Technical Updates: 2014, First Quarter

Combined efforts

The TI team held working meetings (WMs) on January 27 and 28 to discuss the results from two main tasks to be finalized and documented in PEER reports: database and regionalization. Draft report sections were made available to the TI team prior to the meetings. The goals for these WM were to obtain the TI team’s feedback on the completed work and to provide comments for the completion of the documentation. Both meetings were attended by approximately 20 in-person participants and about the same number of on-line participants, most of the latter generally acting as observers (Figure 1). The PPRP members participated online with pertinent questions and provided input also as resource experts when needed. The WMs were concluded with a list of action items and task assignments to key NGA-East participants to finalize the work. More details are provided below in under each technical task. The PPRP provided an informal list of comments to be considered by the TI team.

Figure 1. NGA-East Regionalization Working Meeting, January 27, 2014 at PEER Headquarters.

Drs. Atkinson (UWO) and Goulet (PEER) and Profs. Chapman (VA Tech), Cramer (U. Memphis) and Pezeshk (U. Memphis) were invited to participate in the “Central and Eastern U.S. Seismic Hazard Workshop” held at CERI in Memphis, in late February 2014. They gave presentations on the current and future work for NGA-East and related topics.
Several NGA-East participants were invited to the “Southern Western U.S.” utilities (SWUS) SSHAC Workshop 3 (March 2014). The technical developments of certain methodologies and approaches are tied to NGA-East and is funded by the SWUS project as part of the cost-sharing strategies developed last year.

Data Sets (Task A)

The January 28, 2014 WM was dedicated to the broad topic of Database. The work from the large number of participants involved in the development of the database was presented and reviewed. The key topics covered included: data collection and pre-processing, data reprocessing using updated PEER protocols developed through efforts from NGA-West2 and NGA-East, metadata collection and computation including: appropriate magnitude, distances, site characterization, aftershock identification and source/site region assignment. One additional very important addition to the database metadata is the evaluation of Kappa.
Numerous technical tasks have been initiated following the January 28 meeting:

Figure 2. A single figure with 12 plots summarizes the key components used in the QA process. The three columns correspond to each possible component. The first row shows the original instrument-corrected acceleration traces. The second row shows the processed acceleration time seriesThe displacement is computed by numerical integration and plotted on the third row, also in blue. This allows to check that the final processed data have self-consistent. Finally the Fourier amplitudes retrieved from the processing products are plotted in black (original), blue (processed) and the amplitudes for the noise window is shown in red. A black solid line is anchored at the reported HP corner frequency and a green dashed line is anchored at the LP corner frequency. The microseism window, if applicable, is bounded by a peach lines.

A preliminary flatfile (for internal NGA-East review only) was released. The purpose of this internal distribution is to find any inconsistencies in the flatfile before its release to a larger audience.

Reference rock and site amplification models (Task B)

The Geotechnical Working Group (GWG) continues to meet monthly.
Their paper on reference rock was published: Hashash, Y.M.A, Kottke, A.R., Stewart, J.P., Campbell, K.W., Kim, B., Moss, C., Nikolaou, S., Rathje, E.M, and Silva, W.J. (2014). Reference Rock Site Condition for Central and Eastern North America, Bull. Seis. Soc. Am, Vol. 104, No. 2, pp. 684–701.
Mr. Harmon (PhD Student, UI Urbana-Champaign) presented the group’s station database and preliminary preferred Vs30 summary at the January 28 WM. The group has been updating their work based on the TI team comments and in light of the new information collected (see Database tasks).
The group ran a subset of site response simulations to share with the TI leads (Fourier and response spectra).
The group is further refining the geology-based representative velocity profiles to be used in the large scale ground response simulations.

