Technical Updates: 2012, Second Quarter

Data Sets (Task A)

  • The completion of the ground motion database is a priority for NGA-East. As it is the case with any database development, a number of technical issues have been identified in the last few months during the quality assurance (QA) review. Dr. Goulet (PEER) and Prof. Cramer (CERI, University of Memphis) interact on a weekly basis to address these technical issues and to resolve them.
  • Dr. Goulet developed a re-processing scheme for high priority records (currently defined as horizontal recordings on broadband≥20Hz instruments) that includes appropriate windowing, baseline correction and re-filtering (if needed). The post-processing procedure was tested fully in the development of CENA times series for the finite fault simulation exercise (Task E below). Rules are being developed to automate the process as much as possible, following a large set of sensitivity studies on the effect of post-processing on spectral acceleration values. The project is taking advantage of the knowledge and lessons learned in the NGA-West2 record processing.
  • Because the majority of the database contains ground motions with a frequency band less than 20 Hz (Figure 1), NGA-East is also in the process of developing a relationship to define the natural bandwidth reduction that comes from signal attenuation through natural processes. The intent of this exercise is to identify ground motions recorded in CENA for which the relative narrow band still provides the correct spectral response properties of the recordings. It is expected that this process will allow the inclusion of more recordings into the high quality bin of the data that would have been rejected otherwise due to their apparent narrow band.
  • In addition, Prof. Cramer is finalizing the final deliverables for his contract, which include:
    • A reorganization of flatfile content and definition of static event, station and record numbers
    • The preparation of final report to document the database development process, metadata sources and assumptions
    • The preparation of time series with intermediate processing (instrument-corrected but unfiltered data) for the independent evaluation of filtering bandwidth limits (e.g. corner frequencies)

    Figure 1. NGA-East database, distribution of low-pass filter frequency with epicentral distance for Mw>=4 and Repi<=1000 (~4380 horizontal pairs).

Reference rock and site amplification models (Task B)

  • All the tasks below are being led by the Geotechnical WG (GWG). The group continues to be very active with their tasks and have monthly meetings attended by the TI Lead Dr. Goulet.
  • Reference Rock (B.1): The group has finalized the technical work and the assessment of epistemic uncertainties. A report is expected to be submitted by the Fall for review by the TI team and PPRP.
  • Site Corrections at Recording Stations (B.2): Progress has been made to refine the initial site corrections model. There is a strong interaction with the database QA effort to refine the usable bandwidth for spectral acceleration for this task. The group continues to update the underlying data and procedure for these corrections with a particular focus on constraining Vs30 proxy calculations. A Vs30 proxy based on geology has been developed for the region that is based on large scale geologic maps (Figure 2) and measured Vs30 values (Figure 3). Additionally, the receiver function approach proposed by Ni and Somerville (presentation made at GWG meeting, 2010) has been implemented for sites with recorded ground motion and appears to be promising. Further work is needed to develop the method in a forward sense.

    • Figure 2. Geology classes developed by GWG for site classification, CEUS.

  • Profile Database (part of B.2): work is under way to expand the GWG shear-wave velocity database that will be used to develop idealized soil profiles for the simulations of site response (current stations with Vs data are shown in Figure 3). The collaboration with the independent EPRI project underway is expected to lead to 33 additional site profiles at recording stations. The measurements have been completed and the interpreted data is expected to be available to the Geotech WG in mid-August.

    • Figure 3. Recording stations and location of Vs30 measurements compiled by the GWG.

  • Site Response Analysis Approaches (non-SSHAC): The group continued their development of protocols of the large scale site response simulations that will be conducting including input ground motion and typical soil profiles. They have identified new strain based criteria for defining the threshold for conducting equivalent linear and nonlinear analyses.

