By Charles A. Langston (University of Memphis)

Seismic wave propagation for Earth structures that contain thick sections of near-surface, laterally continuous, unconsolidated sediment does not necessarily conform to the standard source-path-receiver convolutional model that is often assumed in earthquake shaking hazards assessments, particularly for earthquake sources within the region covered by the sediments. In theory, earthquakes that occur within the Mississippi embayment or Gulf of Mexico/Atlantic Coastal Plains may excite large, relatively high frequency surface waves in addition to large S waves that can contribute to strong ground shaking. Excitation of these surface waves may be quite high in amplitude if faulting extends into the sediments. In practice, there have been no large earthquakes recorded in these regions to test this assertion although there are hints that this occurs from shallow, low magnitude seismicity. In addition to the velocity structure of the sediments and impedance contrast at the sediment-basement interface, the anelastic attenuation structure will also be an important factor in controlling the frequency dependence of amplification, excitation, and propagation of both S and surface waves. Conversely, the velocity, attenuation, and nonlinear wave propagation characteristics of thick sediments can be studied experimentally using surface waves from earthquakes, explosions, high amplitude controlled sources and surface waves of the ambient noise field. A short review of past work in the Mississippi embayment, use of horizontal-to-vertical ambient noise, teleseismic receiver functions, explosion sources, controlled sources, and earthquake waveform modeling will be presented. Proposals for future instrumentation facilities and field experiments will also be suggested to address the problem of sediment seismic response.

By Vladimir Graizer (U.S. Nuclear Regulatory Commission)

The deep sediments typically found in basins can significantly amplify ground motions. I implemented a deep sediment thickness effect in ground motion models (GMM) for the 5 percent-damped spectral accelerations (SA) based on the depth to the shear-wave velocity horizon of 2.5 kilometers per second (km/s) (Z_2.5) for the Graizer (2018) GMM. In most cases, the actual basin depth under the seismic station is not known and is usually substituted in GMM by the Z-term: depth to the shear-wave velocity horizon of 1.0 km/s (Z_1.0), 1.5 km/s (Z_1.5), or 2.5 km/s (Z_2.5). I do not recommend using the term “basin effect” in application to current GMM, instead the term “deep sediments effect” is a more appropriate. This can help avoid confusion and misunderstanding when comparing to a basin effect in seismology because (1) none of the Z-terms represent actual basin depths, (2) the basin-edge effect is not considered, and (3) the actual shape of the basin is not considered.

Anelastic attenuation of SA is another issue that needs to be clarified. Apparent attenuation of SA quality factor, Q_SA, is different from the classical seismological Q determined from Fourier spectra of S, Lg, or coda waves. From a physical point of view, the Q_SA quality factor represents the diminution of amplitude in the 5 percent-damped response spectral amplitudes at a certain frequency not necessarily associated with one wave. The seismological Q-factor represents the fraction of energy lost per cycle in a specific wave.

By Youssef Hashash (University of Illinois at Urbana-Champaign), Okan Ilhan (University of Illinois at Urbana-Champaign), Halil Uysal (University of Illinois at Urbana-Champaign), Jonathan P. Stewart (University of California at Los Angeles), Ellen M. Rathje (University of Texas), and Sissy Nikolaou (Williams Sale Partnership Limited—WSP)

Site amplification factors are used to compute a ground motion intensity measure of interest at a surface condition through adjustment of the ground motion intensity measure at the reference condition. Central and Eastern North America, which is known as a stable continental region, but with significant seismic hazard, lacks region-specific site amplification functions due to a dearth of ground motion recordings for empirical characterization of site amplification. As part of NGA-East geotechnical working group, we performed a large-scale parametric study of over 1.7 million linear, equivalent-linear, and nonlinear, one dimensional (1-D) site response analyses to represent variability in site conditions in Central and Eastern North America (Harmon and others, 2019a, b). These simulations were used to propose a suite of conventional simulation-based site amplification functions, which were adapted in the development of semi-empirical linear (Stewart and others, 2020) and simulation-based nonlinear amplification (Hashash and others, 2020) models for National Seismic Hazard Maps applications (Petersen and others, 2020). Recently, this parametric study design is further updated (over 3.6 million linear, equivalent-linear, and nonlinear analyses), and an alternative modeling approach using artificial neural network deep learning techniques is adopted to better capture the simulated amplification dataset (Ilhan, 2020). These studies highlight the need to go beyond the time-averaged shear-wave velocity within the top 30 meters (m) of the Earth’s surface, V_S30, as an input proxy for site amplification and as a minimum, include site natural period to significantly enhance the fidelity of conventional and artificial neural network based ergodic site amplification models.

By Seth Carpenter (Kentucky Geological Survey), Zhenming Wang (Kentucky Geological Survey), and Ed Woolery (University of Kentucky)

Numerous proxies have been proposed and used to account for site response, that is, modification of ground motion by near-surface, low-velocity materials in terms of spectral content, amplitude, and duration. Some proxies are based on portions of a site’s velocity profile, including the time-average shear-wave velocity from the surface to a particular depth or the depth to a particular velocity. Others are based on the surface geology or thickness of unlithified sediments over bedrock. Empirical and theoretical studies have shown that site response can be quantified by the fundamental (f₀) and peak site frequencies (f_peak) and the corresponding amplifications, A₀ and A_peak (Carpenter and others, 2020; Wang and Carpenter, 2021a, 2021b; Zhu and others, 2021). We estimated fundamental and peak weak-motion response parameters from empirical and theoretical 1-D site-responses in the thick sediments (from 100 m to approximately 850 m below the land surface) of the upper Mississippi embayment at the two deep vertical seismic arrays and at eight other seismic stations where velocity profiles to bedrock are available. We then compared the fundamental and peak site frequencies and the corresponding amplifications with those proxies. Our results show that the fundamental site period and peak amplification do not simultaneously correlate with velocity- and thickness-based proxies. For example, although f₀ correlates strongly with depth to bedrock, A₀ does not. Further, there is no correlation between f_peak, A_peak, and sediment thickness. Likewise, the average S-wave velocity in the upper 30 m does not correlate with f₀. Our results suggest that a single proxy cannot capture site response, and reliably predicting linear site response relies on site-specific characterizations. We also found that f₀ and A₀ can be estimated by single-station weak-motion S-wave horizontal-to-vertical spectral ratios (H/V) to within 8 percent and 18 percent on average, respectively. Thus, H/V may provide useful approximations of linear site responses in the embayment.

