WEBVTT Kind: captions Language: en-US 00:00:03.340 --> 00:00:05.040 Hi, everyone. Good morning. 00:00:05.040 --> 00:00:09.360 Welcome to the Earthquake Science Center seminar. 00:00:09.360 --> 00:00:17.180 Today our seminar speaker is Xie Hu. Just before I give a brief bio for Xie, just 00:00:17.180 --> 00:00:21.500 a reminder that there’s no seminar next week. And the week after, 00:00:21.500 --> 00:00:27.060 on the 28th of November, is Gary Fuis, so put that in your calendars. 00:00:27.060 --> 00:00:34.760 So just a quick bio about Xie. Xie kindly accepted giving this seminar, I think two 00:00:34.760 --> 00:00:39.960 weeks ago. So thank you so much for stepping in. We really appreciate it. 00:00:39.960 --> 00:00:44.380 Xie is from Wuhan, China, originally. 00:00:44.380 --> 00:00:48.540 Her bachelor’s degree is from China University of Geoscience. 00:00:48.540 --> 00:00:52.750 Her master’s degree is from Wuhan University in remote sensing. 00:00:52.750 --> 00:00:57.120 And she just finished her Ph.D. in May 2018. 00:00:57.120 --> 00:01:02.460 So she’s a newly minted Ph.D., and she’s now a postdoc at Berkeley. 00:01:02.460 --> 00:01:05.280 And she got her Ph.D. at Southern Methodist University 00:01:05.280 --> 00:01:11.140 in Dallas, Texas, and so she’s a new resident here of the Bay Area. 00:01:11.140 --> 00:01:15.500 So welcome, Xie, and please begin. Thank you. 00:01:19.040 --> 00:01:21.620 - Thank you, Sara, for introduction. 00:01:23.180 --> 00:01:27.840 So today, I’m going to talk about hydrodynamics of landsliding 00:01:27.840 --> 00:01:32.580 and aquifer systems revealed by satellite radar interferometry. 00:01:32.580 --> 00:01:37.990 Basically, it’s about my Ph.D. work. And before I go – before I walk you 00:01:37.990 --> 00:01:42.570 through this, I would like to thank my Ph.D. adviser, Zhong Lu, 00:01:42.570 --> 00:01:47.540 and also my collaborators, Thomas Pierson, Dave George, 00:01:47.540 --> 00:01:53.210 Rebecca Kramer, Teng Wang, Sylvain Barbot, Thomas Oommen, 00:01:53.210 --> 00:01:57.220 Alexander Handwerger, and Roland Bürgmann. 00:01:59.070 --> 00:02:04.240 So first I will give you a brief overview of the geodetic tool that I’m using, 00:02:04.240 --> 00:02:08.670 the synthetic aperture radar. It’s transmitting electromagnetic waves 00:02:08.670 --> 00:02:13.090 to the ground target and then receiving the backscatter. 00:02:13.090 --> 00:02:18.030 The SAR signals are composed of the amplitude and phase information, 00:02:18.030 --> 00:02:23.280 in which the amplitude represents the textures and the moisture of the target, 00:02:23.280 --> 00:02:26.740 and the phase indicates the distance it travels. 00:02:26.740 --> 00:02:31.150 When we have – repeats that image and do the interferometry, we can measure 00:02:31.150 --> 00:02:37.080 the elevation or the displacement along the oblique line of sight. 00:02:38.080 --> 00:02:41.300 And the distance between these two satellites 00:02:41.300 --> 00:02:45.060 when the image is same target is called the baseline. 00:02:45.060 --> 00:02:50.870 The derived interferograms is wrapped, so we need to – generally need to 00:02:50.870 --> 00:02:55.680 unwrap it to get the sensible view of the displacement field. 00:02:55.680 --> 00:02:59.230 And this system, for me, is like a remotely sensed ruler 00:02:59.230 --> 00:03:04.020 and can capture the displacement at millimeter accuracy. 00:03:05.320 --> 00:03:09.980 The generally used spaceborne SAR data are working at three wavelengths – 00:03:09.980 --> 00:03:16.700 X-band, C-band, and L-band in the increased wavelengths. 00:03:16.700 --> 00:03:21.260 And they are comparable to the scale range of the rulers. 00:03:21.260 --> 00:03:24.909 X-band has the shortest wavelengths, so it’s more sensible to the 00:03:24.909 --> 00:03:28.769 small displacement. While the L-band is more 00:03:28.769 --> 00:03:35.120 capable to capture the – to resolve the large displacement gradient. 00:03:37.360 --> 00:03:41.680 Beyond that, different wavelength data have different capability to penetrate 00:03:41.680 --> 00:03:47.239 the canopies. In terms of that, the L-band is more preferable 00:03:47.240 --> 00:03:53.180 to monitoring the displacement in forests or agricultural areas. 00:03:55.840 --> 00:03:59.120 For the information management, so we need to simulate the 00:03:59.120 --> 00:04:03.459 topographic phase component using the digital elevation model 00:04:03.459 --> 00:04:06.459 and get it removed from the interferograms. 00:04:06.459 --> 00:04:09.959 So ideally, the consequent interferograms represents the 00:04:09.959 --> 00:04:16.110 information exclusively, but it’s usually contaminated by the artifacts from the 00:04:16.110 --> 00:04:22.440 topographic error in the DEM, the atmospheric delay or troposphere 00:04:22.440 --> 00:04:27.780 and ionosphere, and also orbital errors and other noise. 00:04:29.760 --> 00:04:34.000 But the good thing is that those signals, they have different characteristics 00:04:34.000 --> 00:04:37.920 in the spatial and temporal domain, and we can use them to separate them out. 00:04:37.920 --> 00:04:42.699 For example, the atmospheric phase screen and orbital error are strongly 00:04:42.700 --> 00:04:47.640 correlated in the spatial domain and uncorrelated in the temporal domain. 00:04:47.640 --> 00:04:54.120 While the DEM error has a mathematic relationship with the spatial baseline. 00:04:54.120 --> 00:04:59.860 So when we have a lot of acquisitions, we can generate many interferograms, 00:04:59.860 --> 00:05:05.060 then we can solve them out through the scheme of inversion. 