Regionalization (Task C)

  • Dr. Walter Mooney (USGS) and his team have mostly completed the compilation of their North American crustal structure database in June adding a total of about 800 points to the original dataset of 1485 (Figure 4). The database includes information gathered from the literature such as shallow and deep profile information, P-wave and S-wave velocity and Q for the CEUS region. The products, currently only used internally, include an Arcmap module for interactive data mining and layered mapping with additional layers for lithology, heat flow, topography and gravity anomaly, to name a few. The group has established a crude initial region definition based on crustal properties and is currently working on further defining regions based on their impact on ground motions. This involves simulations using empirical Green’s functions (eGfs)and looking at response spectral values. This work is coordinated by the TI leads and Prof. Martin Chapman (Virginia Tech) the chair of the Path/Source Working Group. The group meets monthly to monitor progress and redirect tasks to address NGA-East critical issues. This work corresponds to task C.3.

Figure 4. Additional crustal structure data sets added (green and black dot) since Jan2012. Each point corresponds to a profile with associated crustal properties.

Regionalization & Source/Path Studies (Tasks C and D)

Activities from the Path/Source Working Group:

  • The Path Working Group is examining data from recent earthquakes well-recorded by the EarthScope Transportable Array (TA) in the south-central United States. The objective is to identify and to document any significant geographic correlation with known structures of the Gulf coast, Mississippi embayment and central mid-continent. The work aims to quantify Q, geometrical spreading and signal duration in this region of the U.S. The work is closely coordinated with the effort of Dr. Walter Mooney to provide crustal models for the study region. The preliminary work accomplished to date indicates the following (Figure 5):
    • Regions in the Gulf Coastal Plain on thick sediments exhibit strong attenuation of the Lg phase.
    • In comparison to most stations in the Paleozoic Platform, stations in the Great Plains underlain by Cretaceous deposits show larger amplitudes, and those stations in parts eastern Nebraska, Iowa and northwestern Missouri exhibit anomoulsy large Lg amplitudes at frequencies around 1Hz. The latter region is roughly correlated with the mid-continent rift structure.
  • The behavior of the Gulf Coastal region appears consistent with expected wave propagation effects in thick sedimentary sequences. The fact that the effect is large in this case is attributable to the unusually thick sequence of sediments in that particular part of the mid-continent. The work to date suggests that a large component of the attenuation in the gulf coastal area is independent of source-receiver distance, and depends on the thickness of the sediment sequence in the vicinity of the stations.

Figure 5. Plots of receiver terms from regression modeling of Lg Fourier spectral amplitudes at 1.0, 1.4, 2.0 and 2.8 Hz, respectively, from 16 earthquakes that occurred from 2010-2012 in Arkansas, Oklahoma and Texas. Triangles indicate stations with larger than predicted amplitudes, circles indicate stations with smaller than predicted amplitudes.

Additional activities for Path/Source issues:

  • Empirical Green’s functions (eGfs) and source effects: Dr. David Boore (USGS) continued his analysis, in collaboration with Prof. Gail Atkinson (University of Western Ontario), of ground motions from the Saguenay, Riviere-du-Loup, and Val des Bois earthquakes, using eGfs to constrain the stress parameters for each earthquake. The analysis shows that there is a strong azimuthal dependence of the eGf, and this also appears in plots of PSa vs. distance, when grouped by azimuth. This azimuthal dependence complicates the inference of a stress parameter for each earthquake, as the result depends on which group of stations is used.
  • Empirical constraints of effective Q: Dr. Jack Boatwright (USGS) is in the process of documenting the analyses of the M5.8 Mineral earthquake and the set of 17 moderate earthquakes in Northeastern North America.
    • Dr. Boatwright recently obtained MASW site characterizations determined by Dr. Rob Kayen (USGS) for two stations in Virginia that recorded the Mineral main shock, the Reston Fire Station and the Corbin Observatory in Fredricksburg. Dr. Boatwright modeled the 1D site amplification from these site characterizations and obtained a good fit to the observed spectrum for Reston and a partial (good for f > 1 Hz) fit for Corbin.
    • In addition, he has analyzed the recent M3.1 earthquake that occurred in New Brunswick, adding it to the set of moderate CENA earthquakes already analyzed. The earthquake was recorded by a broadband CNSN station (BATG) at 9 km hypocentral distance. Unfortunately, this station was previously identified as exhibiting a resonance at 20 Hz: Figure 6 below compares the source spectrum from station BATG with the average source spectrum obtained from the 11 regional stations (170 < r < 430 km) that recorded this earthquake. The disagreement of the source spectra from 0.1 to 0.7 Hz indicates the relative amplitude of the microseismic noise; the source spectrum for station BATG yields a remarkable fit to the Brune spectrum. The resonant behavior of station BATG is indicated by the dashed line for f > 12 Hz. This comparison eloquently demonstrates the limits of the regional spectral analysis for small earthquakes. The microseismic noise exceeds the source spectral amplitude for regional stations, making the determination of seismic moment uncertain, and the sampling of the broadband instruments (at 40 and 100 sps) is barely sufficient to resolve the corner frequency (10.1 Hz) and the Brune stress drop (8.9 bars). The attenuation obtained from the 12 regional broadband recordings of the New Brunswick earthquake, Q = 527 f 0.4, is very similar to the average attenuation obtained previously for the northeastern part of the Appalachian Province by Boatwright and Seekins (2011), that is, Q = 410 f 0.5 (Figure 6).