By Erin Cunningham (University of Memphis)

The thickness and seismic velocities of sedimentary sequences strongly affect their response during earthquakes and can prolong and amplify ground motions. We characterize the shallow structure of Atlantic Coastal Plain sediments using a passive‐seismic approach based on high‐frequency P‐to‐S receiver functions. Using sediment S‐wave reverberations and P‐to‐S phase arrival times measured directly from the receiver functions, we invert for average S‐ and P‐wave velocity (V_S, V_P) profiles of the Atlantic Coastal Plain sedimentary strata. We find that V_S increases with depth following a power‐law relationship, whereas the increase of V_P with depth is more difficult to constrain using converted wave methods. Finally, we use the variation of measured S‐reverberation amplitudes with depth to validate these velocity profiles.

By Chris H. Cramer (University of Memphis)

Studies of seismic crustal attenuation in the Central and Eastern United States have mainly identified two different attenuation regions: mid-continental and Gulf Coast. The boundaries between regions have generally been defined using criteria other than seismic attenuation observations due to limited observations of crustal quality factor (Q). The USArray (IRIS Transportable Array, 2003) provided sufficient station coverage and data to define the actual boundaries of the Gulf Coast crustal attenuation region. We have shown that the Gulf Coast region does not include the upper Mississippi embayment, northern and central Alabama, Georgia, most of Florida, and does not extend under the Atlantic Coastal Plain, but does extend into western Mississippi, southern Arkansas, east Texas, and parts of Oklahoma.

There are also suggestions of higher attenuation under basins and rifts northwest of the Gulf Coast region. Gulf Coast regional crustal Q has been observed between 0.1 and 10 hertz (Hz) as Q_o at 1 Hz = 259 +24/−22 and the power of frequency (estimated time of arrival) = 0.715 ± 0.013. Intensity observations from historical earthquake magnitude (M) 7 and recent M5 earthquakes are consistent with our crustal Q boundary observations, which provides greater support for our model compared to other recent crustal attenuation models. The ground motions generated by the August 9, 2020, M5.1 earthquake in Sparta, North Carolina, indicate an Atlantic Coastal Plain crustal Q that is similar to the mid-continental Q. The eastern United States attenuation anisotropy observed in the 2011 M5.7 earthquake in Mineral, Virginia, is also observed in the 2020 Sparta, North Carolina earthquake, but with the added observation of lower attenuation to the east toward the Atlantic Coastal Plain.

By Charles Mueller (USGS)

A method for estimating and mapping seismic site response over a region is applied to the United States Atlantic and Gulf Coastal Plains. The geologic structures that control the site response are relatively simple in the study area, where 0–15 km of poorly consolidated Cretaceous Period and younger sediments overlie bedrock. At each site on a grid, a spectrum of Fourier-spectral site amplification factors is computed, using power-law V_S-depth and shear-wave quality factor (Q_S) depth models tied to surface geology and depth to bedrock, and applying a quarter-wavelength procedure that accounts for both amplification and attenuation. These factors are applied with a stochastic methodology to simulate linear site response for peak ground acceleration and 1 Hz, 5 Hz, and 10 Hz spectral response. Results are presented as maps of site response factors for selected earthquake magnitudes and distances. Patterns show complex tradeoffs between surface geology and depth to bedrock. Peak ground acceleration for small source distances is amplified by factors of 2–3 near landward edges of the Atlantic and Gulf Coastal Plains, and amplifications are reduced by factors up to 2 in the lower Mississippi River valley. Peak ground acceleration for large source distances is amplified everywhere, up to factors of 2–3 across broad areas of the Mississippi embayment and Atlantic Coastal Plain. Response for 1 Hz, 5 Hz, and 10 Hz is generally amplified at sites with young sediments or shallow bedrock. High-frequency motions are reduced in the lower Mississippi River valley; up to factors of 5 at the delta where sediments are thickest. Efforts to model nonlinear site response were hampered by difficulties in modeling sediment modulus and damping effects.

DISCLAIMER: This information is preliminary or provisional and is subject to revision. It is being provided to meet the need for timely best science. The information has not received final approval by the U.S. Geological Survey (USGS) and is provided on the condition that neither the USGS nor the U.S. Government shall be held liable for any damages resulting from the authorized or unauthorized use of the information.

By Thomas L. Pratt (USGS)

We examine the effect that the Atlantic Coastal Plain strata have on ground motions in the eastern and southeastern United States. At 217 sites on Atlantic Coastal Plain strata with thicknesses up to nearly 4,000 m, we compute spectral ratios relative to the average of 4 bedrock sites west or northwest of the strata. Results show prominent resonance peaks that define the largest amplifications at specific thicknesses for each frequency (Pratt and Schleicher, 2021). As frequencies increase, the resonance peaks migrate to thinner Atlantic Coastal Plain strata and increase in amplitude. We develop simple equations to define the approximate shape of the amplifications versus Atlantic Coastal Plain thickness. Comparison with other corrections based on sediment thickness show similarities on thin Atlantic Coastal Plain strata but divergence on thicker sediment.

