Session: InSight Seismology on Mars: Results From the First (Earth) Year of Data and Prospects for the Futur

Type: Oral

Date: 4/28/2020

Time: 08:45 AM

Room: 115

Description: InSight Seismology on Mars: Results From the First (Earth) Year of Data and Prospects for the Futur

InSight is the first planetary mission with a seismometer package, SEIS, since the Apollo Lunar Surface Experiments Package. It is delivering 6 axis 20 sps continuous seismic data, sent back to Earth and now made open to the community every 3 months. More than 350 events have been detected by the end of 2019 but less than 10 have amplitudes significantly above the SEIS instrument requirement below 1 Hz. A few have clear and coherent arrivals of P and S waves, enabling location, diffusion/attenuation characterization and receiver function analysis. The event’s magnitudes are likely ≤ 4 and no clear surface waves nor deep interior phases have been identified by end of 2019. This suggests deep events, possibly with scattering along their propagation paths and with propagation properties putting Mars between the Earth and the Moon in term of attenuation and diffusion.

Most of the event’s detections are made possible due to the very low noise achieved by the instrument installation. This very low noise is in addition providing us information on the subsurface and local site, through the detection of site effect resonances and allows also to monitors the elastic and seismic interaction of a planetary surface with its atmosphere, illustrated not only by a wide range of SEIS signals correlated with pressure vortexes, dust devils or wind activity but also by modulation of lander and ground resonances above 1 Hz, amplified by ultra-low velocity surface layers, and by the observation of atmospheric waves generated ground displacement.

After 18 months of operation, the first constraints on the Martian interior have been extracted from this data set and covers at the time of writing the elastic structure of the subsurface and upper crust. Deeper constraints have also been obtained on the attenuation, for both the crust and lithosphere and on the diffusivity of the crust. More deeper information will progressively be accumulated, with the detection of larger magnitude quakes and through the processing of longer time series of seismic noise.

Presenting Author: Philippe Lognonné

Additional Authors

Session: Back to the Future: Innovative New Research with Legacy Seismic Data

Type: Oral

Date: 4/28/2020

Time: 08:45 AM

Room: 230 + 235

Description: Back to the Future: Innovative New Research with Legacy Seismic Data

The Great 1906 San Francisco earthquake (April 18, M7.8) remains a monumental and unique historic event. In the U.S., it began an era in earthquake seismology: direct observations archived in Lawson’s (1908) lasting “Carnegie Report” of surface displacement, combined with analysis of the surrounding crustal deformation, led Reid (1910) to formulate the elastic rebound theory. The earthquake’s grandeur and extent resulted in shaking, damage and ground failures that were systematically studied in a new era of scientific excellence. Documentation includes archived seismic recordings from 96 stations worldwide, thousands of macroseismic observations and hundreds of reported landslides and liquefaction reports—many with evidential photos. For this session on Legacy Seismic Data, we focus on two of the datasets in particular: the macroseismic and ground failure observations. Specifically, we revisit the macroseismic data set aggregated by Boatwright and Bundock (2005) to generate a very well-constrained historical ShakeMap—while noting challenges reproducing these observations with existing ground motion models—and we digitize the locations of ground failure features documented in the Report to evaluate how well globally applicable ground failure models hindcast those effects. In general, the global models fare well, yet beg for regional model development that can accommodate more detailed geotechnical data constraints. We test models applicable at a regional scale to investigate if regional geotechnical constraints—not available at a global scale—significantly improve our landslide and liquefaction estimates. Such historical datasets uniquely allow testing and comparison of our ability to predict shaking-induced hazards given an event that has both well-constrained triggering and ground failure observations. The landslide and liquefaction data digitized as part of this project are archived in An Open Repository of Earthquake-Triggered Ground-Failure Inventories (Schmitt et al., 2017).

Presenting Author: David J. Wald

Additional Authors

Session: From Aseismic Deformation to Seismic Transient Detection, Location and Characterization [Poster]

Type: Poster

Date: 4/29/2020

Time: 08:00 AM

Room: Ballroom

Description: From Aseismic Deformation to Seismic Transient Detection, Location and Characterization [Poster]

An intriguing sequence of a 2-stage SSE in Guerrero and a simultaneous SSE in Oaxaca took place in Mexico in 2017-2019. Three large earthquakes occur during these SSEs adding complexity to the observed surface deformations. The objective of this work is to explain the interaction between the overlapping seismic and aseismic events through the analysis of continuous GPS observations.

We perform kinematic inversion of the GPS time series solving for the cumulative slip distribution on the subduction interface due to two SSEs, using Independent Component Analysis Inversion Method (ICAIM, Gualandi, 2015). The daily position time series for 2017-2019 are obtained by processing continuous data using GAMIT/GLOBK 10.7 (Herring et al, 2018). Strong postseismic signals generated by the following earthquakes 08/09/2017 Mw8.2 in Tehuantepec, 19/09/2017 Mw7.1 in Puebla-Morelos and 16/02/2018 Mw7.2 in Pinotepa are removed using the ICA decomposition.

Our results show complex slip evolution on the subduction interface. We observe a clear change of cumulative seismic moment release rate after large seismic events of 2017 and after the earthquake in Pinotepa in 2018. The occurrence of Mw8.2 and Mw7.1 events notably slowed down the slip propagation of the Guerrero SSE. Continuous SSE in Oaxaca propagates from the northeast near the city of Oaxaca (-97.00°E, 16.70°N) towards the southwest approaching Pinotepa (-98.00°E, 17.00°N). Guerrero SSE migrates from the origin of its 1st phase near Tecpan (-100.50°E, 17.50°N) southeastwards to Acapulco (-99.50°E, 17.20°N) where the 2nd stage develops. Therefore the stress changes induced by the two aseismic events likely triggered the Mw7.2 Pinotepa earthquake (-98.01°E, 16.22°N).

Presenting Author: Ekaterina Kazachkina

Additional Authors

Session: Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska

Type: Oral

Date: 4/28/2020

Time: 05:30 PM

Room: 230 + 235

Description: Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska

The U.S. Geological Survey (USGS) computes design ground motions by combining hazard results from the USGS National Seismic Hazard Models (NSHMs) with the site-specific design code procedures of the NEHRP Provisions. Prior to the 2020 NEHRP Provisions, these design code procedures were applied to the USGS ground motions at three spectral periods (0, 0.2 and 1s) at the reference site class (BC). These procedures included risk-targeted calculations, multiplication by maximum-direction factors and deterministic capping. To obtain design ground motions for other site classes, multiplication by NEHRP site amplification factors was required. Design ground motions for other spectral periods were then derived from a Two-Period-based Response Spectrum (TPRS) shape. As a result, the updates of the USGS NSHM have traditionally given much attention to the behavior of ground motion models (GMMs) at the three mentioned periods and one site class.