00:05:05.060 --> 00:05:08.949 But I have to say that it not always work that well. 00:05:08.949 --> 00:05:12.930 So the challenges really vary case by case. 00:05:12.930 --> 00:05:16.020 But for today’s talk, I won’t dive into details. 00:05:16.020 --> 00:05:19.750 But now you should have a flavor on how InSAR work and how we 00:05:19.750 --> 00:05:23.760 get the tensile displacement and also velocity field. 00:05:24.740 --> 00:05:27.460 So now let’s move to landslide. 00:05:28.180 --> 00:05:33.960 The first landslide that I’m going to talk about is Cascade landslide complex. 00:05:33.960 --> 00:05:39.240 It’s located along the Columbia River on the Washington side. 00:05:39.240 --> 00:05:45.000 In the first phase of this study, I used two overlapping L-bands – 00:05:45.000 --> 00:05:47.840 ALOS-1 tracks of data. 00:05:48.360 --> 00:05:54.440 This landslide always attracts the geoscientists with the legends – 00:05:54.440 --> 00:05:59.240 the Bridge of the Gods. In early 15th century, 00:05:59.250 --> 00:06:02.669 the ancient Bonneville Landslide have dammed the Columbia River 00:06:02.669 --> 00:06:05.810 and form a bridge so that the Native Americans 00:06:05.810 --> 00:06:08.960 can walk across the river by feet. 00:06:08.960 --> 00:06:13.980 It was later breached by the flood, but we can still see the remnants 00:06:13.990 --> 00:06:21.360 at the toe area and also these three small islands, which is Cascade Rapids. 00:06:21.360 --> 00:06:26.030 The monitoring the stability of this landslide is important due to its potential 00:06:26.030 --> 00:06:30.280 threats to the local residents at North Bonneville and 00:06:30.280 --> 00:06:35.720 Stevenson as well as the potential damage to the pipelines, 00:06:35.720 --> 00:06:40.260 roads, railways go through the shoreline as well as 00:06:40.260 --> 00:06:44.780 the Bonneville Dam, now situated on the Cascade Rapids. 00:06:47.840 --> 00:06:54.620 The most recent failure event is Greenleaf Rock – Greenleaf Basin Rock 00:06:54.629 --> 00:07:00.720 Avalanche. It occurred on January 3rd, 2008, at the head scarp. 00:07:00.720 --> 00:07:06.920 The phase information is not useful here because it’s totally decorrelation. 00:07:07.760 --> 00:07:15.020 So instead, we extract amplitude at the targets with large dispersion. 00:07:15.020 --> 00:07:18.340 From the time-series amplitude, we see there’s a sudden drop 00:07:18.349 --> 00:07:23.830 by 15 dB in the end of 2007, suggesting that the fractures 00:07:23.830 --> 00:07:29.320 may have initiated one month prior to the reported date. 00:07:30.900 --> 00:07:35.140 Here’s the result of time-series displacement using these 00:07:35.150 --> 00:07:38.380 two tracks of ALOS-1 data sets. 00:07:38.380 --> 00:07:44.330 Both results agree that only a part of about a 4-square-meter area 00:07:44.330 --> 00:07:48.110 was in motion, and we name is Crescent Lake Landslide. 00:07:48.110 --> 00:07:54.070 And the cumulative landslide displacement from 2007 to 2011 is 00:07:54.070 --> 00:07:59.900 about 30 centimeters. And the motions mainly occurred in the wintertime. 00:08:01.560 --> 00:08:07.740 When use SAR data to image the slope terrain, the slope may facing to or facing 00:08:07.740 --> 00:08:13.560 away from the oncoming radar passes. But for both cases, the downslope 00:08:13.560 --> 00:08:19.620 slip vector S is always larger than that of the radar look vector L. 00:08:19.620 --> 00:08:23.069 And the amplification factor can be calculated using the 00:08:23.069 --> 00:08:28.360 geometries of the radar-looking vector and the slope topography. 00:08:28.360 --> 00:08:32.800 After we apply the amplification factor on our original line-of-sight 00:08:32.810 --> 00:08:36.660 displacement, we can derive the slope-parallel displacement. 00:08:36.660 --> 00:08:42.080 So we see not only the magnitude of the motion has been amplified, 00:08:42.080 --> 00:08:44.380 the spatial pattern also changed. 00:08:44.380 --> 00:08:46.660 There’s another area concentrated with 00:08:46.660 --> 00:08:50.780 increased motion at the west lobe of the landslide. 00:08:50.780 --> 00:08:55.260 Next, I’m going to show you the time series slope-parallel displacements at 00:08:55.260 --> 00:09:02.040 location P on the landslide with the off- slide GPS observation across the river. 00:09:03.580 --> 00:09:08.450 I removed the linear trend and get the seasonal waveform, 00:09:08.450 --> 00:09:12.150 which is comparable to that of the precipitation. 00:09:12.150 --> 00:09:17.250 It seems that the landslide start to slip when the 30-day precipitation 00:09:17.250 --> 00:09:23.820 total exceeds 300 millimeter. And the drought year, 2010, 00:09:23.820 --> 00:09:28.600 witnessed the most variable and the least amount of precipitation. 00:09:28.600 --> 00:09:30.350 And correspondingly, the landslide motion 00:09:30.350 --> 00:09:34.440 started and stopped it repeatedly that year. 00:09:35.260 --> 00:09:39.880 Assuming that the off-slide GPS represents the region motion 00:09:39.890 --> 00:09:45.410 due to hydrologic loading, I converted into the slope-parallel 00:09:45.410 --> 00:09:49.030 direction using the same geometry at location P 00:09:49.030 --> 00:09:51.380 and also removed the linear trends. 00:09:51.380 --> 00:09:57.310 It turns out that the seasonal waveform is similar to that of the InSAR result. 00:09:57.310 --> 00:10:02.960 However, the magnitude of GPS is much less than 00:10:02.960 --> 00:10:06.420 what’s expressed in the InSAR result. 