We also test the reliability of H/V spectral ratios using ambient noise and earthquake arrivals to estimate the fundamental resonance peak of the Atlantic Coastal Plain strata, compared to spectral ratios relative to bedrock. We find that H/V ratios accurately define the frequency of the fundamental resonance peaks but consistently underestimate the amplitudes. Correcting the H/V ratios by multiplication with the vertical to horizontal spectral ratio of bedrock sites to partially correct for the source P-wave to S-wave amplitude ratio brings the H/V ratios more in line with bedrock spectral ratios. After this correction, the bedrock spectral ratios are about a factor of 1.5 greater than the H/V ratios, although scatter is wide in the data. We also find that H/V ratios using ambient noise are remarkably consistent during one-half hour time windows throughout a day, and in one-half hour time windows each day throughout nearly a month. These latter observations indicate that H/V ratios using ambient noise are an effective method for identifying the frequencies of the fundamental resonance peaks, although the amplitudes again show considerable scatter.

Finally, we are using joint inversions of dispersion curves and H/V ratios to invert for the velocity structure of the Atlantic Coastal Plain strata along the arrays of the Eastern North American Margin experiment (Lynner and others, 2019; Magnani and others, 2014). The inversion is done using a genetic algorithm that evaluates initial random models, chooses components from the best models, and evaluates this as a new generation of models by comparison with the dispersion curves and H/V ratios. After 150 generations, each consisting of 250 models (37,500 models total), we choose the average of the best 750 models (best 2 percent) as our result. Results clearly show the Atlantic Coastal Plain strata having velocities of less than 300 meters per second (m/s) at the surface to nearly 1,000 m/s in the deepest parts of the Atlantic Coastal Plain, below which velocities rapidly increase to more than 2,500 m/s.

By Martin C. Chapman (Virginia Polytechnic Institute and State University) and Zhen Guo (Virginia Polytechnic Institute and State University)

The Atlantic and Gulf Coastal Plains in the southern and southeastern United States contain extensive Cretaceous Period and Cenozoic Era sedimentary sequences of variable thickness. We investigated the difference in site response in the Atlantic and Gulf Coastal Plains relative to sites outside that region using Fourier spectral ratios from 17 regional earthquakes occurring during 2010–18 recorded by the EARTHSCOPE transportable array (IRIS Transportable Array, 2003) and other stations. We used mean coda and Lg spectra for sites outside the Coastal Plain as a reference. We found that Coastal Plain sites experience amplification of low-frequency ground motions and attenuation at high-frequencies relative to average site conditions outside the Coastal Plain (Guo and Chapman, 2019). The spectral ratios at high frequencies gave estimates of the difference between kappa, a measure of ground motion attenuation, at Coastal Plain sites and the reference condition. Differential kappa values determined from the coda are correlated with the thickness of the sediment section and agree with previous estimates determined from Lg waves. Averaged estimates of kappa reach about 120 milliseconds (ms) at Gulf Coast stations overlying ~12 km of sediments. Relations between Lg spectral ratio amplitudes versus sediment thickness in successive frequency bins exhibit consistent patterns, which were modeled using piecewise linear functions at frequencies ranging from 0.1 to 2.8 Hz. For sediment thickness greater than about 0.5 km, the spectral amplitude ratio at frequencies higher than approximately 3 Hz is controlled by the value of kappa. The peak frequency and maximum relative amplification at frequencies less than about 1.0 Hz depend on sediment thickness. At 0.1 Hz, the mean Fourier amplitude ratio (Coastal Plain/reference) is about 2.7 for sediment of 12 km thickness.

The model for the Fourier amplitude ratio was used develop a corresponding peak spectral acceleration (PSA) response spectral ratio model, by direct time-domain simulations via the stochastic model. The PSA ratio model for the Coastal Plain region is referenced to a site condition defined as the mean response for site locations outside the Coastal Plain. The model is strongly dependent on sediment thickness. The model can be used to predict PSA response at sites in the Atlantic or Gulf Coastal Plain with a known thickness of Coastal Plain sediment, given a ground motion prediction model for conditions outside the Coastal Plain region of the Central and Eastern United States. Results will be presented in terms of residual plots using PSA observations from 17 earthquakes and the NGA-East median ground motion prediction model developed specifically for the USGS.

By Roland LaForge (LaForge GeoConsulting)

The Central and Eastern United States Seismic Source Characterization (CEUS-SSC) source model was developed over the period 2008–11 by the Nuclear Regulatory Commission (Electric Power Research Institute, 2012), to provide a regional seismic source model for use in probabilistic seismic hazard analyses (PSHA) for nuclear facilities in the eastern United States. Key components of any PSHA are the ground motion models, which translate a magnitude and distance into a horizontal ground motion. The NGA-East Ground Motion project, completed and published by the Pacific Earthquake Research Center in 2018 (Goulet and others, 2018), comprises the most recent and comprehensive set of ground motion models for the eastern United States and southeastern Canada. Computer code has been developed that applies the NGA-East ground motion models to the CEUS-SSC source model.

The issue of reduced ground motions for ray paths within, and traversing, the Gulf Coast region was of particular concern and focus on the NGA-East project. Details are contained in the “Next Generation Attenuation” report (Goulet and others, 2018), but to summarize, two independent groups derived corrections to non-Gulf Coast ray paths based on simulations and the sparse available observations. The correction is based on the fraction of total distance the ray path spends in the Gulf Coast zone. Weights are applied to the two groups corrections, and two geographic definitions of the Gulf Coast zone. To inform the goals of this workshop, several source-site paths traversing the Gulf Coast regions will be executed in the PSHA code, with and without the correction, to show the effect of the NGA-East correction scheme on hazard curves and uniform hazard spectra.