For the 2020 NEHRP Provisions, the design code procedures have been updated to use a Multi-Period Response Spectrum (MPRS) in order to overcome the recognized shortcomings of the TPRS and site amplification factors. As a result, the site amplification factors were eliminated from the provisions, which now rely on the USGS to provide hazard results for more periods (22) and site classes (8). We discuss how the MPRS affects the selection and modification of GMMs in the 2018 and future updates of NSHMs. The 2018 NSHM update for the conterminous U.S. (CONUS), which is used to compute the design ground motions of the 2020 NEHRP Provisions, selected and modified GMMs such that they are applicable at all periods and site classes of interest and in some regions considered sedimentary basin effects. Outside of the CONUS (e.g., Hawaii and Alaska), where the USGS has not yet developed new hazard models, the design ground motions of the 2020 NEHRP Provisions are calculated using newly developed generic spectral shapes to achieve estimates of MPRS.

Presenting Author: Sanaz Rezaeian

Additional Authors

Session: Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska [Poster]

Type: Poster

Date: 4/28/2020

Time: 08:00 AM

Room: Ballroom

Description: Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska [Poster]

The U.S. Geological Survey’s National Seismic Hazard Model Project (NSHMP) plans to next update the National Seismic Hazard Model (NSHM) for the conterminous U.S. in 2023. The timing of the update will allow for (1) two years of development, (2) stakeholder feedback early and often (e.g., kick-off workshop in January 2021 and regional workshops throughout the update process), (3) one year for external review of the NSHM documentation and publication and (4) coordination with the next update of the NEHRP Recommended Seismic Provisions for New Buildings and Other Structures (and the 2028 ASCE 7 Standard and the 2023 International Building Code). The timeline for the 2023 NSHM update is posted on our website (https://www.usgs.gov/natural-hazards/earthquake-hazards/hazards).

The previous update in 2018 was focused on updating central and eastern U.S. (CEUS) ground motion models (GMMs) and incorporating basin amplifications in four urban areas in the western U.S. (WUS). The 2023 update will focus on updating the WUS source model (e.g., new faults) and subduction-zone GMMs (i.e., NGA-Subduction). Additional updates may include improved Atlantic and Gulf Coast site-effect models, the Uniform California Earthquake Rupture Forecast (UCERF) with updated catalogs and other logic tree modifications, GMMs derived from physics-based 3D simulations, non-ergodic aleatory ground motion uncertainty, additional basin-effect models for urban areas in the WUS, updated geodetic models, new magnitude-area scaling relations and improved treatment of the WUS-CEUS attenuation boundary. The NSHMP requests that the earthquake hazards community bring to our attention new source and ground motion data or models that could be included in the 2023 NSHM. The deadline for publication of external data or models to be considered by the NSHMP for this update is December 31, 2020.

Presenting Author: Allison M. Shumway

Additional Authors

Session: Earthquake Early Warning: Current Status and Latest Innovations [Poster]

Type: Poster

Date: 4/29/2020

Time: 08:00 AM

Room: Ballroom

Description: Earthquake Early Warning: Current Status and Latest Innovations [Poster]

We have developed a global seismic monitoring capability based on real-time measurements from ~ 1,000 GNSS receivers to rapidly characterize large earthquakes and, where relevant, tsunami. It complements traditional seismic monitoring by allowing moment release to be quantified while fault rupture unfolds. Position time series from stations distributed across six continents are continuously estimated within an earth center of mass-fixed reference frame and streamed as local north, east and vertical coordinates into a variety of seismic monitoring algorithms and also rebroadcast for third-party use. Average positioning latency, which includes satellite observable acquisition, telemetry and processing, averages about 1.4 seconds on our production system and 0.5 seconds on a development system. Ongoing efforts through 2020 will up the number of global stations to ~2500.

This system captured the 2019 Ridgecrest California M7.1 earthquake and determined its coseismic deformation of up to 70 cm on 12 nearby stations within 22 seconds of event nucleation. Those 22 seconds comprise the fault rupture time itself (roughly 5-10 sec), another ~5s for propagation delay between various regions of slip and GNSS stations, another 5-10s for dynamic displacements to dissipate such that coseismic offsets ‘settle down,’ plus another 1.4 seconds for telemetry and data analysis latency. Comparison of coseismic deformation estimated within 25 seconds to that determined with post-processing using several days of post-processing show that the real-time offsets were accurate to within 10% of the post-processed “true” offsets.

While ~1 sec determination of GNSS coseismic offsets could not help ShakeAlert improve its initial magnitude assessment made several seconds after nucleation, high-M6 magnitude at ~25 seconds could have nonetheless triggered or revised an alert before S-waves reached the LA Basin. This highlights how GNSS can improve magnitude estimation for events whose duration and extent of rupture is difficult using only P-wave amplitudes.

Presenting Author: Timothy I. Melbourne

Additional Authors

Session: Near-Surface Effects: Advances in Site Response Estimation and Its Applications

Type: Oral

Date: 4/28/2020

Time: 02:30 PM

Room: 110 + 140

Description: Near-Surface Effects: Advances in Site Response Estimation and Its Applications

It is well known that the characteristics of the shallow geology can affect the seismic wave propagation and may produce large ground motion amplifications during earthquakes. This is mainly due to their low velocity compared to the underlying bedrock. In sedimentary basins, their geometry can furthermore lead to seismic wave focusing, hence complexifying ground motion amplification. The assessment of these site effects is important for seismic hazard evaluation. Numerical simulations of seismic wave propagation can be used to understand and predict site amplification. The reliability of simulations depends on how the velocity model used reflects the reality. For long period or large-scale simulations, smooth velocity models are generally adequate. At short periods, where the effect of small-scale soil heterogeneities on seismic wave propagation can be important and their consideration becomes crucial. Due to the difficulty to map these heterogeneities, statistical methods are often used to model them. Using a detailed velocity model of the Nice (France) sedimentary basin, we performed 2-D numerical simulations of seismic wave propagation to investigate the effect of short-wavelength heterogeneities on surface ground motion. Velocity heterogeneities within the basin are modeled statistically as a stationary, homogeneous random process with a 2-D Von Karman autocorrelation function. We investigate how different parameters of this function affect ground motion in terms of peak ground velocity, pseudo-spectral acceleration and standard spectral ratios. We show that the presence of velocity heterogeneities increases ground motion variability and may increase or decrease the spectral amplification depending on the local configuration of the media’s statistical parameters.