00:10:06.420 --> 00:10:10.560 This is because the GPS is next to the river, and there is 00:10:10.560 --> 00:10:16.760 very little variation in the water storage. However, the landslide body has a much 00:10:16.760 --> 00:10:23.640 thicker on saturated storm, and thus magnified hydrological loading effect. 00:10:25.800 --> 00:10:28.940 Later, we obtained more spaceborne SAR data. 00:10:28.940 --> 00:10:32.760 So now we have two timeframes. 00:10:32.760 --> 00:10:39.160 And we add another timeframe from 2014 to 2016. 00:10:41.820 --> 00:10:47.000 We also obtained the on-slide GPS data from USGS Cascades Volcano 00:10:47.000 --> 00:10:54.480 Observatory, and the data has been collected since the late 2014. 00:10:54.480 --> 00:10:58.320 And you can see that the InSAR result for our Sentinel-1 data set has a 00:10:58.320 --> 00:11:01.270 perfect agreement with that of the GPS. 00:11:01.270 --> 00:11:05.960 And there’s an interesting signal leading the wet season, which is 00:11:05.960 --> 00:11:12.540 followed by and in contrast to the later more drastic downslope slip. 00:11:12.540 --> 00:11:17.380 And there is a question mark here. So we want to [inaudible] that means, 00:11:17.380 --> 00:11:20.770 if this suggests an [inaudible] of motion before that. 00:11:20.770 --> 00:11:24.890 To better analyze this, we projected the GPS data 00:11:24.890 --> 00:11:28.860 into the slope fit caught in the system. 00:11:28.860 --> 00:11:33.190 So u here is the slope aspect that shows the direction with 00:11:33.190 --> 00:11:40.390 the largest topographic gradient. And w is outwardly perpendicular to u. 00:11:40.390 --> 00:11:46.360 And v is normal to u and contained in the slope-parallel plane. 00:11:47.160 --> 00:11:52.620 After we do the conversion, we find that that precursory motion is actually mainly 00:11:52.620 --> 00:11:58.110 due to the slope-normal subsidence. Afterwards, when the pore pressure 00:11:58.110 --> 00:12:03.900 elevated at the basal surface in a timeframe of less then one month, 00:12:03.900 --> 00:12:07.300 the more drastic downslope slip occurred. 00:12:07.300 --> 00:12:12.380 And accompanied by some lateral propagation. 00:12:12.380 --> 00:12:17.200 At that moment, both u and v components surged. 00:12:19.380 --> 00:12:24.940 So why InSAR can capture this two-phase progressive motion? 00:12:24.940 --> 00:12:30.150 This is because the descending line of sight is looking nearly straight 00:12:30.150 --> 00:12:35.740 at the landslide motion at this particular location of GPS. 00:12:35.740 --> 00:12:40.670 In addition, the slope-normal subsidence [inaudible] to moving away 00:12:40.670 --> 00:12:44.350 from the satellite, while the downslope slip [inaudible] 00:12:44.350 --> 00:12:49.270 to move in toward the satellite. And this contrast makes pronounced 00:12:49.270 --> 00:12:53.200 differences in the line-of-sight perspective. 00:12:53.200 --> 00:12:57.790 In terms of the slope-normal subsidence, there are two possible explanations. 00:12:57.790 --> 00:13:02.120 It can be due to the mass loading by the rainfall or the contraction 00:13:02.120 --> 00:13:06.810 of the loose soil upon shearing. However, here, where the 00:13:06.810 --> 00:13:12.240 subsidence was observed is about – the depth is about 100 meters 00:13:12.240 --> 00:13:15.420 and thus subjected to high normal stress. 00:13:15.420 --> 00:13:21.800 In addition, the soil material is chemically altered and clay-rich. 00:13:21.800 --> 00:13:27.720 And where the slip has been observed intermittently for decades. 00:13:27.730 --> 00:13:33.560 Given these constraints, the stress and the porosity should have reduced 00:13:33.560 --> 00:13:39.020 to a residual level in which the further contraction is highly unlikely. 00:13:42.280 --> 00:13:48.440 We also summarized the rainfall conditions at those inflection points 00:13:48.440 --> 00:13:53.510 [inaudible] slope-normal subsidence occurred was about cumulative of 00:13:53.510 --> 00:13:59.820 140 millimeters of antecedent rainfall collected since September. 00:13:59.820 --> 00:14:06.690 Afterwards, was a cumulative about 270 millimeters of rainfall collected in 00:14:06.690 --> 00:14:13.120 less than one-month frame, the more drastic downslope slip occurred. 00:14:14.820 --> 00:14:18.580 Now we focus on the reactivated part of the landslide 00:14:18.580 --> 00:14:22.220 marked by this array of circles. 00:14:22.940 --> 00:14:27.880 Here’s the result of the landslide displacement velocity 00:14:27.890 --> 00:14:31.430 for each data track. The first row shows the 00:14:31.430 --> 00:14:35.250 ascending result, and the second row shows the descending result. 00:14:35.250 --> 00:14:38.950 So the area of interest is moving into the opposite direction 00:14:38.950 --> 00:14:43.140 with respect to the location of the sensors. 00:14:45.380 --> 00:14:50.260 One limitation of the polar-orbiting spaceborne InSAR data is that 00:14:50.260 --> 00:14:55.240 the landslide displacement is not sensitive to the north-south motion. 00:14:55.250 --> 00:14:57.580 Because you see the landslide is looking nearly to the – 00:14:57.580 --> 00:15:00.960 either to the east or to the west. 00:15:01.750 --> 00:15:04.140 So strictly speaking, it’s difficult for us. 00:15:04.140 --> 00:15:07.530 We only have two independent measurements, and – which are 00:15:07.530 --> 00:15:12.420 insufficient for us to retrieve the three-dimensional displacement. 00:15:12.420 --> 00:15:17.060 So we proposed a solution by utilizing the topography. 00:15:17.060 --> 00:15:21.170 Now let’s come back to this slope fit coordinate system. 00:15:21.170 --> 00:15:28.800 Here we assume that the displacement along direction v is negligible. 