By Oliver S. Boyd (USGS) and Sarah Mills (Colorado School of Mines)

Earthquake ground motions along the Atlantic and Gulf Coastal Plains show significant frequency-dependent deviations relative to other areas in the Central and Eastern United States (Guo and Chapman, 2019; Pratt and Schleicher, 2021). Ground motions have been shown to be strongly correlated with sediment thickness, where sediment is generally defined as being of Cretaceous Period or younger age. Studies have found that greater sediment thickness leads to higher attenuation and lower ground motion amplitudes at short periods, but higher ground motion amplitudes at longer periods (Guo and Chapman, 2019; Pratt and Schleicher, 2021). With the recent implementation of basin depths for select areas in seismic hazard analyses for the western United States, a USGS working group has been formed to consider whether a sediment thickness model can be constructed for the Atlantic and Gulf Coastal Plains and successfully implemented in Central and Eastern United States seismic hazard analyses. In this presentation, we will detail the many existing sediment thickness datasets that have been gathered by our working group to date and the construction of a composite sediment thickness model for the Atlantic and Gulf Coastal Plains. Most of the data are available as printed maps covering one or more States. Several datasets consist of digitized shapefiles covering larger areas. Contours and sediment thickness values from wells on the printed maps are rectified and digitized. They are combined with the previously available digitized datasets (Mills and others, 2020), and values are averaged where multiple datasets overlap. Smooth transitions are applied to minimize artifacts from the edges of datasets, for example, at State boundaries. The standard deviation of differences between datasets, where overlap occurs, averages about 100 m, providing some estimate of uncertainty. When comparing thicknesses from the composite surface to those produced from a rough digitization of similar surfaces applied in the work of Guo and Chapman (2019) and Thomas L. Pratt (U.S. Geological Survey, written commun., 2020), we find close one-to-one agreement, although values from Pratt’s original analysis on the Atlantic Coastal Plain tended to be on the shallow side for larger sediment thickness values. Pratt’s sediment thickness values for his analysis have since been updated with the new composite surface (Pratt and Schleicher, 2021).

DISCLAIMER: This information is preliminary or provisional and is subject to revision. It is being provided to meet the need for timely best science. The information has not received final approval by the U.S. Geological Survey (USGS) and is provided on the condition that neither the USGS nor the U.S. Government shall be held liable for any damages resulting from the authorized or unauthorized use of the information.

By Morgan P. Moschetti (USGS), Eric Thompson (USGS), Dan McNamara (USGS), and David Churchwell (USGS)

We have compiled an initial ground motion dataset of earthquakes greater than M4 since 2010 near the Atlantic Coastal Plain to evaluate the Atlantic Coastal Plain amplification models being proposed for incorporation into USGS seismic hazard models. The NGA-East ground motion database (Goulet and others, 2021), which is the most widely accepted dataset for earthquake ground motions in the Central and Eastern United States, contains few observations from the Atlantic Coastal Plain, and its most recent records correspond to events in 2011. Since 2010 broadband seismic stations of the USArray Transportable Array (IRIS Transportable Array, 2003) were installed across the Central and Eastern United States, and data from numerous temporary seismic networks have become available (see https://www.fdsn.org for more information). More than 100 earthquakes greater than M4 have occurred in the Central and Eastern United States since that time. Many of these events are potentially induced earthquakes in Oklahoma, Kansas, Texas, and Arkansas, but about 30 events have occurred within about 500 km of the Atlantic and Gulf Coastal Plains. We applied automated ground motion processing to the seismic data from these events to produce a ground motion dataset for the region. The processing workflow used the GMProcess software (Hearne and others, 2019), developed internally and publicly available, which has been applied to earthquake sequences in Alaska, Hawaii, southern California, and the Pacific Northwest to produce Fourier and response spectral amplitudes. The ground motion processing method has been verified with records in the NGA-East and Next Generation Attenuation relationships for the Western United States-2 (NGA-West-2) ground motion databases. Incorporation of additional site metadata (for example, sediment thicknesses) permits evaluation of the amplification functions proposed for use with existing ground motion models (for example, NGA-East) used in the National Seismic Hazard Model.

By Peter M. Powers (USGS) and Jason M. Altekruse (USGS)

We present implementation details and hazard implications of the Guo & Chapman (2019) Atlantic and Gulf Coastal Plain site amplification and attenuation model. The model, derived from spectral ratios, provides adjustments to median hard-rock site condition ground motions as a function of sediment thickness at a site, the earthquake magnitude and hypocentral distance, and response spectral period. When computing hazard, we apply the adjustments to median ground motions derived from the GMM used in the Central and Eastern United States for the 2018 update of the NSHM. We compare CPA-adjusted ground motions to ground motions modified using the 2018 NSHM Central and Eastern United States site effects model, which is a function of V_S30, at a variety of sites throughout the southeastern United States. Further work would be beneficial (and we welcome discussion) on possible adjustments to Central and Eastern United States GMM aleatory variability models and on comparisons to NGA-East Gulf Coastal Plain site effects models. These latter models define CPA site effects as a function of the source-to-site path length in the Gulf Coastal Plain as opposed to sediment thickness at a site.

DISCLAIMER: This information is preliminary or provisional and is subject to revision. It is being provided to meet the need for timely best science. The information has not received final approval by the U.S. Geological Survey (USGS) and is provided on the condition that neither the USGS nor the U.S. Government shall be held liable for any damages resulting from the authorized or unauthorized use of the information.

10:00 a.m. Workshop Overview

Oliver S. Boyd, USGS

Oliver S. Boyd stated the purpose and goals of the workshop and discussed the agenda.

The goal is to address the question of how wave propagation effects in the Coastal Plain region of the Eastern and Central United States can be incorporated into future (possibly 2023) USGS National Seismic Hazard Maps. Specific items to be addressed are as follows:

Allison Shumway provided an introduction for the attendees on using the various functions of Microsoft Teams for the virtual workshop.

Mark D. Petersen thanked the attendees and explained that the USGS has been working on improving the modeling of the effect of geologic structure on earthquake ground motions in the Atlantic and Gulf Coastal Plains for some time. Mark D. Petersen emphasized that one of the workshop goals is to determine whether the work done to date is mature enough to include in the next update of the National Seismic Hazard Model.

10:10 a.m. Participant Introductions

Oliver S. Boyd led an introduction of workshop attendees, where each attendee had the opportunity to relate affiliation, position, and scientific interest.