Presenting Author: Flomin N. Tchawe

Additional Authors

Session: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise

Type: Oral

Date: 4/29/2020

Time: 08:30 AM

Room: 230 + 235

Description: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise

We perform a suite of 0-5 Hz deterministic simulations of the 2014 Mw 5.1 La Habra, CA, earthquake, with the parallel AWP-ODC-GPU code in a 28 km by 28 km by 12 km mesh from the Southern California Earthquake Center (SCEC) Community Velocity Model CVM-S4.26-M01 using a minimum shear-wave velocity (Vs) of 200 m/s. The finite-fault source is obtained from the Graves and Pitarka (2016) kinematic rupture generator, with statistics constrained by dynamic rough fault simulations. We include statistical distributions of small-scale crustal heterogeneities (SSHs) with an anisotropic von Karman autocorrelation to represent the effects of realistic velocity and density perturbations, parameterized by a standard deviation, Hurst number, correlation length and H/V anisotropy. In addition, frequency-dependent attenuation Q(f) = and surface topography using a curvilinear grid (O’Reilly et al., 2019) are included in our simulations. Strong motion data is available for comparison at a series of sites in our model domain.

We compare our simulation results to a reference case with a flat free-surface, frequency-independent Q = 100 Vs (Vs in km/s) and no SSHs included. We use an initial model with Q(f) = 100 Vs f^0.6 and SSHs with a standard deviation of 5%, a Hurst number of 0.1, correlation length of 100 m and a horizontal-to-vertical anisotropy of 5. Our results show that, as expected, the effects of topography, SSHs and Q(f) become increasingly important as frequencies increase toward 5 Hz. We find that SSHs, topography and Q(f) all contribute significantly to the synthetic ground motion by generally increasing values of duration-dependent metrics (i.e., cumulative energy) toward those derived from the recorded time series. In addition, at most stations, the best fit are obtained by including topography, SSHs and Q(f) in the simulations. Our results indicate that topography, SSHs and Q(f) need to be included in ground motion modeling for frequencies as high as 5 Hz.

Presenting Author: Zhifeng Hu

Additional Authors

Session: Early Results from the 2020 M6.4 Indios, Puerto Rico Earthquake Sequence [Poster]

Type: Poster

Date: 4/30/2020

Time: 08:00 AM

Room: Ballroom

Description: Early Results from the 2020 M6.4 Indios, Puerto Rico Earthquake Sequence [Poster]

Puerto Rico has been recently affected by an earthquake sequence that began on December 28, 2019 and to this writing has not ended. So far, more than 2,000 earthquakes have affected the island, of which 32 earthquakes have had magnitudes of 4.5 or greater. The main shock of 6.4 Mw occurred on January 7 and has been the most destructive earthquake to hit Puerto Rico in 100 years. One person has died, a hundreds of buildings have been damaged. Although earthquake damage is evident, we are unaware of any reported on land surface ruptures. Four epicenters of magnitudes 4.9, 5.0, 5.6 and 5.9 were located on land or close to the coast. We use Multi Aperture Interferometry (MAI) and DInSAR to obtain 3D estimation maps of the ground displacement. We used Synthetic Aperture Radar data of the Copernicus Sentinel-1 satellite operated by the European Space Agency and processed it with Environment for Visualizing Images software. We then produced line-of-sight and along-track displacement maps, which yielded 3D coseismic displacement maps. In addition, we spatially compared the location of ground displacement with previously mapped faults on geological maps and on a digital elevation model produced from Lidar at 10 m resolution. The study area is covered by Oligocene to Miocene clastic and carbonate rocks that contain NE-SW to E-W striking normal faults. Underneath these rocks lie a Cretaceous volcanic and sedimentary sequence with NW-SE striking sub-vertical faults with normal and left-lateral senses of movement. The focal mechanisms calculated by the U.S. Geological Survey for the four on land >4.8 M earthquakes indicate NE-SW striking normal faulting and NW-SE left-lateral faulting, which suggests a similar strain regime compared to the mapped faults in the study area. Results show the greatest ground displacement occurred west of the city of Ponce up to 12 cm between 12/28/2019 and 01/09/2020 and up to 19 cm between 01/02/2020 and 01/14/2020 in the line sight of the radar.

Presenting Author: Zelalem Demissie

Additional Authors

Session: What Can We Infer About the Earthquake Source Through Analyses of Strong Ground Motion?

Type: Oral

Date: 4/29/2020

Time: 05:15 PM

Room: 115

Description: What Can We Infer About the Earthquake Source Through Analyses of Strong Ground Motion?

In addition to the M6.4 and M7.1 events, the Ridgecrest sequence produced several moderate magnitude (~M5) earthquakes that were also well-recorded by the roughly 300 stations of the Southern California Seismic Network (SCSN). Using these data, we examine the adequacy of existing Southern California Earthquake Center (SCEC) 3D Community Velocity Models (CVMs) to simulate long period (T > 1 sec) ground motions across southern California. Preliminary analysis of the motions from moderate magnitude events indicates that sites in the epicentral area have very simple waveforms, indicating that these events can be treated as point sources at the longer periods and that the subsurface structure near these sites is relatively simple. However, sites to the west of the epicentral area in Ridgecrest exhibit significant later-arriving, long-period motions, suggesting these arrivals may be basin-generated surface waves. Further to the south in the Los Angeles region, the recorded motions exhibit significant amplification (factors of 3-4) and long shaking durations for sites within the deep basins as compared to sites located outside of the basins. In Ridgecrest and across the Mojave Desert, both SCEC CVMs (cvmh and cvmsi) are comprised of smoothly varying velocity structure that is constrained primarily by waveform tomography. To the south, this structure transitions into several deep low-velocity basins in the Los Angeles region, although the details of these structures differ somewhat between the two CVMs. In the first phase of our work, we simulate the long-period motions for the moderate magnitude events using both CVMs and assess the ability of each model to reproduce the observed motions at sites across the southern California region. In the second phase, we extend our analysis to the M6.4 and M7.1 events in order to explore our ability to capture both complex finite-faulting effects along with 3D wave propagation effects.

Presenting Author: Robert Graves

Additional Authors

Session: What Can We Infer About the Earthquake Source Through Analyses of Strong Ground Motion?

Type: Oral

Date: 4/29/2020

Time: 03:15 PM

Room: 115

Description: What Can We Infer About the Earthquake Source Through Analyses of Strong Ground Motion?