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

The January 27 WM was dedicated to Regionalization. The work from two teams was present to the TI team: Dr. Mooney (USGS) and Prof. Chapman (VA Tech). Presentations included those on crustal structure, seismic velocity and Q models as well as results from simulations. The crustal data provided the basis for an initial proposal to divide CENA into four regions: Atlantic Coastal Plain, Appalachians, Central US, and Gulf Coast Province. The simulation results showed that the regions could potentially be combined into two regions: the Gulf Coast and everything else in CENA. Additional presentations were also made. Prof. Chapman showed a summary of azimuthal variability of ground motions from his simulations. Dr. Boatwright showed a summary of his work on depth-dependent stress parameter (see below).
Several tasks have been initiated following the January 27 meeting:
- Numerous exchanges involving the TI leads and Dr. Mooney’s team were aimed to better constrain the statistics on variation of ground motions across regions.
- Follow-up calls and web meetings were help by Dr. Goulet and both teams (USGS and VA Tech) to ensure that the updated products were as needed.
- Dr. Mooney is finalizing the report which was updated by his team following the recommendations from the January 27th working meeting.
- Prof. Chapman’s team has rerun a subset of computations.
- Prof. Chapman has integrated the codes developed by Dr. Mooney’s team for the post-processing of simulations data. He is working on his report.
- Dr. Hollenback (PEER post-doctoral research fellow) conducted a detailed literature review and compiled a list of attenuation models for CENA.
Dr. Boatwright re-evaluated moments and stress drops for the complete set of CENA earthquakes that have been analyzed, which now includes 71 events reaching back to 1986. This effort was summarized at the PEER NGA-East WM in January and updated over the following weeks (Figure 3 ). The earthquakes in the Charlevoix and Lower St Lawrence region have the highest stress drops, and these stress drops correlate (r = 0.75) with depth. The earthquakes in upstate New York and the Western Quebec Seismic Zone have somewhat lower stress drops and no correlation (r = 0.02) of stress drop on depth.

Figure 3. The stress drops of these earthquakes are plotted below as a function of source depth. The line shows the linear regression of stress drop on depth, which increases from 20 bars at 1 km to 500 bars at 30 km depth for the Eastern Quebec earthquakes. In contrast, the stress drops for the earthquakes in Western Quebec show no dependence on source depth. What is striking about the stress drops of the Eastern Quebec earthquakes is the cluster of large stress drops for earthquakes h > 20 km. These large stress drops argue that the “anomalous” stress drop of the 1988 Saguenay earthquake is determined by its source depth, not its source size. Seven earthquakes, ranging in size from M2.5 to M4.2 and in depth from 14 to 27 km, have stress drops that equal or exceed the stress drop of the Saguenay earthquake. We note that the stress drop estimates for the small earthquakes are uncertain, owing to the limited frequency band (f < 40 Hz) of the recordings.

Finite Fault and Stochastic Simulations (Tasks E and F)

The collaborative simulations group (PEER, SWUS and SCEC) kept very active in the last few months and continued to have their weekly phone calls.
- Figure 4. Observed Sa (1 sec) plotted as a function of Rjb distance. The purple symbols indicate St. Lawrence sites. The level of amplification at the St. Lawrence sites is about 4-5 relative to the closer in locations. It is very difficult to explain this solely with source characterization. Simulations results are shown for alternate fault geometries (8×3 and 3.7×3.7) km2. The 8×3 simulation does somewhat better in capturing the distance dependence of the observations, but results still noticeably underpredicts ground motions at the St. Lawrence sites.
A new pair of models by Drs. Silva and Darragh (finite-fault and point-source stochastic) have been added to the methods to be evaluated. The validation is being completed outside of the BBP, but following the same input and rules.
Dr. Frankel is continuing to make broadband synthetic seismograms for ENA earthquakes and comparing their spectral accelerations with those for the western U.S. Figure 5 shows the 0.2, 1.0 and 3.0 second SA values for M6.5 earthquakes for some of the ENA simulations for a hard-rock site condition (Vs30= 2800 m/s), compared to those from the WNA simulations for a firm-rock site condition (Vs30= 620 m/s). For this example, the simulations were run to 200 km distance; I am working on extending the ENA simulations to 1000 km. He is extending the distance range of the ENA synthetics out to 1000 km. He is examining how the slip velocity (for long-period deterministic synthetics) and the stress drop (for short-period stochastic synthetics) affects the spectral accelerations. Dr. Frankel also is investigating how focal mechanism affects the rate of decay of ground motions using the broadband synthetics he has calculated for ENA earthquakes. The radiation pattern affects the low-frequency portion of the ground motions determined from the reflectivity code. Observations from the Charlevoix earthquakes indicate that radiation pattern strongly alters the distance decay for frequencies of 1 Hz. This is also found in the synthetic ground motions for two Charlevoix earthquakes.