10:20 a.m. “Proposed Models for the Implementation of Ground Motion Amplification on the Atlantic and Gulf Coastal Plains” by Oliver S. Boyd

Oliver S. Boyd briefly summarized the current models the CPA WG is considering. This includes the models by Pratt and Schleicher (2021), Chapman and Guo (2021), and the Pacific Earthquake Engineering Research Center (Goulet and others, 2018) and unpublished work by Charles Mueller (U.S. Geological Survey, written commun., 2019). He noted that the USGS has developed an Atlantic and Gulf Coastal Plain sediment thickness database and map, which is not yet published. Oliver S. Boyd proposed to the group the idea of giving equal weight to all the current models. He noted that one of the issues to be addressed involves adjustment to these models for variations in V_S30 using National Earthquake Hazard Reduction Program or other recently developed site amplification factors.

10:40 a.m. Group Discussion

Walter Mooney asked about the definition of sediment. Charles Mueller and Oliver S. Boyd responded by saying that the Atlantic and Gulf Coastal Plain sediments are those units (mostly of Cretaceous Period and younger age) lying above the post-rift unconformity. Eric Thompson had questions about larger magnitude events than those in the existing dataset, the importance of surface waves, and the possibility of nonlinear response in the Coastal Plain sediments. Walter Mooney pointed out that the practice of using V_S30 for site response definition is quite old and incomplete and may not be particularly useful in the Central and Eastern United States. Shahram Pezeshk asked about the purpose of the mapping effort regarding regional versus site-specific hazard assessment.

5-minute break

11:00 a.m.–12:15 p.m. 25-minute participant presentations on relevant research

11:00 a.m. “Seismic Response of Thick Unconsolidated Sediments” by Charles A. Langston

Charles A. Langston pointed out that the standard source-path-site convolutional approach often used for earthquake hazard assessment can break down in cases were a fault ruptures into a sedimentary structure, such as the Mississippi embayment or Atlantic and Gulf Coastal Plain. Such an event can generate large surface waves, and the database does not contain examples of this potential event. He emphasized the importance of using full wavefield simulations to quantify ground shaking hazard. He also stressed the important datasets needed to accomplish the modeling, including velocity well logs, seismic reflection data to constrain sediment thickness, and average velocity measurements of the sediment section and the resonant frequencies. Methods include receiver functions and phase modeling of micro-earthquakes. He advocated the use of controlled experiments to better improve our understanding of wave propagation in the Mississippi embayment and Coastal Plain. He finds that Q for S and P waves is approximately 100 and 200, respectively, in the Mississippi embayment, from surface-wave analysis. Also, the upper Mississippi embayment acts to trap surface waves, increasing amplitude and duration.

11:25 a.m. “Deep Sediments Thickness Correction and Apparent Anelastic Attenuation in Applications to Ground Motion Modeling” by Vladimir Graizer

Vladimir Graizer discussed the use of Z_1.0, Z_1.5, and Z_2.0. These are the depths in a sedimentary basin to shear-wave velocities of 1.0 km/s, 1.5 km/s, and 2.0 km/s, respectively. He recommends the term “deep sediment effect” rather than the term “basin effect” in the context of estimating response in GMM due to sedimentary structures because none of the Z terms represents depth to the bottom of the basin. Basin edge and basin shape effects are not considered. He noted that the V_S30 scaling term in GMM often already incorporates implicit basin amplification effects to some degree. He also pointed out the difference between what he refers to as Q_SA (quality factor derived from engineering SA response spectra) and the traditional seismological quality factor Q, which is often derived from the Fourier amplitude spectrum of ground motion.

11:50 a.m. “Site Amplification Models for Central and Eastern North America” by Youssef M.A. Hashash, Okan Ilhan, Halil Uysal, Jonathan P. Stewart, Ellen M. Rathje, and Sissy Nikolaou

Youssef Hashash reviewed the parametric study of more than 1.7 million linear, equivalent-linear, and nonlinear 1-D site response analyses to quantify variability of site response in the Central and Eastern United States. This led to the development of linear and non-linear amplification models for the latest USGS National Seismic Hazard Maps. Recently, that work has been updated using artificial neural network deep learning techniques. These studies highlight the need to go beyond V_S30 as a proxy for site amplification. The study underscores the importance of site period. Models for site response in terms of both linear and nonlinear Fourier amplitude spectra have been developed.

12:15 p.m.–1:00 p.m. Lunch

1:00 p.m. “Primary Site-Response Parameters and an Evaluation of Site-Response Proxies at Selected Sites in the Upper Mississippi Embayment” by Seth Carpenter

Seth Carpenter presented results from site-response studies in southwest Kentucky (the Jackson Purchase) where they looked at various proxies for V_S30 when estimating site response. They made the point that V_S30 is inadequate to characterize site response by showing several profiles, both modeled and real, showing that similar V_S30 values can come from much different velocity structures that produce much different site responses. The main issue is that V_S30 does not account for deeper interfaces that can cause strong resonances. They presented results from two deep borehole sites (VSAP—Vertical Seismic Array Paducah; and CUSSO—Central U.S. Seismic Observatory) in the upper Mississippi embayment where they had seismometers both at the surface and at depth. One dimensional (1-D) models of site response reproduced the observed site response well, which is consistent with the flat layers of the upper embayment showing little influence from three dimensional (3-D) effects. Maps of fundamental frequency and amplification in the Jackson Purchase were shown to have little correlation with each other; fundamental frequency correlates well with depth, but amplification and peak frequency do not. There was also little correlation between V_S30 and amplification. They conclude that fundamental frequency and amplitude match 1-D models well, and the impedance ratios in the models work well. Estimating the amplitude of the fundamental peak using H/V overestimates the amplitude compared to full resonance models made from the borehole data, but dividing the H/V amplification by a factor of 1.5 brings the H/V into a better match with the models. They summarized that V_S30 does not work well for estimating site response in the Mississippi embayment and that there is no single proxy that works well.