Figure 6. Comparison of source spectrum from station BATG with the average source spectrum obtained from 11 regional stations (170 < r < 430 km) that recorded the New Brunswick M3.1 event.

Finite Fault Simulations (Task E)

  • The final-fault simulation validation exercise (E.1-3) has expanded tremendously and is currently underway in collaboration with the Southern California Earthquake Center (SCEC) with funding by sources outside of NGA-East.
    • The validation exercise now includes 23 scenarios from Western U.S., Central and Eastern North America, Japan, New Zealand, China and Turkey with most scenarios tested against a carefully selected subset of 40 recordings.
    • NGA-East is playing a major role in many technical and logistical aspects of the project, applying many of the “lessons learned” from last year’s validation exercise. The TI Leads, TI member Rob Graves and many Simulations Working Group members join weekly calls to make this large scale, fast-paced project a success. Dr. Goulet is leading the coordination efforts between SCEC and PEER.
    • Most of the simulations will be completed on the SCEC BroadBand platform, which is currently being improved to allow instant goodness-of-fit feedback.
    • Additional simulations methodologies are being implemented on the platform while the remaining modelers will participate by providing their results directly to NGA-East.
    • A NGA-East SSHAC Working Meeting is scheduled on December 11, 2012 to discuss the preliminary results from this exercise.
    • In addition to the coordination and liaison role played by Dr. Goulet for NGA-East, PEER is funding Prof. Jonathan Stewart (UCLA) on two tasks: (a) To carry out site corrections of the selected recording sites to provide response spectra at a reference “rock” with Vs30 near 1000 m/s, depending on the specific scenario. This will facilitate the simulation validation efforts as the ground motion simulators will only have to simulate response spectra at the reference rock condition rather than implementing a local site amplification. (b) The second effort by Prof. Stewart is his role as a liaison for PEER to make sure the simulation efforts are in harmony with various other on-going technical tasks including the NGA-West3 project. The funding for Prof. Stewart is coming from PEER’s resources outside of NGA-East.
  • Dr. Frankel (USGS) is working independently, developing broadband synthetic seismograms using his own simulation methodology. He is currently working on the Saguenay and Riviere-du-Loup earthquakes and comparing them with observed recordings. The source model for Saguenay has a random distribution of slip and stress drop along the fault plane. The fault geometry and crustal model of Hartzell et al. (1994) are initially being used. Long-period synthetics are being calculated with the method of Zhu and Rivera (2002). Short-period synthetics are calculated by summing point-source synthetics derived from the stochastic procedure of Boore (1983). Time domain and frequency-domain comparisons will be made between the synthetics and observed records. A key question is what slip velocity to use for the long-period synthetics. The slip velocity should be constrained by the peak amplitudes and pulse widths of the data.
  • Dr. Graves is continuing his modeling of low frequency (f < 1 Hz) wave propagation and attenuation effects for CEUS (excluding Mississippi embayment). The calibration of the 1D velocity and Q structure complements ongoing activities in Tasks C and D (Regionalization and Source/Path Studies) and will provide a useful resource for finite-fault simulation of scenario earthquakes. In addition, the comparison of the observed DYFI intensities for recent earthquakes (Mt. Carmel, IL and Mineral VA) with those estimated from waveform simulations provide a framework for re-examining the 1811-1812 New Madrid earthquakes using finite-fault rupture simulations. The current phase of work is focused on the ground motions recorded during the 2011 Mineral, VA earthquake. High quality broadband data are available for 89 sites covering the distance range 233 to 1110 km. A preliminary analysis indicates that the 1D crustal velocity and attenuation model developed for the 2008 Mt. Carmel, IL earthquake does well at reproducing the median level observed spectral accelerations at periods of 2, 3 and 5 seconds for the Mineral, VA event. Summary points:
    • Waveform simulations using a calibrated 1D velocity and Q model are able to reproduce the observed low frequency (f < 1 Hz) ground motion levels for the Mw 5.23 Mt. Carmel earthquake at distances out to about 850 km.
    • Key characteristics of the observed motions including flat attenuation between 70 and 150 km, and generally low attenuation with distance, are captured by the simulation model.
    • Simulated MMI values obtained by the conversion relations of Atkinson and Kaka (2007) provide a better overall match to the observed intensities than those obtained using the Dangkua and Cramer (2011) conversion relation.