00:15:28.800 --> 00:15:32.760 And, in this way, we add a constraint on the horizontal plane, and we can 00:15:32.760 --> 00:15:36.900 solve for the [inaudible] three-dimensional displacement. 00:15:36.900 --> 00:15:39.080 Here’s the result. 00:15:39.080 --> 00:15:41.260 The color means the vertical displacement. 00:15:41.260 --> 00:15:46.220 So you see the upper-to-central lobe shows some subsidence and 00:15:46.220 --> 00:15:51.440 some localized [inaudible]. At the toe shows slight uplift. 00:15:51.440 --> 00:15:55.660 And the arrows means horizontal displacement. 00:15:55.670 --> 00:16:01.440 And the direction of the arrows is superimposed by the topography. 00:16:01.440 --> 00:16:05.600 And the size of the arrows means the magnitude of horizontal velocity. 00:16:05.600 --> 00:16:10.270 The maximum horizontal velocity is at the place 00:16:10.270 --> 00:16:13.920 where we see the maximum subsidence. 00:16:14.510 --> 00:16:22.200 We then use the 3D displacement as the input to invert for the landslide 00:16:22.200 --> 00:16:27.050 thickness based on the given equation of mass conservation. 00:16:27.050 --> 00:16:31.070 And this equation states that the mass flux divergency 00:16:31.070 --> 00:16:34.620 is balanced by the rate of thickness change. 00:16:34.620 --> 00:16:38.120 We used the rheological parameter f to approximate 00:16:38.120 --> 00:16:41.780 the depth average velocity using the surface velocity – 00:16:41.780 --> 00:16:45.470 f ranges between zero and 1 and depends on the thickness 00:16:45.470 --> 00:16:50.120 of the lower yield zone and the overlying plug region. 00:16:52.370 --> 00:16:58.140 Here is the result by applying different f. Larger f [inaudible] smaller thickness. 00:16:58.140 --> 00:17:02.600 We see here the thick zone terminates abruptly against 00:17:02.600 --> 00:17:08.320 the subsurface escarpment [inaudible] by the cross-hatched zone. 00:17:08.320 --> 00:17:12.560 And the derived basal surface is [inaudible]. 00:17:12.560 --> 00:17:19.610 We also draw a profile along the dashed line in which the slope aspect 00:17:19.610 --> 00:17:26.130 along this profile is fairly uniform. We see that the landslide depth – 00:17:26.130 --> 00:17:30.070 the dependence of the landslide depth on the rheological 00:17:30.070 --> 00:17:37.020 parameter f is more pronounced at the downslope part of this profile. 00:17:40.270 --> 00:17:46.400 In terms of the uncertainty of this study is mainly coming from the angles 00:17:46.400 --> 00:17:52.350 of the slope and aspect, and we allow them to vary by 3 degrees. 00:17:52.350 --> 00:17:58.100 And the consequent landslide, the thickness and volume will vary by 8%. 00:17:58.100 --> 00:18:02.140 However, the spatial pattern won’t change much. 00:18:04.140 --> 00:18:09.500 We also estimate the landslide mobility, which is given by the ratio between 00:18:09.500 --> 00:18:15.600 the runout distance and the elevation from the head to the toe. 00:18:15.600 --> 00:18:20.170 Given the upper boundary relationship between the landslide mobility and the 00:18:20.170 --> 00:18:26.820 landslide volume, the maximum runout distance can reach 7,000 meters. 00:18:26.820 --> 00:18:31.640 This suggests that a highly mobile runout at this landslide 00:18:31.640 --> 00:18:36.260 could potentially, again, block the Columbia River. 00:18:38.110 --> 00:18:44.160 So we apply the similar methods on the Monroe Landslide in northern California. 00:18:44.160 --> 00:18:49.240 The tension cracks at the toe were first detected in mid-2002. 00:18:49.240 --> 00:18:55.390 Subsequently, a first failure occurred in that December after the rainstorm and 00:18:55.390 --> 00:19:02.420 flooding and destroyed a county road. And two years later, after the 2004 Sims 00:19:02.420 --> 00:19:09.990 fire, which is north to the active lobe, a second failure struck the same place. 00:19:09.990 --> 00:19:16.630 So in this study, from InSAR results, we noticed that the active motion is – 00:19:16.630 --> 00:19:25.430 the area of active motion is far larger than just the failure block downslope. 00:19:25.430 --> 00:19:28.870 Here is the three-dimensional displacement result. 00:19:28.870 --> 00:19:33.640 The largest velocity and ground subsidence are near the longitudinal 00:19:33.640 --> 00:19:37.840 center at the [inaudible] element of 3. 00:19:37.850 --> 00:19:42.760 And there, there is a slightly over-steepened facet. 00:19:42.760 --> 00:19:48.440 Here you can see the slope angle. It’s increased suddenly at this block. 00:19:48.440 --> 00:19:51.840 And downslope – immediately downslope of it, 00:19:51.840 --> 00:19:54.960 there is some surface uplift. 00:19:56.820 --> 00:20:02.220 From the result of landslide thickness inversion, we see that there’s a 00:20:02.220 --> 00:20:08.420 thickening of the landslide near the transition from the surface subsidence 00:20:08.420 --> 00:20:14.940 to surface uplift. Around here, there’s a thickening of the landslide body. 00:20:15.790 --> 00:20:21.600 We also applied a one-dimensional pore pressure diffusion model to 00:20:21.600 --> 00:20:27.340 investigate the interaction between the rainfall and landslide velocity. 00:20:27.340 --> 00:20:31.640 And the surface boundary condition is that the transient 00:20:31.640 --> 00:20:35.690 pore pressure at the surface is modulated by the rainfall and 00:20:35.690 --> 00:20:40.300 calibrated by our infiltration scaling factor. 00:20:40.300 --> 00:20:46.930 And the best-fit result suggests that, in this landslide, the hydraulic diffusivity 00:20:46.930 --> 00:20:52.700 is about 5 to 8 times 10 to the negative 5 square meter per second. 