The strong ground motion parameters, such as peak ground acceleration (PGA) and peak ground velocity (PGV), contain important high frequency source information that is difficult to be extracted with conventional finite fault studies. We note that the magnitude dependence of mean log(PGA) and mean log(PGV) estimated using the NGA West-2 database features a sharp change in slope at about M 5.3, which cannot be reproduced by the stochastic ground-motion modeling using a self-similar double-corner-frequency (DCF) source spectrum model. Here we present a DCF model JA19_2S that has two magnitude ranges; this model can well match this abnormal feature. Its lower corner frequency f_c1 satisfies for log(f_c1)=1.474-0.415M for 3.3<M<5.3; log(f_c1)=2.375-0.585M for 5.3<M<7.3. Its high corner frequency f_c2 follows log(f_c2)=3.25-0.5M, for 3.3<M<7.3. f_c1 is related with the average source duration T d =1/\pi f_c1 and f_c2 is related with the average rise time T_R ~1/ f_c2 .T_d then increases with seismic moment as M₀^1/3+a , with a=0.614 for 3.3<M<5.3 but a= -0.436 for 5.3<M<7.3. This can be the result of either rupture velocity increasing 47% or static stress drop increasing 3.2 times when earthquake magnitude increases from 3.3 to 5.3 and subsequently decreasing by the same amount when magnitude further increases from 5.3 to 7.3. The scaled energy (ratio of seismic radiated energy and seismic moment) of this DCF model scales with two corner frequencies as M₀f_c1² f_c2. This model then also implies that the scaled energy varies from 2.2x10^-5 to 4.8x10^-5 when magnitude increases from 3.3 to 5.3 and subsequently decreases to 2.2x10^-5 when magnitude further increases to 7.3. While these predictions can be evaluated with independent seismic observations, for 5.3<M<7.3 the predicted acceleration spectrum using JA19_2S remarkably agrees with the additive DCF model proposed by Atkinson and Silva (2000) for California earthquakes, though they were constructed with entirely different approaches.

Presenting Author: Chen Ji

Additional Authors

Session: Earthquake Early Warning: Current Status and Latest Innovations

Type: Oral

Date: 4/29/2020

Time: 09:00 AM

Room: 115

Description: Earthquake Early Warning: Current Status and Latest Innovations

Current EEW algorithms are continuously optimized to strive for both timely and accurate ground motion estimates while also avoiding false alarms. However, they may still create missed or false events for complicated earthquakes. We aim to develop an algorithm that is able to handle such complicated cases, specifically foreshock-mainshock pairs and swarm-like sequences. Our algorithm finds earthquake source parameter estimates by matching the incoming observed ground motion envelope with two other types of envelopes that are based on the Cua-Heaton (CH) GMPEs and a catalog of earthquake history. To ensure that optimal parameter estimates are not missed, the algorithm is split into two independent approaches that check each other’s results: (i) a grid search that finds the best-fitting CH envelopes and (ii) a catalog search of envelopes from past real earthquakes and synthetic ground motions. In this talk, we focus on the second approach.

While the grid search gives average results, the second approach of the algorithm, the catalog search, provides more station-specific results. Here, we look for similar envelopes from nearby events. We find that in many cases, the catalog search reduces both error and computation time of the estimates given by the grid search. To ensure the catalog search works adequately and can handle complicated cases, we extend the catalog to cover a large enough range of earthquakes. We accomplish this by using Brune’s scaling model to produce unavailable data of large earthquakes from data of smaller ones. Also, we generate foreshock-mainshock pairs to include in the extended catalog by taking different combinations of envelopes and time delays. The goal is to create a catalog of complex sequences.

The advantage of matching envelopes is that it helps visualize when an abrupt signal arrives, such as a second, larger earthquake. With these two approaches working together, we hope to achieve an EEW algorithm that will not miss arrivals of multiple events.

Presenting Author: Becky H. Roh

Additional Authors

Session: Early Results from the 2020 M6.4 Indios, Puerto Rico Earthquake Sequence

Type: Oral

Date: 4/30/2020

Time: 11:30 AM

Room: 115

Description: Early Results from the 2020 M6.4 Indios, Puerto Rico Earthquake Sequence

On January 7, 2020 a Mw6.4 earthquake occurred on an oblique strike-slip fault, approximately 8km from Indios, Puerto Rico. To date this is the largest earthquake in a sequence of events that began on December 28^thand continues through the submission of this abstract. The Mw6.4 was recorded by many strong motion stations at distances from 19 km to more than 500 kms. The available strong motion data will be processed, reviewed and the recorded motions will be compared to predicted ground motions based on the Next Generation Attenuation Relationships-West 2 (NGA-West2) by Abrahamson et al. (2014), Boore et al. (2014), Campbell and Bozorgnia (2014) and Chiou and Youngs (2014). Additional comparisons between measured Arias Intensity and the 2016 GMPE for Arias Intensity by Abrahamson et al. (2016) will be made. and compare ground motion intensity parameters, D_5-75, D_5-95, Arias Intensity and Cumulative Absolute Velocity based on Bommer et al. (2009), Kempton and Stewart (2006) and Abrahamson and Silva (1996). Lastly, we will rotate the available recorded motions to Fault Normal and Fault Parallel and summarize our findings.

Presenting Author: Jennie Watson-Lamprey

Additional Authors

Session: Earthquake Source Parameters: Theory, Observations and Interpretations [Poster]

Type: Poster

Date: 4/30/2020

Time: 08:00 AM

Room: Ballroom

Description: Earthquake Source Parameters: Theory, Observations and Interpretations [Poster]

Chile experiences considerable seismic hazard from interplate, intraplate and crustal earthquakes. Due to their large magnitudes and frequent occurrence in the Nazca subduction zone, interplate and intraplate earthquakes have been widely studied with broadband and strong motion networks. However, the smaller-magnitude crustal seismicity has been studied much less, and the lack of sufficiently dense seismic networks over most of the country makes comprehensive studies of crustal events very challenging.

In Northern Chile, the interseismic gap since the last giant thrust earthquake in 1877 has motivated the deployment of several temporary and permanent seismic networks since the early 2000s, thus providing new opportunities to carry out seismological studies in the area, especially between latitudes 18°S and 24°S.

We select more than 800 M_L ≥ 3.0 crustal earthquakes that occurred between 2005 and 2018, reported in two recently-published earthquake catalogs for the area, and calculate their focal mechanisms using P-wave polarities and S/P amplitude ratios from broadband and short-period waveform data recorded by local stations.