1:30 p.m. “1-D Velocity Profiles of Atlantic Coastal Plain Sedimentary Strata from Receiver Functions” by Erin Cunningham

Erin Cunningham described her work on using receiver function analyses to estimate sediment thickness and velocity in the Atlantic Coastal Plain east and south of the Appalachians, which was described in her 2020 Bulletin of the Seismological Society of America paper (Cunningham and Lekic, 2020). They assume 1-D velocity structure and used receiver functions to recover the impedance contrast and resonances within the sediments. In particular, the autocorrelation of the receiver function works well to estimate the frequency and amplitude of the primary reverberations, both for S waves and P waves. She then showed maps of fundamental frequency at USArray sites and the estimated reflection coefficient at the base of the sediments. She also computed a velocity versus depth function under the assumption of flat layers, with average velocity to a particular depth being constant and differences to bedrock being the result of successively deeper layers.

2:00 p.m. “Central and Eastern United States Attenuation Boundaries and Coastal Plain Crustal Seismic Attenuation Models” by Chris H. Cramer

Chris H. Cramer gave a presentation on the crustal boundaries in the Gulf Coast region, which was largely a summary of his 2018 Bulletin of the Seismological Society of America paper (Cramer, 2018). He also added some observations of ground motions from the recent M5.1 Sparta, North Carolina, earthquake. For the boundary work, he used the vertical component of $$USArray data to look at the 5 Hz versus 1 Hz boundaries west of Alabama based on attenuation (the inverse of the quality factor, Q). He presented several regressions of Q in different areas in the form of Q=Q_ofⁿ, where Q_o is the quality factor at 1 Hz, f is shaking frequency, and n is an exponent that controls how quickly Q increases with frequency. He concluded that the Gulf Coast region has higher attenuation in an area smaller than was previously thought. The high-attenuation region is confined largely to the Gulf Coast of Texas, Louisiana, and Mississippi. Alabama, Georgia, and Florida do not have reduced amplitudes because they lie outside of the high-attenuation area. There is a change at the approximate Georgia-Florida border, however, with Q being higher in Florida. The pattern he sees from analysis of USArray data are consistent with intensity observations from the 1886 Charleston, South Carolina, earthquake, which showed a rapid decrease in intensities at the edge of the Gulf Coast region. There is also a boundary in southern Oklahoma, with higher Q north of this boundary.

He presented a focal mechanism for the Sparta, North Carolina earthquake, with a focal plane that is consistent with the small fault scarp at the surface. Depth seems to be very shallow, with a stress drop of about 22 mega Pascals (MPa) that seems similar to other earthquakes in the Central and Eastern United States investigated by Cramer such as induced earthquakes in Oklahoma. There is a strong azimuthal variation in ground motions from the earthquake, with higher attenuation to the northwest than in other directions, although the ground motions are least attenuated to the northeast along strike of the Appalachians. Q seems to be lower than in Atkinson and Boore (2014).

2:30 p.m. “Mapping Site Response in the U.S. Coastal Plain” by Charles Mueller

Charles Mueller gave a presentation on his work in estimating site response in the Atlantic and Gulf Coastal Plain. He put extensive work into digitizing maps of the sediment thickness, mostly from the Decade of North American Geology (DNAG) volumes from 1980–90 (Klitgord and others, 1988; Olsson and others, 1988; Salvador, 1991). He tried to produce simple models of ground motion amplification with a map of sediment thickness, S-wave velocity, and Q for the region and using the SMSIM (Fortran Programs for Simulating Ground Motions from Earthquakes—Boore, 2000) point-source simulation code. He defined the base of the sediments as the post-rift unconformity, assuming sediments were deposited on uniform basement rocks. Power law V_S versus depth and Q versus depth were assumed for the Coastal Plain. He also incorporated geologic maps of the surface units (Schruben and others, 1994).

He used a Thompson-Haskell matrix method and the quarter-wavelength method to estimate site response throughout the Atlantic and Gulf Coastal Plain region, concluding that the quarter-wavelength method is preferred because it had less influence from higher-mode resonance peaks. He then computed site responses, showing a suite of models with some areas of amplification and others of deamplification. The results were not compared to recordings of ground motions, so they are strictly model-based results.

10:00 a.m. “Atlantic Coastal Plain Amplification and Velocity Studies” by Thomas L. Pratt

Thomas L. Pratt measured teleseisms and regional earthquakes (shorter time windows) from four arrays: Eastern North America seismic experiment (Lynner and others, 2019; Magnani and others, 2014), Southeastern Suture of the Appalachian Margin Experiment (SESAME; Fischer and others, 2010), EarthScope Temporary Array (IRIS Transportable Array, 2003), and Washington DC seismic array (Pratt and others, 2017) to calculate sediment-to-bedrock spectral ratios. Bedrock sites had a flat response whereas sites on the Atlantic Coastal Plain showed a prominent resonance peak, f_o and low amplification at values greater than 5 Hz. As sediments get thicker, the resonant peak decreases in amplitude and frequency. Teleseisms work well for looking at the spectral ratios, even without a lot of frequency content (die out at 6–7 Hz).

He showed plots of f_o versus thickness, amplification of resonance peak versus thickness, and amplification of resonance peak versus frequency. At 0.8 Hz and above, there is a need for site-specific studies, as amplifications attenuate. Scatter is large at higher frequencies (for which the thicknesses are not known as well) and may be due to near-surface layers varying spatially.

He showed that the Atlantic Coastal Plain is different in North Carolina and Virginia when compared to Georgia and Florida.

He characterized observations with equations: for h<h_o, for h_o>h<2h_o, and for h>2h_o and compares with Harmon and others (2019b) corrections. Correction curves do well out to several hundred meters thickness, then start missing resonance peaks. Comparisons with sediment-to-bedrock spectral ratios show that H/V spectral ratios tend to be about three quarters amplitude of bedrock spectral ratios.

H/V with ambient noise—He can identify resonance peaks using ambient noise, probably because they are dominated by a 1-D velocity substructure, but it takes averaging over several windows to avoid weird peaks from noise.