00:20:54.520 --> 00:20:59.120 So, to conclude, SAR amplitude is useful to monitoring the tension cracks 00:20:59.130 --> 00:21:03.830 and avalanche in landslide areas, especially if there is no vegetation. 00:21:03.830 --> 00:21:09.060 Also, InSAR and GPS results reveal rainfall-triggered seasonal 00:21:09.060 --> 00:21:13.240 motion as well as subsidence prior to the downslope slip. 00:21:13.240 --> 00:21:17.120 We also show how we can derive the three-dimensional displacement 00:21:17.120 --> 00:21:22.480 from InSAR and use it to further invert for the basal geometries. 00:21:24.460 --> 00:21:29.020 So now let’s come to the next part – the aquifers – 00:21:29.020 --> 00:21:33.200 the natural containers of the groundwater in the Earth. 00:21:35.680 --> 00:21:39.790 The focus of my study is at Salt Lake Valley in Utah. 00:21:39.790 --> 00:21:45.600 It’s an alluvial basin bounded by the mountain ranges and the Great Salt Lake. 00:21:45.610 --> 00:21:51.110 It accommodates the rigorous aquifer systems and can be 00:21:51.110 --> 00:21:54.200 classified into the water discharge area, 00:21:54.200 --> 00:21:58.240 the primary recharge area, and secondary recharge area. 00:21:59.160 --> 00:22:06.450 The primary – the water recharge area generally at the mountain front where 00:22:06.450 --> 00:22:11.800 the hydraulic gradient is downward. And the water discharge area 00:22:11.800 --> 00:22:16.260 are generally the lower – at the lower elevated terrain 00:22:16.260 --> 00:22:21.370 at the north of this valley and also the river bank of the 00:22:21.370 --> 00:22:26.200 Jordan River that trisects the central access of the valley. 00:22:26.200 --> 00:22:28.650 There, the hydraulic gradient is upward, 00:22:28.650 --> 00:22:34.260 and groundwater will flow into the Jordan River or the Great Salt Lake. 00:22:35.080 --> 00:22:40.170 The second – the secondary recharge area is situated between where the 00:22:40.170 --> 00:22:44.770 shallow and deeper aquifers are not clearly distinguished. 00:22:44.770 --> 00:22:50.310 And also, this Salt Lake Valley also accommodates the very famous 00:22:50.310 --> 00:22:55.290 Wasatch Fault Zone at the mountain front of Wasatch Range 00:22:55.290 --> 00:23:01.080 and also the West Valley Fault Zone in the site of this valley. 00:23:02.220 --> 00:23:09.830 For this study, I used two SAR data sets – the Envisat and the Sentinel-1. 00:23:09.830 --> 00:23:14.380 I have two timeframes – covers almost one decade. 00:23:16.180 --> 00:23:21.020 Here’s the result. The first figure shows the long-term velocity 00:23:21.020 --> 00:23:25.380 during 2004 to 2010 is highlighting an area of 00:23:25.380 --> 00:23:30.770 net uplift by 15 millimeter per year. 00:23:30.770 --> 00:23:34.280 And the second figure shows the seasonal amplitude 00:23:34.280 --> 00:23:39.300 that’s corresponding to one summer season in 2015. 00:23:40.080 --> 00:23:46.220 And we also applied some statistic methods to distinguish the targets that’s 00:23:46.230 --> 00:23:51.770 shown with seasonal displacement, and here they are marked in the red dots. 00:23:51.770 --> 00:23:59.280 So both three – all those three figures highlight the same area that’s contained 00:23:59.280 --> 00:24:05.580 in the water discharge part of the aquifer, and it’s well-bounded by the faults. 00:24:05.580 --> 00:24:11.860 In addition to this, from the discontinuity in the displacement, we find 00:24:11.860 --> 00:24:18.120 another new fault segment, which is immediately west to the Salt Lake City. 00:24:19.980 --> 00:24:26.980 In order to explain the net uplift at this location, we collected 00:24:26.980 --> 00:24:32.460 the water level change maps. So all these maps are corresponding to 30 00:24:32.460 --> 00:24:40.490 years’ timeframe, and they are collected since 1970 with five years intervals. 00:24:40.490 --> 00:24:46.460 So we see the dotted pattern means water level increase, and the outer 00:24:46.460 --> 00:24:50.420 shadows at different [inaudible] means water level decline. 00:24:50.420 --> 00:24:56.000 And basically, the principle aquifer experienced a water level decline. 00:24:56.720 --> 00:25:01.140 For the – so, for the area of the net uplift, 00:25:01.140 --> 00:25:09.530 it has the best overlap with the water level increase during 1975 to 2005 00:25:09.530 --> 00:25:14.800 and 1980 to 2010, rather than the years before or after that. 00:25:14.800 --> 00:25:19.061 So it suggested that the ground uplift is 00:25:19.061 --> 00:25:24.720 probably due to the prolonged water recharge decades ago. 00:25:26.320 --> 00:25:32.460 We also performed analytical modeling of the cuboid vertical shear zone. 00:25:33.380 --> 00:25:37.360 We first applied a single cuboid and applied a thickness 00:25:37.370 --> 00:25:43.450 as 500 and 600 meter. And the strain rate is constrained 00:25:43.450 --> 00:25:51.510 to be 2 times 10 to the negative 5 and 1.7 times 10 to the negative 5. 00:25:51.510 --> 00:25:55.890 And we then applied four cuboids and assumed the homogeneous 00:25:55.890 --> 00:26:02.270 subsurface with a constant strain rate, and the result shows that the 00:26:02.270 --> 00:26:08.180 west side has a much thicker shearing zone than the east side. 00:26:09.420 --> 00:26:14.340 Another simpler approximation of the subsurface volume changes come 00:26:14.340 --> 00:26:21.040 from the assumption of the point-source dilatation in elastic half-space. 00:26:21.040 --> 00:26:25.350 And this – the result is similar to that of the analytical modeling. 00:26:25.350 --> 00:26:32.920 The annual volume change is larger than 5.4 times 10 to the 5 cubic meter. 