Sigma (Standard Deviation) (Task J)

Uncertainty from simulated ground motions (J.1):

  • Dr. Paul Spudich (USGS) is working on the quantification of uncertainty of ground motions from simulations. He is continuing on a previous task funded by the NRC and is now focusing on a review of the error-estimation theory of Yagi and Fukuhata (2011). In their paper, the authors introduce an error estimate by modifying the data covariance matrix, but their method does not take into account the fact that Green’s function error increases with time into the Green’s functions. Dr. Spudich is investigating how this work can be integrated in his uncertainty assessment. This task also has strong ties with task E.6-7.

Sigma and single-station sigma using NGA West-2 dataset (J.2):

  • Preliminary data exploration was performed using the SMM and LM NGA West-2 datasets. The 2 datasets were combined to create a single flatfile that identifies repeatable stations.
  • Preliminary regression analyses were performed with respect to a simple functional form based on the Campbell and Bozorgnia (2008) model and single-station within-event residuals were calculated. Trends of the single-station within-event residuals with magnitude, distance, Vs30, and number of recordings per station were examined and compared to ergodic within-event residuals of the interim NGA West-2 Abrahamson & Silva model.
  • A progress meeting of the single-station sigma working group is scheduled for August 10, 2012. Before the working meeting, residual analyses will be repeated using the new version of the SMM and LM flatfiles and preliminary results will be presented during the progress meeting to get feedback from other members of the working group.

References

    Atkinson, G. M. and Kaka, S. I. (2007). Relationships between Felt Intensity and Instrumental Ground Motion in the Central United States and California, Bull. Seism. Soc. Am., 97: 497-510.

    Boatwright, J., and L. Seekins (2011). Regional spectral analysis of three moderate earthquakes in northeastern North America. Bull. Seism. Soc. Am., 101: 1,769–1,782.

    Boore, D. (1983). Stochastic simulation of high-frequency ground motions based on seismological models of the radiated spectra, Bull Seism. Soc. Am., 73: 1865-1894.3

    Boore, D. M. (2003). Prediction of ground motion using the stochastic method. Pure and Applied Geophysics, 160: 635–676.

    Campbell, K. W. and Y. Bozorgnia (2008). NGA ground motion model for the geometric mean horizontal component of PGA, PGV, PGD and 5% damped linear elastic response spectra for periods ranging from 0.01 to 10s, Earthquake Spectra, 24: 139-171.

    Dangkua, D.T., and C.H. Cramer, 2011, Felt Area versus Instrumental Ground Motion: A Difference between California and Eastern North America?, Bull. Seism. Soc. Am., 101: 1847-1858.

    Hartzell, S., C. Langer, and C. Mendoza (1994). Rupture histories of eastern North American earthquakes, Bull. Seism. Soc. Am., 84: 1703-1724.

    Yagi, Y. and Fukahata, Y. (2011), Introduction of uncertainty of Green’s function into waveform inversion for seismic source processes. Geophys. J. Int., 186: 711–720.

    Zhu, L. and Rivera, L. A. (2002), A note on the dynamic and static displacements from a point source in multilayered media. Geophys. J. Int., 148: 619–627.