Conclusions were as follows (1) low frequencies (<0.8 Hz) can be characterized well as amplitude versus thickness, (2) above ~0.8 Hz, site-specific studies need to be done (amplifications are not predictable in a regional sense), (3) H/V using ambient noise or earthquakes identify resonance frequency well, amplitude of H/V is about 0.75 times that of sediment-to-bedrock spectral ratios, (4) H/V can be partially corrected using vertical to horizontal spectral ratios from bedrock site(s), (5) future work is planned in Charleston (SUGAR—Suwanee Suture and Georgia Rift basin—experiment, velocity inversions) and 200 station embayment nodal array from bedrock to bedrock across the Mississippi embayment.

Velocity inversions using Eastern North America seismic experiment dataset is a work in progress—joint inversion of dispersion and H/V curves.

Inflection point moves to lower frequencies with increase in sediment thickness.

A total of 37,500 randomized models are considered in the genetic algorithm that is used to solve for the best-fitting velocity profile.

He shows an Atlantic Coastal Plain shear-wave velocity profile, which is a work in progress. The surface has a velocity of 300 m/s and bedrock is 1,000 m/s. Within the sedimentary wedge, there are 300–600 m and 600–1,000 m layers, but these results need refinement.

Planned future work: additional velocity inversions (Charleston, SUGAR experiment) and embayment nodal array (about 200 stations spanning 250 km of Mississippi embayment).

10:30 a.m. “A Response Spectral Ratio Model to Account for Amplification and Attenuation Effects in the Atlantic and Gulf Coastal Plain” by Martin Chapman

Martin C. Chapman started out showing three of his papers that provide background for this work (Chapman and Conn, 2016; Guo and Chapman, 2019; Chapman and Guo, 2021). They want to quantify the response of the Atlantic and Gulf Coastal Plains sites relative to reference sites. They use a traditional spectral ratio approach for coda and Lg waves. The reference condition is not a rock site, but an average response of stations located outside the Atlantic and Gulf Coastal Plains region. The motivation is the need to develop correction factors for Atlantic and Gulf Coastal Plains sites that can be applied to existing GMM for Central and Eastern North America.

This study has two elements: (1) determine models for the ratios of Fourier amplitude spectra for stations inside the Atlantic and Gulf Coastal Plains with respect to the average Fourier amplitude spectrum of stations outside the Atlantic and Gulf Coastal Plains and (2) use the Fourier amplitude spectral ratio model to develop a model for PSA spectral ratios as functions of earthquake magnitude (M), source-to-site distance (R), and sediment thickness that can be used as correction factors for existing GMMs for response at sites outside the Atlantic and Gulf Coastal Plains.

There were 17 earthquakes used in study, mostly induced seismicity events in Oklahoma and a few tectonic events (2011 Mineral, Virginia, and earthquakes in South Carolina, Eastern Tennessee Seismic Zone, Kentucky, and Delaware). Comparison of coda and Lg waves recorded in the Atlantic and Gulf Coastal Plains and at a reference site show amplification at low frequencies and attenuation at high frequencies.

For an example of the Fourier amplitude spectra ratio model, they look at coda and Lg waves recorded in the Atlantic and Gulf Coastal Plains and at a reference site (Iowa). The coda spectral ratio represents the response of the Atlantic and Gulf Coastal Plains site relative to the reference site condition. They also consider a kappa model for inside and outside Atlantic and Gulf Coastal Plains. They take the ratio of two stations within Atlantic and Gulf Coastal Plains and find that kappa increases with sediment thickness.

At low frequencies, the coda spectral ratios at large lapse time contain surface wave energy and may not be a good estimate of the relative site response for the higher amplitude parts of the seismograms that are associated with the Lg phase. They showed Lg spectral ratio versus sediment thickness. At 0.1 Hz, ratios increase to 2.5 km thickness, and continue to increase to 12 km thickness. At 0.5 Hz, spectral ratio decreases with increasing thickness above the transition thickness, which becomes obvious above 1 Hz—deamplification at higher frequencies with increasing thickness. At high frequency (>2.68 Hz), the spectral ratio reflects deamplification due to a delta kappa term that is a function of sediment thickness. He summarized by saying that with increasing thickness, amps go up then turn over at higher frequencies. There are good amps at low frequencies, then attenuate at high frequencies.

He then moved on to discussing how to go from Fourier to PSA by using a stochastic method of ground motion time series simulation to develop synthetic acceleration time series. To do this, a range of earthquake magnitude, distances, and sediment thickness is needed. He went on to show PSA response spectrum for the Atlantic and Gulf Coastal Plains divided by PSA spectra for reference rock condition assuming a stochastic phase model and Lg duration model (same slope as Herrmann, 1985). The PSA response spectra ratios were calculated for thousands of magnitudes, distances, and thickness. The residuals are relative to the reference calculated GMM, which we used for exploring the effectiveness of the PSA spectral ratio model and is the weighted mean NGA-East model developed specifically for the USGS (V_S30 = 3,000 m/s) with the Stewart and others (2017) linear site amp factor for V_S30 = 760 m/s added.

Conclusions: The PSA spectral ratio model can correct the existing dataset for Atlantic and Gulf Coastal Plains stations to be in good agreement (on average) with predictions of the NGA-East model developed specifically for the USGS (on average) for stations outside the Atlantic and Gulf Coastal Plains using the Stewart and others (2017) V_S30 = 760 m/s amplification corrections. If you do not use the Stewart and others (2017) correction factors at high frequencies, you get a small, positive bias in the residuals. The existing dataset is limited to M5.8 for testing the model, and the model is for linear response only.

11:00 a.m. “Effects of NGA-East Ground Motion Models’ Gulf Coast Corrections on Hazard Curves and Uniform Hazard Spectra, as applied to the CEUS-SSC Source Model” by Roland LaForge

NGA-East developed Gulf Coast corrections as part of the project. Two groups came up with correction factors. DASG (Darragh, Abrahamson, Silva, and Gregor) Gulf Coast ground motion model (Darragh and others, 2018) uses a theoretical-based approach based on the point source stochastic model, constraining the point-source parameters by data inversion from the Gulf Coastal Plain. They ran ground motion simulations for a wide range of M (4.5–8.5) and Joiner-Boore source-to-site distance (1–1,000 km) for both regions and computed PSA ratios. Their model is frequency dependent. The Pacific Earthquake Engineering Research Center (PEER) Gulf Coast model (Hollenback and others, 2018) used an empirically based Fourier amplitude spectrum functional form, which was recalibrated with the Gulf Coastal Plain data through a residual analysis procedure. It is not frequency dependent. NGA-East also defines two gulf zone geographic models (large and small).