00:26:36.000 --> 00:26:40.440 Since I’ve shown that we suspect the ground deformation is 00:26:40.450 --> 00:26:48.400 due to the water process a long time ago, so we used a decay coefficient to 00:26:48.400 --> 00:26:53.980 describe the delayed poroelastic response to head change and can be characterized 00:26:53.980 --> 00:26:59.920 by modeling the vertical displacement as an exponential function of time. 00:26:59.920 --> 00:27:05.990 So here, k is the decay coefficient, and it ranges from minus 1 to zero. 00:27:05.990 --> 00:27:14.120 And it tell us how fast the displacement approach to be leveling off. 00:27:15.220 --> 00:27:20.280 Here is the result. The first figure shows the decay coefficient. 00:27:20.290 --> 00:27:23.110 The second figure shows the magnitude coefficient. 00:27:23.110 --> 00:27:29.809 And the third one shows the residual between the model and the observations. 00:27:29.809 --> 00:27:35.840 So from these figures, the first two, apart from the previously mapped 00:27:35.840 --> 00:27:42.100 one fault segment, we are able to identify another fault segment, 00:27:42.100 --> 00:27:47.550 which is orthogonal to the previous one. And this too have to make a better 00:27:47.550 --> 00:27:54.350 connection between the Wasatch Fault Zone and the West Valley Fault Zone. 00:27:54.350 --> 00:27:57.920 In turn, it suggests that the faults may disturb 00:27:57.920 --> 00:28:02.740 the groundwater flow and partition the hydraulic units. 00:28:03.360 --> 00:28:08.500 When we focus on this decay coefficient map, it has some 00:28:08.500 --> 00:28:14.090 patchy distribution there, but just to focus on the area with the seasonal 00:28:14.090 --> 00:28:19.930 features, marked by these yellow dotted lines inside this block, 00:28:19.930 --> 00:28:26.550 we see the east part of the fault has a much smaller decay coefficient. 00:28:26.550 --> 00:28:32.360 And it suggests that there’s a faster response to a given head change. 00:28:35.920 --> 00:28:41.120 Now we focus on the time series deformation at the area of the net uplift. 00:28:41.120 --> 00:28:47.000 We remove the linear trend and get the second flow figure. 00:28:47.000 --> 00:28:52.150 And we find the seasonal waveform is situated in between the water 00:28:52.150 --> 00:28:56.620 discharge data and the precipitation, which we consider the main 00:28:56.620 --> 00:29:00.960 source of water recharge. So it suggested that the 00:29:00.960 --> 00:29:05.720 ground displacement is modulated simultaneously 00:29:05.720 --> 00:29:10.320 by the water discharge and recharge process. 00:29:10.320 --> 00:29:13.900 Another important parameter is the groundwater system 00:29:13.910 --> 00:29:19.220 is the storage coefficient. So it describes the amount of the 00:29:19.220 --> 00:29:25.610 water drained from the system with the per-unit decline the water had. 00:29:25.610 --> 00:29:32.420 And here in this study, we focused on the elastic part of the storage coefficient. 00:29:32.420 --> 00:29:34.510 And it can be modeled by correlating the 00:29:34.510 --> 00:29:38.920 detrended water level and the displacement. 00:29:39.660 --> 00:29:44.680 The skeletal specific storage coefficient is given by the ratio between the storage 00:29:44.740 --> 00:29:50.580 coefficient and the aquifer thickness. And another representation of 00:29:50.580 --> 00:29:56.420 that is related with the bulk aquifer compressibility. 00:29:57.040 --> 00:30:01.660 In this valley, we have four water-level gauges that have 00:30:01.670 --> 00:30:08.100 frequently enough sampling for us to correlate with the displacement. 00:30:08.100 --> 00:30:13.190 So here is the result. The first figure is the interpolation result 00:30:13.190 --> 00:30:19.750 from the four gauges by correlating the displacement and water level. 00:30:19.750 --> 00:30:23.420 And thickness is from the isopach map of the 00:30:23.420 --> 00:30:27.140 unconsolidated and semi-consolidated layers. 00:30:27.140 --> 00:30:31.000 Then we can derive the specific storage coefficient. 00:30:31.000 --> 00:30:34.600 And the porosity is digitized from the written literature. 00:30:34.600 --> 00:30:40.480 And finally, we can obtain the bulk aquifer compressibilities. 00:30:40.480 --> 00:30:44.940 And all those derive the values in the reasonable range, 00:30:44.940 --> 00:30:49.540 considering the database of other aquifers around the world. 00:30:51.200 --> 00:30:56.520 During this study, we also identified some localized subsidence signals. 00:30:56.520 --> 00:31:01.610 The first one is in North Salt Lake. So you can see that the displacement 00:31:01.610 --> 00:31:09.840 is cross-linear at a rate about 2 centimeter per year during 2004 to 2011. 00:31:09.840 --> 00:31:14.080 And it is since accelerated to 3 centimeter per year in more 00:31:14.080 --> 00:31:18.660 recent data sets from 2015. And there’s no correlation 00:31:18.660 --> 00:31:22.740 with the precipitation. Instead, we find a group of 00:31:22.740 --> 00:31:28.600 round-top buildings there, probably related to industry or production. 00:31:29.440 --> 00:31:34.360 And the second localized subsidence is in Lehi. 00:31:34.360 --> 00:31:36.300 And they are, like, the first one – 00:31:36.300 --> 00:31:43.620 the subsidence only visible in the data sets from 2015. 00:31:44.600 --> 00:31:50.220 And we collected real images over this location, and it’s interesting to 00:31:50.220 --> 00:31:58.240 see that there’s no cracks in the 2005. But afterwards, since 2010, 00:31:58.240 --> 00:32:04.280 the cracks just show up suddenly and just keep growing to now. 00:32:07.340 --> 00:32:12.900 We used the temporal features of the displacement to characterize 00:32:12.900 --> 00:32:16.660 the displacements due to different source. 