Our initial results show a variety of faulting geometries, with some trends becoming evident. Several offshore crustal events show north-south strike with thrust mechanisms, indicating strong influence of the main plate subduction. However, several onshore events near 21°S show a clear east-west strike, with north-south horizonal pressure axes, which may indicate the existence of additional local tectonic features in compressive regime. Further inland, persistent seismicity occurs in the Andes mountains in a small area near latitude 19.7°S, which could be related to the 2001 Mw 6.3 Aroma crustal earthquake.

Our compilation of crustal-earthquake results for Northern Chile can provide clues to local and regional faulting patterns, and serve as a basis for crustal stress mapping studies in the region.

Presenting Author: Carlos Herrera

Additional Authors

Session: Innovative Seismo-Acoustic Applications to Forensics and Novel Monitoring Problems

Type: Oral

Date: 4/29/2020

Time: 04:45 PM

Room: 240

Description: Innovative Seismo-Acoustic Applications to Forensics and Novel Monitoring Problems

In July of 2019, staff at the Oklahoma Geological Survey (OGS) documented an unusual observation in their field blog: seismic stations distributed over the state of Oklahoma (OK) had recorded sequences of repetitive, quasi-similar signals almost daily over the previous spring and summer months. These signals appeared after 16:00 UTC in the 5-15Hz band as wavetrains of short duration pulses (~1-3s widths) that were separated by 20s intervals and moved near acoustic propagation speeds over widespread regions of OK. The onset of a wavetrain at a particular station was generally marked by up to 26 distinct pulses, each separated in time by integer multiples of 20s, recorded for 1200s per day and for multiple months. These observations drew attention from multiple media platforms that include National Geographic and Scientific American. Our research found that these seismo-acoustic signals originated from demolition explosions (shots) that are regularly carried out by the Army at the McAlester Ammunition Plant (McAAP), in designated shot pits. The observation of these seismo-acoustic signals on sensors up to ~350km from McAAP, their report over media and concern from OK residents gave this so-called “acoustic anomaly” visibility to the public and media. We extend the initial cursory analysis of this acoustic anomaly from a public interest project to a scientific one and leverage 9 months of waveform observations on ≤ 115 seismic stations to perform an empirical perturbation analysis of atmospheric variability, over a range of spatial and temporal scales that span the troposphere to the thermosphere and that span seconds to seasons in time. We verify these observations against ground truth data documented in McAAP blast-logs to reveal a spatial-temporal variability in infrasonic return paths that is not captured by low resolution space-borne and weather balloon observations. Our assessment demonstrates that explosion-triggered pulses can resolve changes in the atmospheric state and winds with sub-minute fidelity.

Presenting Author: Joshua D. Carmichael

Additional Authors

Session: Recent Development in Ultra-Dense Seismic Arrays With Nodes and Distributed Acoustic Sensing (DAS)

Type: Oral

Date: 4/30/2020

Time: 04:15 PM

Room: 110 + 140

Description: Recent Development in Ultra-Dense Seismic Arrays With Nodes and Distributed Acoustic Sensing (DAS)

In November 2016, a M5 earthquake stuck the town of Cushing, the largest crude oil storage facility in the world. The earthquake caused significant building damages, especially some historic buildings. Preliminary analysis suggests that the earthquakes are mostly shallower than 3 km, which could cause higher shaking near the epicenter. To further understand the fault properties and monitor earthquake activities, 130 Fairfield 3C nodes were deployed in 5-by-5 km wide area on top of the fault system. The instruments recorded continuously from November 5 to December 15. Due to logistical and property right constraints, the instruments were deployed along county roads. The resulting survey grid comprises 4 E-W lines and 5 S-N lines with receiver spacings of 150 and 250 meters. In this study, we present preliminary results using the dense nodal array to analyze a M2.5 earthquake sequence that occurred close to the array in November 2019. We also apply interferometric ambient noise analysis and calculate teleseismic receiver functions to map the subsurface structure in the area.

Previous experiments proved the suitability of dense deployments with low-frequency geophones for deep earth studies, and we analyze our data for reflections from the outer and inner core. These phases can not only reveal information about the structure of the core but can also be used to image the shallow reflectivity structure below the recording array.

Presenting Author: Michael Behm

Additional Authors

Session: Innovative Seismo-Acoustic Applications to Forensics and Novel Monitoring Problems [Poster]

Type: Poster

Date: 4/29/2020

Time: 08:00 AM

Room: Ballroom

Description: Innovative Seismo-Acoustic Applications to Forensics and Novel Monitoring Problems [Poster]

The Dynamic Network Experiment 2018 (DNE18) was a virtual experiment designed to quantitatively assess current capabilities of multi-modal data processing for local and regional nuclear explosion monitoring. The experiment was a collaborative effort between Los Alamos National Laboratory (LANL), Sandia National Laboratories (SNL), Lawrence Livermore National Laboratory and Pacific Northwest National Laboratory. One component of the experiment was the establishment of a baseline for infrasonic signal detection and event identification. This baseline was established using proprietary infrasonic data processing packages at both LANL (InfraPy) and SNL (Bloodhound). The DNE18 data set spanned December, 2010 through February, 2011 and included data collected on the University of Utah seismic network. The analysis of this data set using a consistent set of processing parameters allows for a direct comparison of the algorithms utilized in each processing pipeline. For each processing toolkit, automated detection and event catalogs were created. The detection catalogs have been compared against both each other and an analyst-produced catalog to quantify detector performance in terms of precision and recall. Subsequent event association and location results in the event catalogs have been compared in terms of completeness and measures of error. This comparison expands upon previously published studies quantifying the performance of automatic infrasonic signal detectors and also presents a comparison of infrasonic detection, association and localization analyses in a functioning pipeline.

Presenting Author: Fransiska K. Dannemann Dugick

Additional Authors

Session: Seismicity and Tectonics of Stable Continental Interiors

Type: Oral

Date: 4/30/2020

Time: 09:00 AM

Room: 240

Description: Seismicity and Tectonics of Stable Continental Interiors

At its nearest, northern Australia is just over 400 km from an active convergent plate margin. This complex and unique tectonic region combines active subduction and the collision of the Sunda-Banda Arc with the Precambrian North Australian Craton (NAC) near the Timor Trough and continues through to the New Guinea Highlands. Ground-motions generated from earthquakes on these structures have particular significance for northern Australian communities and infrastructure projects, with several large earthquakes in the Banda Arc region having caused ground-shaking-related damage in the northern Australian city of Darwin over the historical period.