Figure 5. Spectral acceleration values (SA) for M6.5 earthquakes using broadband synthetics for Eastern North America (ENA; hard-rock site condition, Vs30= 2800 m/s) and Western North American (WNA; firm-rock site condition, Vs30= 620 m/s). The ENA values have been offset to the right for easier viewing. The 1.0 and 3.0 s values are controlled by the deterministic synthetics, whereas the 0.2 s values are controlled by the stochastic synthetics. The deterministic and stochastic synthetics are combined with a crossover frequency of 2 Hz, for these runs. A stress drop of 200 bars was used for ENA and 100 bars for WNA.

Prof. Atkinson and her student Hassani have a paper in preparation on the use of referenced empirical method to understand differences between NGA-East database and NGA-West2 GMPEs, to shed further light on constraints on point-source (and finite-fault) parameters for ENA. Early conclusion is that ENA and WNA motions are similar at R<70km, except at high frequencies (>5 Hz and PGA). Motions in southeastern Canada and central U.S. are similar, but different for the Gulf Coast region.
Prof. Atkinson is also working with graduate students Yenier and Singh on more specific use of point-source methods to model ENA ground motions. The aim is to eventually merge this work/understanding with the finite-fault simulations.
Dr. Boore submitted a paper (coauthored with Dr. Thompson, SDSU) to BSSA for publication: “Path Durations for Use in the Stochastic-Method Simulation of Ground Motions “ While the application was to data from the NGA-West2 project, the methodology might be applicable to data from the NGA-East project. A summary of findings was shared with the NGA RVT technical group (see GMPE tasks).
Dr. Boore also finalized a draft paper with Drs. Di Alessandro (Geopentech) and Abrahamson (PG&E) as coauthors: “A Generalization of the Double-Corner-Frequency Source Spectral Model and Its Use in the SCEC BBP Validation Exercise”. The paper presents two generalizations of a double-corner frequency source spectral model, intended for use in stochastic simulations of ground motions. The data used to illustrate the use of the models are from PEER’s database for active crustal regions, but the source models are relevant for the NGA-East project, because the GMPEs for that project will be developed using stochastic method simulations.

Development of Median Ground Motion Prediction Equations (Task H)

Dr. Bozorgnia (PEER) assembled the NGA Random Vibration Theory (RVT) technical group to address issues in RVT. The group consists of Drs. Boore (USGS), Goulet (PEER), Kishida (PEER), Kottke (Bechtel), Kuehn (PEER), Silva (PEA) and Bozorgnia (PEER) as well as Profs Der Kiureghian (UC Berkeley) and Rathje (UT Austin).
- The group had a teleconference followed by an in-person meeting to devise a clear plan for key tasks:
  - Appropriate choice of spectrum for development of GMPEs with RVT: Fourier Amplitude Spectra or Power Spectral Density and choice of component (single horizontal, vector sum, etc.).
  - Appropriate duration to apply to go from frequency domain to PSA space; development of a duration model.
  - Appropriate choice of frequencies (e.g. downsampling) and level of smoothing to use.
  - Drs. Boore, Jottke and Kuehn prepared materials which they presented at the in-person meeting.
  - Follow-up work has been shared with the group by Profs. Rathje and Der Kiureghian and Dr. Boore.

Evaluation of Epistemic Uncertainty(Task K)

Dr. Kuehn (PEER) is refining the Sammon’s maps visualization tool. This is currently supported by SWUS. New developments for NGA-East will include the consideration of simulated ground motions.