00:32:16.660 --> 00:32:22.220 For example, the net uplifting hydrological units show with seasonality, 00:32:22.230 --> 00:32:25.890 and the larger residual of the exponential decay model. 00:32:25.890 --> 00:32:32.340 The large residual is come from the wildly varied displacement waveform. 00:32:32.340 --> 00:32:34.620 Where the other hydrological units are shown 00:32:34.630 --> 00:32:38.180 without seasonality and small residuals. 00:32:38.180 --> 00:32:41.160 On the other hand, for the industrial fields, 00:32:41.160 --> 00:32:44.880 there is no seasonality, but there is large residuals. 00:32:44.890 --> 00:32:48.620 And that is coming from either the cross-linear or 00:32:48.620 --> 00:32:52.220 even accelerated trend of the displacement. 00:32:54.340 --> 00:33:01.200 Another very drastic subsidence signal we observed during this study is 00:33:01.200 --> 00:33:06.220 at the tailings impoundment in the vicinity of the Great Salt Lake. 00:33:06.220 --> 00:33:09.360 The eastern flank of the Oquirrh Mountains accommodate the 00:33:09.360 --> 00:33:12.740 Bingham Canyon Mine, which contributes to 00:33:12.740 --> 00:33:16.680 a quarter of the copper to the United States. 00:33:16.680 --> 00:33:23.220 20 kilometer north of it, the company built the tailings impoundment 00:33:23.230 --> 00:33:31.190 to contain its own economic waste. And it has been in operation since 1906. 00:33:31.190 --> 00:33:35.180 One big concern about this facility is due to 00:33:35.180 --> 00:33:38.360 the potential earthquake-induced failure. 00:33:38.360 --> 00:33:43.260 There are at least three failure – happened in the history. 00:33:43.260 --> 00:33:48.049 And you can also notice that there is a cluster of micro-earthquakes 00:33:48.049 --> 00:33:52.179 concentrated beside the impoundment. And this impoundment – 00:33:52.180 --> 00:34:01.060 this impoundment also bounded by highway of I-80 and 201. 00:34:01.060 --> 00:34:07.580 Around the year 2000, the south pond was abandoned and reclaimed by 00:34:07.590 --> 00:34:13.600 vegetation at the top surface. And the active deposition has been 00:34:13.600 --> 00:34:17.920 transferred from the south pond to the north pond here. 00:34:19.420 --> 00:34:26.060 For this study, I add another track of data from L-band ALOS-1. 00:34:26.070 --> 00:34:31.300 Here is the result. So all those three data sets pinpoint 00:34:31.300 --> 00:34:36.640 the same location of the compaction peak at the northeast corner. 00:34:36.640 --> 00:34:43.000 And it also shows the deceleration in the subsidence. 00:34:43.010 --> 00:34:47.510 The compaction peak – the compaction peak rate 00:34:47.510 --> 00:34:50.980 has been decreased from 20 centimeter to year 00:34:50.980 --> 00:34:58.140 during 2004 to 2011, to 10 centimeter per year during 2015-2016. 00:35:00.160 --> 00:35:07.810 We then used geotechnical modeling to detail the – 00:35:07.810 --> 00:35:11.990 differentiate the settlement process. 00:35:11.990 --> 00:35:16.270 From the written literature, the south pond is stratified into five layers, 00:35:16.270 --> 00:35:21.990 and we consider the top 6 meters spigotted tailings as the load. 00:35:21.990 --> 00:35:27.619 And we collect a series of properties for each layers. 00:35:27.619 --> 00:35:33.460 Geotechnically, the total sediment is from three stages. 00:35:33.460 --> 00:35:38.420 The immediate settlement, which occurs instantly after the load is applied. 00:35:38.420 --> 00:35:42.420 And then, it’s the primary consolidation occurred due to the 00:35:42.420 --> 00:35:45.500 gradual dissipation of the pore pressure. 00:35:45.500 --> 00:35:49.530 And then lastly, the secondary consolidation may also occurred 00:35:49.530 --> 00:35:55.340 at a constant effective stress due to the re-arrangement of the particles. 00:35:55.340 --> 00:36:01.480 Through this forward model, we can characterize the displacement 00:36:01.480 --> 00:36:06.990 at each location during the time. And it’s worth mentioning that there 00:36:06.990 --> 00:36:12.040 is [inaudible] drains installed at northeast corner to enhance 00:36:12.040 --> 00:36:15.520 the vertical drainage between these layers. 00:36:16.520 --> 00:36:22.660 Our InSAR results can be perfectly matched and were explained 00:36:22.670 --> 00:36:27.830 by the model. And also, it helped to constrain this model. 00:36:27.830 --> 00:36:35.200 And this model can give implication about the development of the 00:36:35.200 --> 00:36:40.460 settlement process in the near future. The settlement rate in 2020 is 00:36:40.460 --> 00:36:45.160 expected to be less than half of the amount one decade ago. 00:36:46.870 --> 00:36:54.360 By narrowing down the scale range of the color bar, we can identify other 00:36:54.370 --> 00:37:03.050 localized signals, such as there’s highway segments now is subsiding. 00:37:03.050 --> 00:37:11.910 And also, there is a sedimentation pond, which is just east to the south tailings 00:37:11.910 --> 00:37:17.950 impoundment, which is also – there is also normal subsidence there. 00:37:17.950 --> 00:37:21.870 And there’s water level gauges here, but the water level data 00:37:21.870 --> 00:37:27.320 is no relationship with those localized subsidence. 00:37:28.540 --> 00:37:33.060 And to conclude, we image the spatiotemporal deformation 00:37:33.060 --> 00:37:36.800 over the Salt Lake City – Salt Lake Valley, and we modeled 00:37:36.800 --> 00:37:39.680 the net uplifting groundwater reservoir, 00:37:39.680 --> 00:37:42.970 constraining the geometry and the strain rate. 00:37:42.970 --> 00:37:47.530 We also derived the hydrological properties, mapped new fault segments, 00:37:47.530 --> 00:37:51.550 and characterized the deformation due to different sources. 