There are very few, if any, present-day tectonic analogs where cold cratonic crust abuts a convergent tectonic margin with subduction and continent-continent collision. Ground motions recorded from earthquakes in typical subduction environments are highly attenuated as they travel through young sediments associated with forearc accretionary prisms and volcanic back-arc regions. In contrast, seismic energy from earthquakes in the northern Australian plate margin region are efficiently channelled through the low-attenuation NAC, which acts as a waveguide for high-frequency earthquake shaking. As such, it is difficult to select models appropriate to the region for seismic hazard assessments.

The development of a far-field ground-motion model to support future seismic hazard assessments for northern Australia is discussed. In general, the new model predicts larger ground motions in Australia from plate margin sources than models used for the 2018 National Seismic Hazard Assessment of Australia, none of which are considered fully appropriate for the tectonic environment. Short-period ground motions are strongly dependent on hypocentral depth and are significantly higher than predictions from commonly-used intraslab ground-motion models at comparable distances. This depth dependence in ground motion diminishes with increasing spectral periods.

Presenting Author: Trevor I. Allen

Additional Authors

Session: Earthquake Source Parameters: Theory, Observations and Interpretations

Type: Oral

Date: 4/30/2020

Time: 11:30 AM

Room: 120 + 130

Description: Earthquake Source Parameters: Theory, Observations and Interpretations

We present an updated focal mechanism catalog for 129,430 events that occurred in southern California from 2011 to 2018. The approach uses S/P amplitude ratios and first-motion P wave polarities to determine focal mechanisms with the HASH method of Hardebeck and Shearer (2002, 2003). In addition to the broadband channels used by Yang et al. (2012), three-component short period channels (EH-) are also used to obtain P- and S-wave amplitudes. Moreover, we utilize a convolutional-neural-network (CNN) phase picker (Ross et al., 2018b) and polarity picker (Ross et al., 2018a), which are trained on millions of manually labelled seismograms from Southern California, to detect more phase arrivals and polarities and further improve the focal mechanism catalog. For all events, we detect additional P- and S-wave arrivals using the CNN phase picker to obtain more S/P amplitude ratios. For 126,999 small magnitude events (M < 2.8), extra CNN P-wave first-motion polarities are also added into focal mechanism calculations. More than 60% of the events have at least 5 additional polarities and S/P amplitude ratios compared with Yang et al., leading to 63% more (59,679) mechanisms than 36,668 focal mechanisms in Yang et al. (2012), with >90% quality A and B common events having rotation angles less than 35 degree, which is the uncertainty value used to define quality B focal mechanism. The number of quality B, C and D focal mechanisms increased by 43%, 84% and 63%, respectively. We also apply CNN pickers to 65 co-located events with similar waveforms as well as 782 M > 3 events with moment tensor solutions to test the accuracy improvement. The solved focal mechanisms are more consistent after implementing CNN pickers. The results indicate that CNN pickers can significantly improve the quality and quantity of the obtained focal mechanisms. The procedure will be applied to the rest of the catalog events in southern California to improve the focal mechanisms data base and related analysis results.

Presenting Author: Yifang Cheng

Additional Authors

Session: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise

Type: Oral

Date: 4/29/2020

Time: 03:00 PM

Room: 230 + 235

Description: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise

We have developed an empirical spatial correlation model of normalized within-event residuals for the Fourier Amplitude Spectra (FAS), as FAS is a more straightforward representation of the ground motion frequency content than the traditionally-used pseudo-spectral acceleration. The spatial correlation model is developed using the orientation-independent Effective Amplitude Spectra from the Pacific Earthquake Engineering Research Center Next Generation Attenuation West2 database. The database includes shallow crustal earthquakes (M>3) in active tectonic regions. A linear coregionalization model (with three functions representing short-range, long-range and nugget effects) is used to fit the semivariogram from data using the Goulard-Voltz algorithm (Goulard and Voltz, 1992) for frequencies from 0.1 Hz to 20 Hz and separation distances up to 120 km. No significant magnitude dependence is found for the spatial correlation coefficients.

Since exclusion of spatial correlation in ground motion simulations can result in an under-estimation of the seismic risk, the developed frequency-dependent spatial correlation has been implemented into the San Diego State University ground motion generation module on the Southern California Earthquake Center Broadband Platform using a post-processing method. This method makes use of three independent two-dimensional Gaussian random variables with covariance corresponding to the three functions of the spatial correlation model. We present results from applying the method to several historical earthquakes in the western USA. The results indicate that the spatial correlation model for FAS is consist with the model for spectral acceleration (Loth and Baker, 2013), providing an additional improvement of the spatial correlation in spectral accelerations.

Presenting Author: Nan Wang

Additional Authors

Session: Leveraging Advanced Detection, Association and Source Characterization in Network Seismology [Poster]

Type: Poster

Date: 4/30/2020

Time: 08:00 AM

Room: Ballroom

Description: Leveraging Advanced Detection, Association and Source Characterization in Network Seismology [Poster]

Detecting induced seismicity using surface arrays located near injection wells is challenging mainly due to the noise introduced by the anthropogenic activity. Additionally, induced events with magnitudes M<0 typically have small amplitudes and short durations (~ 0.5-1 s), which complicate event detection in the time-domain and inflate false detections. To overcome these issues, we introduce a new frequency-domain-based algorithm for detecting induced events with small magnitudes using dense nodal arrays. Our proposed method takes advantage of the high energy carried by S waves, the known location of the injections and the geometry of the array. The processing stages involved in event detection are: (1) Preprocess the raw data by filtering, removing the mean and linear trend and rotating the horizontal components to the direction of the injection well, (2) Compute 2-hr long spectrograms for each station, (3) Sum the amplitudes for a given frequency range and return the envelope of the stacked amplitude - time series across the array and (4) Run a sliding window and declare detections above a threshold. We apply the new method during an injection phase at the Utah Frontier Observatory for Research in Geothermal Energy (FORGE) in southcentral Utah, where a dense circular array consisting of 151 5-Hz three component nodal geophones over a 5 km aperture was in operation. The geophones were distributed in five concentric rings centered around well 58-32 and were recording continuously at 1 kHz, during the different injection stages. Event detections derived by the new method are compared to the catalog provided by Schlumberger compiled using a deep borehole 12-level geophone string.

