Abstract: The Collaboratory for the Study of Earthquake Predictability (CSEP) is global platform for conducting prospective earthquake forecasting and prediction experiments. Since its inception in 2006 in California, CSEP has grown to encompass four testing centers around the world, and is now prospectively evaluating over four hundred models in testing regions in California, Italy, New Zealand, Japan, the western Pacific and at global scale. Here, we review a decade of CSEP achievements, and state our priorities for future activities. Achievements encompass new scientific insights into earthquakes and their predictability, progress in evaluation methodology, and broader impacts in seismic hazard assessments and risk reduction strategies. Scientific highlights include: (1) the numerous small earthquakes provide information about future moderate-to-large earthquakes; (2) geodetic strain rate models capture earthquake potential, and, when coupled with smoothed seismicity models, provide the most informative forecasts; (3) a new generation of physics-based Coulomb/rate-state models are now able to compete with statistical models in forecasting the space-time evolution of earthquake sequences;. More broadly, CSEP has highlighted the benefits and scientific importance of prospective and independent testing to establish credible benchmarks of the forecast skill of competing hypotheses and models. These and other CSEP results have effected changes to several earthquake source models of official seismic hazard models, including in California, New Zealand and Italy. In this way, CSEP has contributed to safer and better-informed societies. Future activities will be guided by three main objectives: (1) Improve the discrimination capability of forecast testing by expanding the spatial and temporal distribution of earthquake data. (2) Develop procedures and requisite cyberinfrastructure for testing earthquake forecasts worldwide, focusing on new types of earthquake forecasts. (3) Test key hypotheses that underlie earthquake forecasting models.
Slidecast:
https://vimeo.com/277546924
Abstract: Spatial and temporal distributions of aftershocks were studied for recent moderate earthquakes that occurred at shallow depth onshore of Japan and were well recorded by the regional networks. These events include the 2000 Western Tottori (Mw 6.7), 2004 Niigata Chuetsu (Mw 6.6), 2005 Fukuoka (Mw 6.6), 2007 Noto Peninsula (Mw 6.7), 2007 Niigata Chuetsu-oki (Mw 6.8), 2008 Iwate-Miyagi-ken (Mw 6.8), 2016 Kumamoto (Mw6.2) and 2017 Tottori (Mw6.2) earthquakes. All of these earthquakes are approximately of similar size, however, the rates of aftershock activity are quite different. The 2004 Niigata and 2008 Iwate-Miyagi earthquakes have significantly more aftershocks than the other 7 events. In the spatial locations of the aftershocks, these two earthquakes have more complex spatial distributions with more aftershocks occurring away from the mainshock fault plane. There appears to be a correlation between the rate of aftershock activity and the spatial complexity of the locations. The sequences with higher rates of aftershock occurrence may be associated with aftershocks triggered in a volume around the mainshock. In contrast, for the other sequences, aftershocks occur mainly in a planar pattern close to the mainshock fault plane. The early time sequences of the aftershocks for these events were also examined. Using continuously recorded seismograms from nearby borehole stations of Hi-net, aftershocks were identified and counted. From about one minute following the mainshock origin time, we estimate that we can identify aftershocks with magnitudes down to Mj 3.5. For the first few minutes the rate of aftershocks is quite similar for all of the mainshocks. The higher rate of aftershocks for the 2004 Niigata and 2008 Iwate-Miyagi earthquakes appears to begin about 10 minutes after the mainshock. This suggests that the enhanced triggering of aftershock for these two earthquakes may be caused by some changes in the aftershock region several minutes after the mainshock.
Slidecast:
https://vimeo.com/277548143
Abstract: The current deployment of the EarthScope Transportable Array (TA) in Alaska affords an unprecedented opportunity to study explosive volcanic eruptions using a relatively dense regional seismo-acoustic network. Active volcanism in the Aleutian Arc poses a risk to both regional and international air traffic. Infrasound monitoring has demonstrated utility for the detection and characterization of explosive volcanism, but previous studies have utilized relatively sparse networks of infrasound arrays in comparison to the TA in Alaska (which uses single-sensor stations). Here we present capabilities for the detection, location, and characterization of remote explosive volcanic eruptions using seismic, infrasonic, and ground-coupled airwave phases. We combine data from the TA and additional regional networks, including data from the Alaska Volcano Observatory (AVO) and Alaska Earthquake Center (AEC). We implement a Reverse Time Migration (RTM) technique to locate explosive eruptions in Alaska, with a focus on the recent explosive activity at locally-unmonitored Bogoslof volcano (December 2016 – August 2017). More than 60 eruptive events from Bogoslof provide a unique validation dataset, allowing experimentation and optimization of different RTM strategies. We also apply RTM to eruptions from other Alaskan volcanoes (Cleveland, Pavlof) and Kamchatkan volcanoes (Bezymianny, Shiveluch). We are experimenting with different strategies and parameter choices for the RTM; challenges include varying signal durations and amplitudes, the source-receiver geometries, and most volcanic eruptions occurring outside the network. We employ Receiver Operating Characteristic (ROC) curves to characterize parameter choices, and investigate coherence weighting of infrasound for signal cleaning and selection. Our methods are useful for both (1) event detection using real-time data and (2) scanning data archives to identify and discriminate volcanic and non-volcanic events.