00:37:51.550 --> 00:37:57.270 And finally, I would like to thank these – all those institutions for providing us 00:37:57.270 --> 00:38:01.100 the data sets and the funding. Thank you. 00:38:01.100 --> 00:38:06.780 [Applause] 00:38:08.120 --> 00:38:10.120 [Silence] 00:38:10.780 --> 00:38:13.700 - Thank you so much. Anybody have any questions? 00:38:13.700 --> 00:38:16.220 [audio feedback] Whoa. 00:38:17.920 --> 00:38:24.200 [Silence] 00:38:25.020 --> 00:38:28.980 - That was a really nice talk. You’ve done a lot of work on this. 00:38:28.980 --> 00:38:35.480 I got a question about the newly identified faults in the Salt Lake area. 00:38:36.220 --> 00:38:42.240 Have you found any geologic evidence of these – of these new faults since then? 00:38:42.240 --> 00:38:48.380 And do you have any sort of idea of – based on the recurrence, 00:38:48.380 --> 00:38:55.200 or based on the GPS rates and the InSAR rates, any idea on recurrence and sort of 00:38:55.200 --> 00:38:58.640 the magnitude of events that they could – these could produce? 00:39:01.320 --> 00:39:08.400 - So for the – for the newly mapped faults, we – 00:39:08.400 --> 00:39:14.520 purely from the displacement map. And also, we derived the 00:39:14.520 --> 00:39:18.170 hydrological properties from that map, we derived this. 00:39:18.170 --> 00:39:23.650 We don’t have the geological evidence on that, but from – just from here, you 00:39:23.650 --> 00:39:30.880 can see it match also pretty well with the existent already-mapped fault segments. 00:39:30.880 --> 00:39:39.020 And in terms of the – we actually did some planar work – 00:39:39.020 --> 00:39:43.240 I’m going to show here – about how the groundwater reservoir 00:39:43.240 --> 00:39:50.900 may affect the Wasatch Fault Zone. Here are some preliminary results. 00:39:50.900 --> 00:39:55.220 You can see that the – we calculated the Coulomb stress rate 00:39:55.220 --> 00:39:58.660 at the Salt Lake City segment of the Wasatch Fault Zone. 00:39:58.660 --> 00:40:02.470 The stress rate is not uniform, so the impact on the earthquake 00:40:02.470 --> 00:40:09.000 nucleation is very complex issue here. It’s a part of the ongoing job. 00:40:09.000 --> 00:40:11.080 Thank you. 00:40:15.360 --> 00:40:18.480 - Anybody else have any questions? 00:40:21.160 --> 00:40:22.400 - Hi. - Yeah. 00:40:22.400 --> 00:40:25.440 - I have a question about your work on northern California 00:40:25.440 --> 00:40:27.440 on the landslide there. - Yes. 00:40:27.440 --> 00:40:32.390 - So when you introduced the landslide, you were saying that there’s a lot of 00:40:32.390 --> 00:40:35.640 tension cracks that were observed. Is it – was it in the 00:40:35.640 --> 00:40:38.060 upper head scarp area? - No. 00:40:38.060 --> 00:40:42.390 - Throughout the body? - Yeah. For this one, 00:40:42.390 --> 00:40:47.980 the first tension cracks detected is in mid-2002 at the toe area. 00:40:47.980 --> 00:40:49.520 - Okay. - Here is the toe. 00:40:49.520 --> 00:40:54.720 - I was just curious because, when you were doing your pore pressure modeling, 00:40:54.720 --> 00:40:59.700 you were using a 1D diffusion model, so I was just curious if you think that’s 00:40:59.700 --> 00:41:03.800 totally appropriate given the fact that there’s large cracks and how that might 00:41:03.800 --> 00:41:07.130 affect the pore pressures at the landslide base, and if there’s a way you could 00:41:07.130 --> 00:41:09.950 potentially test what may be in those [inaudible]. 00:41:09.950 --> 00:41:14.080 - Yes. At this moment, we just use very simple 1D diffusion model 00:41:14.080 --> 00:41:19.680 because we are in lack of a lot of knowledge about this landslide. Yeah. 00:41:22.780 --> 00:41:25.440 - Any more questions? 00:41:28.160 --> 00:41:31.280 - So to make you jump back to Utah – sorry. 00:41:31.280 --> 00:41:39.480 When you map hydrogeologic properties, how deep do you think that represents? 00:41:39.480 --> 00:41:44.040 Or what depth, you know, does that – do those properties represent? 00:41:44.040 --> 00:41:51.100 - Yes. It’s 5 to 6 meter – 5 to – 500 to 600 meter, yeah. 00:41:51.109 --> 00:41:59.849 - Okay. So the newly mapped faults, is that – do you think those are 00:41:59.849 --> 00:42:03.050 crustal-scale faults? Or are they just shallow features 00:42:03.050 --> 00:42:09.721 that are sort of controlling shallow groundwater flow and that kind of thing? 00:42:09.721 --> 00:42:16.730 - For this part, I couldn’t give you a answer now yet in terms of 00:42:16.730 --> 00:42:20.540 the depth of the faults. Yeah. - Thanks. It’s really interesting 00:42:20.540 --> 00:42:24.000 data that you’re showing, though. - Thank you. 00:42:27.020 --> 00:42:28.760 - Did you have a question, or were you putting … 00:42:28.760 --> 00:42:33.660 - No. - [laughs] All right. Any more questions? 00:42:35.000 --> 00:42:40.760 If not, if any of you would like to join us for lunch with the speaker or would like 00:42:40.760 --> 00:42:46.970 to get on her schedule for the afternoon, please come up and see myself and Sara, 00:42:46.970 --> 00:42:50.830 or the schedule – the schedule is online, and you can – you can get to that 00:42:50.830 --> 00:42:53.200 from the announcement email that we sent out. 00:42:53.200 --> 00:42:55.980 All right. Please join me in thanking our speaker again. 00:42:55.980 --> 00:43:01.840 [Applause] 00:43:04.640 --> 00:43:10.500 [Silence]