Presenting Author: Kristine Pankow

Additional Authors

Session: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise [Poster]

Type: Poster

Date: 4/29/2020

Time: 08:00 AM

Room: Ballroom

Description: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise [Poster]

We present an implementation of finite element solver using an octree data structure which incorporates topography and strain-dependent rheologies using a conforming octree-based mesh. The former are features of interest for earthquake engineering and seismological applications. The performance of this solver (ATLAS) is grounded on the variable polynomial order of the field approximation implemented, as in most standard finite element codes, together with a conforming mesh, which preserves the semistructured character of octrees and their inherent advantages in memory usage.

The analysis of a canonical example, including topography, computed using the Indirect Integral Equation Method shows an improvement in the convergence of ATLAS when compared against octree-based non-conforming low order code. In addition, we analyze different alternatives to ameliorate the spurious modes introduced in displacement-based approximations when dealing with large p- to s-wave ratios. We conclude that a stress-velocity formulation might offer better performance for seismological applications.

Presenting Author: Leonardo Ramirez-Guzman

Additional Authors

Session: Earthquake Source Parameters: Theory, Observations and Interpretations

Type: Oral

Date: 4/30/2020

Time: 01:30 PM

Room: 120 + 130

Description: Earthquake Source Parameters: Theory, Observations and Interpretations

A primary way of investigating the enigmatic nature of deep-focus earthquakes is by studying the source properties of these events. At these great depths, the orientations of rupture planes are difficult to determine given the limited knowledge of preexisting faults and the lack of aftershocks generally associated with large deep-focus events. This lack of rupture plane information hinders many source modeling methods and limits insights that may constrain the mechanisms that promote these enigmatic events. In this study, we utilize P waves recorded at the Hi-net array in Japan with a back-projection method to study the source properties of the 2015 Mw 7.9 Bonin deep-focus earthquake. To limit artifacts caused by seismic phases other than the direct P waves, a genetic algorithm method is applied to the selection of data that maximizes the point source nature of back-projection results from small events in the study region. This approach combined with the excellent aperture of the local Hi-net array allows for high-resolution 4D imaging of the earthquake source. This analysis identifies three rupture segments of this event based upon the locations of imaged energy. Segments 1 and 3 propagate with a horizontal orientation to the northwest and are separated by 15 km. Segment 2 exhibits both westward and upward propagation and terminates at the initiation point of segment 3. Combined, the orientations of these three segments cannot be fit with a single plane, suggesting either failure of multiple rupture planes during this event or failure on a curved surface. Interestingly, the orientations of segments 1 and 3 are consistent of the near vertical nodal plane of the focal mechanism of this event, while segment 2 is only consistent with the near horizontal nodal plane. One interpretation of these results is that the earthquake involved failure on conjugate rupture planes. This failure may have been induced by extensional stresses in the heel of the subducting slab as it is deflected to a horizontal orientation upon encountering the 660 km discontinuity.

Presenting Author: Eric Kiser

Additional Authors

Session: Recent Advances in Very Broadband Seismology

Type: Oral

Date: 4/30/2020

Time: 11:30 AM

Room: 215 + 220

Description: Recent Advances in Very Broadband Seismology

Unusual for a Global Seismograph Network (GSN) borehole station, MSVF (Monasavu, Fiji) includes only a single 100 m borehole, which hosts the station's primary seismometer. Despite high seismic noise levels there, MSVF thus offers no noise environment comparable to that of the primary sensor where a traditionally-installed secondary seismometer can be deployed. However, a non-traditional installation of the secondary sensor---in the borehole's unoccupied upper portion---is possible. But can it be done without loss of primary seismometer data quality and without risking the removal of either sensor?

Here we report on the successful testing and evaluation of a new holelock designed to allow the installation of a Streckeisen STS-5A or STS-6A seismometer above a previously-deployed seismometer. The holelock, which does not require power to operate, features a massive and robust all-mechanical fail-safe design that also provides for safe management of cables from the downhole sensor.

A prototype of the new holelock was lab tested by Project IDA at UCSD in late Fall 2019. In late November, along with a Streckeisen STS-5A, it was deployed at the XPFO test facility at Pinyon Flat, California, at a depth of 10 m in a 100 m cased borehole already occupied by a Streckeisen STS-6A near the bottom. Noise spectra from the two sensors are consistent with their relative vertical positioning in the borehole, and the noise performance of the downhole STS-6A noise remains unchanged. Deconvolved waveforms from the 15 December 2019 M6.8 Magsaysay (Philippines) event at a distance of 109 deg are essentially identical, with R² > 0.99.

The deployment of these two sensors and holelocks at MSVF is scheduled to take place during the second calendar quarter of 2020. This installation method offers simplicity in addition to logistical and cost benefits and could serve as a model for network operators when considering future seismic stations.

Presenting Author: C W. Ebeling

Additional Authors

Session: Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska [Poster]

Type: Poster

Date: 4/28/2020

Time: 08:00 AM

Room: Ballroom

Description: Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska [Poster]

We present a non-ergodic GMPE for shallow induced earthquakes in the Oklahoma region. The high-rate of earthquakes in the Oklahoma region provides the data needed to constrain the source-, path- and site-specific aspects of the GMM. The selected data set includes earthquakes with magnitudes from M 3.0 to 5.0, rupture distances less than 50 km and a minimum of five recordings per earthquake. Using the Abrahamson et al. (2014) crustal GMM adjusted for regional induced events as the reference ergodic model, we derive nonergodic adjustment terms based on a varying-coefficient model (VCM, following Landwehr et al., 2016) and the cell-specific approach by Kuehn et al. (2019) for anelastic attenuation. The VCM models a source-specific constant, as well as a site-specific constant and Vs30-scaling using a Gaussian process. The parameters required for the nonergodic adjustment terms (variances and length-scales of the Gaussian process and cell-specific attenuation coefficients) are estimated via Bayesian inference. We pay close attention to the uncertainty in all of the parameters of the model.

In a forward application, at different coordinates the estimated spatially varying coefficients allow to calculate both the median and epistemic uncertainty for the predicted ground-motion parameter as it varies spatially across the region. For coordinate locations that have no data, the non-ergodic model returns to the ergodic mean model but with large epistemic uncertainty.

Presenting Author: Nicolas M. Kuehn

Additional Authors

Session: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise [Poster]

Type: Poster

Date: 4/29/2020

Time: 08:00 AM

Room: Ballroom

Description: Numerical Modeling of Rupture Dynamics, Earthquake Ground Motion and Seismic Noise [Poster]

The USGS is working towards incorporating regional-specific seismic hazard into the U.S. National Seismic Hazard Model. The large dataset of ground motions generated from simulations can serve to supplement empirical data in areas where observed ground motion data is lacking and help constrain trends in intensity where geologic structures complicate seismic wave propagation. Machine learning provides an avenue to incorporate the plethora of information from synthetic datasets, without enforcing specific mathematical formulations that ground motion trends must follow, potentially biasing ground motion models (GMMs).