Poster:
WED.Brickell.1700.Sanderson
Abstract: Path-specific ground-motion models, or fully non-ergodic ground-motion prediction equations (GMPEs) are the next step in improving ground-motion estimation, in particular for critical facilities where design targets fall at very low probabilities of exceedance. Statistical approaches in regions with high seismicity rates have been shown to be robust methods for moving towards fully non-ergodic models, however in seismically quiet regions, these methods will not reduce uncertainty in path terms; we work to include known, regional properties into GMPEs to reduce uncertainty. These results are then included as local path term adjustments, to demonstrate the GMPE’s uncertainty reduction. With a relationship between crustal properties and ground-motion amplitudes, obtainable physical properties which remain constant on human timescales may be included into GMPEs as path effects. We test our method in the seismically active Southern California region, which hosts a myriad of studies on crustal properties (seismic velocity, attenuation, fault and geologic mapping). With dense station coverage from four networks, we compute PGA for over 130,000 earthquake recordings (0.5 < M < 4.5, 5 < Rrup < 180km). We develop a regional GMPE with a mixed effects approach, and simultaneously obtain event, site, and path residuals, as well as 3D raypaths for each recording. Next, we compute values of the velocity gradient along each path to compare to each recording’s path residual. We observe a moderate negative correlation between the path residual and that path’s gradient of velocity, implying a relationship between amplitude and structural heterogeneity. In addition, we compute a path tomography model, or 3D volume with the mean path residual in each grid cell. We compare this to a volume of crustal velocity (Vs), and find that for velocity alone there is no correlation. Finally, we repeat the analysis with a Q model to investigate the effects of attenuation on path residuals.
Slidecast:
https://vimeo.com/277547279
Abstract: The search for key elements characterizing the earthquake rupture process is challenged by the specificities of each individual event. This results in a large diversity when looking at earthquakes as a whole. This diversity is well documented by the moment rate functions (or Source Time Functions – STF), one of the most robust seismological observables of the rupture process. Teleseismic STFs also have the potential to be automately extracted for each earthquake with magnitude above 5.7-6, since the development of the digital global broadband seismic networks. This potential access to thousands of STFs, in all earthquake contexts and depths, motivated the development of the SCARDEC method, which simultaneously retrieves the static source parameters (depth, focal mechanism and magnitude) together with the STF. More precisely, the SCARDEC method retrieves apparent STFs (for each location and P/S phase), which also offers the possibility to track first-order features of the space-time rupture process. The present study uses these STFs to further document how, and at which velocity, rupture develops. In a first step, we will review the information provided by a systematic search of the average rupture velocities. Second, we will focus on the most energetic phase of the STFs. Even if the times at which rupture strongly accelerates appears unpredictable (resulting in very different STF shapes), the characteristics of this acceleration provide constraints on the dynamics of the rupture. As a matter of fact, this acceleration is on average faster than what is predicted by a classical self-similar rupture growth (where the STF grows quadratically with time). This shows that during the main phase, rupture velocity and/or slip rate increase.
Poster:
WED.Brickell.1430.Vallée
Abstract: The Oklahoma Geological Survey (OGS) monitors seismicity throughout the state of Oklahoma utilizing permanent and temporary seismometers installed by OGS and other agencies, while maintaining an earthquake catalog. In Oklahoma, prior to 2009 background seismicity rates were about 2 M3.0+ earthquakes per year, which increased to 579 and 903 M3.0+ earthquakes in 2014 and 2015, respectively. The peak in the seismicity rate has since fallen to 624 and 304 M3.0+ earthquakes in 2016 and 2017, respectively. The catalog is complete down to M2.4 from mid-2014 to present, despite the significant workload for a primarily state-funded regional network. Unique challenges associated with being the de-facto earthquake information source include providing regulators, i.e. the Oklahoma Corporation Commission, with earthquake locations and magnitude within minutes of an event so that their “traffic-light” protocol may be effectively applied. We are transitioning to running our information center solely on virtual machines (or “cloud-based” computers), which have the added advantage that memory, CPU, and hard-disk resources may be added a la carte and on-the-fly. In addition, we have begun a citizen-scientist driven, educational seismometer program by installing Raspberry Shake geophones throughout the state at local schools. Educational aspects of that program include teacher-driven curriculum development facilitated during professional development workshops for teachers. We will connect the Raspberry Shakes to our earthquake information center, and utilize the data for earthquake locations and research in areas of the state with sparse broadband or short-period seismographs. The future seismic hazard of the state portends a continued need for expansion and densification of seismic monitoring throughout Oklahoma.
Slidecast:
https://vimeo.com/277555760
