the structures that we can identify from the repeaters around 00:13:15.840 --> 00:13:19.040 Cloverdale – there are a couple of places where it looks like there 00:13:19.040 --> 00:13:22.616 are dipping structures picked out in the repeaters. 00:13:22.640 --> 00:13:26.880 We haven’t really dug deeply into what these might represent, 00:13:26.880 --> 00:13:31.520 but it seems that there are parallel or sub-parallel dipping trends that 00:13:31.520 --> 00:13:35.120 have some of these repeating sources on them, which might then, 00:13:35.120 --> 00:13:39.256 therefore, be creeping. Something to look into. 00:13:39.280 --> 00:13:44.320 Further north near Willits, where we – where there was the excitement in the 00:13:44.320 --> 00:13:48.880 fall last year, we know that there’s surface creep evidence there, 00:13:48.880 --> 00:13:54.480 and of course we see it in the InSAR. But also you can see dipping trends 00:13:54.480 --> 00:13:58.640 in the repeaters in these places too. You can see, for example, in A2-B2, 00:13:58.640 --> 00:14:04.136 this fault’s dipping steeply to the east, and then A3-B3, something similar. 00:14:04.160 --> 00:14:09.360 Curiously, some of the repeaters seem to form some kind of vertical structure 00:14:09.360 --> 00:14:15.120 which surfaces a few kilometers – 5 kilometers to the east of the 00:14:15.120 --> 00:14:17.967 surface trace of the Maacama Fault. 00:14:18.880 --> 00:14:23.680 Most intriguingly, there may be some kind of Quaternary fault scarp 00:14:23.680 --> 00:14:28.456 5 kilometers to the east of the Maacama Fault near Willits. 00:14:28.480 --> 00:14:31.760 I don’t have time to talk about that, but it’s an intriguing thought that 00:14:31.760 --> 00:14:36.616 perhaps this has repeating earthquakes and might be creeping at depth. 00:14:36.640 --> 00:14:41.600 I realize I’ve taken all of my time, so I’m just going to say you can pause 00:14:41.600 --> 00:14:46.716 the video and read my conclusions. Thank you very much for listening. 00:14:48.789 --> 00:14:54.141 [silence]