Here, we use a machine learning approach to build a ground motion model (GMM) from a synthetic database of ground motions. First, we build a dataset of predictor variables and corresponding ground motion intensities extracted from the Southern California CyberShake study. An artificial neural network (ANN) is used to find the optimal weights and coefficients that best fit the target data (without overfitting), with predictor variables chosen to match that of state-of-the-art GMMs. We compare our synthetic-based GMM with empirically based GMMs derived from the globally based Next Generation Attenuation West2 dataset, finding near-zero median residuals and similar amplitude and trends (with period) of total variability. Additionally, we find that the ANN GMM has similar bias and variability to empirical GMMs from records of the recent 2019 Mw 7.1 Ridgecrest event, which neither GMM has included in its formulation. As simulations continue to more accurately model broadband ground motions and expand across a wider range of source, site, and path conditions, machine learning provides a way to utilize the vast amount of synthetically generated data to better constrain ground motions where empirical data is sparse and to identify the relationships and relative importance between parameters for future parameterization of GMMs.

Presenting Author: Kyle Withers

Additional Authors

Session: Leveraging Advanced Detection, Association and Source Characterization in Network Seismology

Type: Oral

Date: 4/30/2020

Time: 04:30 PM

Room: 120 + 130

Description: Leveraging Advanced Detection, Association and Source Characterization in Network Seismology

Periods of enhanced seismicity consume the resources of a regional seismic network (RSN) and diminish its ability to perform other core monitoring responsibilities. Two common scenarios for enhanced seismicity encountered by RSNs are swarm and aftershock sequences. Since the temporal onset of such seismic sequences is inherently unpredictable, it is important to proactively develop a strategy for expeditiously detecting, locating and estimating the magnitude of the events comprising the sequence. When successful, such a workflow can free up analyst time to perform important high-order response tasks, such as moment tensor estimation, while also lowering the magnitude of completeness of the sequence catalog.

This abstract summarizes the salient processing features comprising the University of Utah Seismograph Stations’ anticipated response to such a period of enhanced seismicity. The overarching goals of the workflow are to detect, locate and compute the magnitude of additional events in an earthquake sequence. The first major component in the workflow is a template matching scheme for event detection. An implicit advantage of template matching over deep-learning approaches is that template matching directly leverages a priori catalog information such as observed travel times and magnitudes from a subset of the events in the sequence. To compute high-precision locations, we compute differential traveltimes by comparing the observed traveltimes with the onsets of the detected event. Additionally, we compute a relative magnitude by comparing the amplitudes of the template waveforms with the amplitudes of the detections. To demonstrate the proposed workflow in a realistic context, we apply our pipeline to a swarm of earthquakes that occurred in the San Rafael swell region of the Colorado Plateau in March-April of 2019.

Presenting Author: Ben Baker

Additional Authors

Session: Explosion Seismology Advances [Poster]

Type: Poster

Date: 4/29/2020

Time: 08:00 AM

Room: Ballroom

Description: Explosion Seismology Advances [Poster]

We present a constructive method to fuse and evaluate multiple waveform statistics that improves our capability to detect small, aboveground explosions over similar methods that use single waveform statistics. Our method advances Fisher's Combined Probability Test (Fisher's Method) to operate under both null and alternative hypotheses of a binary test on noisy data, which allows researchers to forecast their ability to screen fused explosion signatures from noise, against source size. Our data show that (1) a fused multi-physics waveform statistic that combines radio, acoustic and seismic waveform data can identify explosions ~ 0.4 magnitude units lower than the most capable, constituent radio emission detector for the same detection probability and (2) that we can quantitively predict how this fused, multi-physics statistic performs with Fisher's Method. This work offers a baseline, quantitative foundation for event detection that supports multi-phenomenological explosion monitoring (multiPEM) that and applicable to any binary hypothesis testing problem in observational geophysics.

Presenting Author: Joshua D. Carmichael

Additional Authors

Session: Leveraging Advanced Detection, Association and Source Characterization in Network Seismology [Poster]

Type: Poster

Date: 4/30/2020

Time: 08:00 AM

Room: Ballroom

Description: Leveraging Advanced Detection, Association and Source Characterization in Network Seismology [Poster]

Small seismic events are abundant and can provide a lot of information on the subsurface. The ability to automatically detect small seismic events in an accurate and efficient way is of great interest for the study of geological structure and fault process. Small seismic events are difficult to detect using conventional methods as the amplitudes of signals are relatively small. Here we present two efficient models: a detector that could accurately extract small seismic signals from continuous records and a locator that determines the geographical position of the extracted event. Both of our models are based on a convolution neural network with a skip-connected structure. Features are learned from spectrograms of signals and then used to classify windows of seismic records. We demonstrate the performance of our algorithms using dataset from Oklahoma, where seismicity has increased dramatically over the last decade. We show that our algorithms are efficient at detecting and locating small seismic events, resulting in a much more comprehensive catalog compared with the Oklahoma Geological Survey (OGS) catalog.

Presenting Author: Xiaofei Ma

Additional Authors

Session: Applications and Technologies in Large-Scale Seismic Analysis [Poster]

Type: Poster

Date: 4/28/2020

Time: 08:00 AM

Room: Ballroom

Description: Applications and Technologies in Large-Scale Seismic Analysis [Poster]

Traditionally, the ability to study seismic phenomena is dependant on both the available hardware and time for processing data needed to produce a research grade catalogue. Consequently, shortages in either of these resources constrain the scope of studies available to the research scientist. This is becoming especially challenging as networks become larger and more dense as the community moves towards Large-N networks and arrays. We will look at alternative solutions to address these resource constraints and open the scientist up to a broader field of study.

Ownership of equipment or waiting in a queue for loan pool assets are the two most common methods for acquiring the hardware necessary to conduct a scientific study. Further, once the data has been collected, a good deal of time is spent processing that data to produce a catalogue before the scientific inquiry can begin.

There is now an alternative model for acquiring and processing data in seismology that shortens the time and effort necessary to produce a research grade catalogue. We will demonstrate how we can customize acquisition arrays to meet experimental goals and apply proven processing models and AI techniques to deliver a bespoke research grade catalogue at a fraction of the time and cost of traditional acquisition and processing methods. This removes several of the challenging aspects of running an experiment in order to enable researchers to get straight to their science and shortening the time to publication.

Presenting Author: Dario Baturan

Additional Authors