Roland LaForge implemented the NGA-East GMM with (and without) the Gulf Coast correction factors into his code for the Nuclear Regulatory Commission to run hazard using the CEUS-SSC source model (very complex model with large logic trees).

He looked at hazard results for the city of New Orleans, Louisiana, to see how hazard compared when using the Gulf Coast corrections or not using the corrections. Corrections were determined by horizontal distance of ray paths in the Gulf zones. For the weighted reduction model, DASG was given a weight of 0.33 and PEER a weight of 0.67. The large Gulf zone geometry was given a weight of 0.6 and the small Gulf zone geometry was given a weight of 0.4. No site response was included (hard rock sites are assumed). Tracing ray path is complicated as the path can go in and out of the gulf zones more than once.

Roland LaForge presented hazard results for short return periods (475, 975, and 2,475 years) and long return periods (10,000; 50,000; and 100,000 years). He showed results for DASG, PEER, and then the weighted reduction model. He then showed a bunch of tables comparing the corrected and uncorrected results.

This analysis gives a rough estimate of how the Gulf Coastal Plain ground motion reductions, as implemented in NGA-East and applied to the CEUS-SSC model, affect uniform hazard response spectra for a large range of return periods from the USGS (short) to the Nuclear Regulatory Commission (long). Maximum reductions in long return period/ Nuclear Regulatory Commission hazard spectra are on the order of 0.05 the acceleration of gravity (g) at 1–2 Hz. Maximum reductions in short return period/USGS hazard spectra are <0.02 g at 1–2 Hz. Both reductions are on the order of 30 percent. No site response was included—how would this affect the results? This analysis should be useful in comparing any future Gulf Coastal Plain reduction algorithms to those in the NGA-East.

11:30 a.m. “Sediment Thickness along the Atlantic and Gulf Coastal Plains” by Oliver Boyd

Oliver S. Boyd presented what he and Sarah Mills did to put together a sediment thickness surface along the Atlantic and Gulf Coastal Plains. They used Guo and Chapman (2019) and Pratt and Schleicher (2021) sites in their analysis. These studies rely on the same underlying datasets for sediment thickness as what Oliver S. Boyd uses, but some of their sites have different thickness values. Charles Mueller also had thickness values (from digitizing DNAG maps). Sarah Mills (student intern from Colorado School of Mines) obtained and digitized datasets—contour maps from six publications. Oliver S. Boyd would like to include Maryland, Delaware, and the DNAG maps for the Atlantic and Gulf Coast. Data also were available online (well data). Sarah Mills georeferenced the wells onto the contour maps and then digitized them all. Some of the maps had overlapping data (up to four models in the same location); therefore, epistemic uncertainty could be assessed. The average standard deviation is about 100 m. Multiple models were given equal weight. They came up with a composite surface and compared with Guo and Chapman (2019) and Pratt and Schleicher (2021) and got a one-to-one agreement, mostly. There are larger differences with Mueller values, which may be due to fewer DNAG values available in certain areas. They need to digitize more DNAG maps.

Six contour maps and four well datasets are georeferenced and digitized and they obtained two hydrogeologic models and a North Carolina well database. These sources of data are combined with Mueller’s DNAG data to produce a composite surface of the elevation of the basement at the top of the post-rift unconformity (pre-Cretaceous Period). When multiple datasets overlap, the average standard deviation is 100 m with deviations over 500 m in limited areas. The resulting composite surface is consistent with Guo and Chapman (2019) and preliminary analysis of Pratt and Schleicher (2021) in which they roughly estimate these values from contour maps, which is not surprising as Oliver S. Boyd and Sarah Mills are using nearly the same datasets. Planned additional work: the surface needs to be extended into west Texas, digitizing more of the DNAG contours would improve the model, and the model needs to be published.

12:00 p.m.–1:00 p.m. Lunch

1:00 p.m. “A Ground Motion Database to Support Evaluation of Atlantic Coastal Plain Amplification Factors for Use in Seismic Hazard Assessments” by Morgan P. Moschetti

Morgan P. Moschetti presented plans to make a database for evaluation of Atlantic Coastal Plain amplification factors. GMProcess: new code/software being used in our office for automated signal processing, on github, software data release by Hearne and others (2019). Processing uses the same workflow as was used for NGA-East and NGA-West-2 projects. Significant efforts were made to read different data formats, pick signal/noise windows, and ensure useable signal. Software provides reports. Code is verified using data from NGA-West-2 and NGA-East projects. Same code has been used in many other ongoing projects such as Oklahoma induced earthquakes, Alaska in-slab, Hawaii caldera collapse, Ridgecrest. Atlantic Coastal Plain data: showed preliminary data (processed examples shown) from NGA-East database and a channel network map with recordings between 2010–20. For events greater than M4, 32 events, <500 km → 1,200 records. Very few records for <100 km. Max M5.75. Mostly focused on future plans: compile the database for testing Atlantic Coastal Plain, calculate ground motion residuals using Atlantic Coastal Plain adjusted NGA-East model, and evaluate amplification/ deamplification. They suspect bias between NGA-East and Martin C. Chapman reference rock sites. They will investigate trends with distance, magnitude, V_S30, and sediment thickness, and will come up with recommendations.

1:30 p.m. “Implementation of the Guo & Chapman (2019) Coastal Plain Amplification Model” by Peter M. Powers

Peter M. Powers presented the preliminary implementation of the Guo and Chapman (2019) model (GC19; preliminary model was provided to the USGS). He discussed the following. (1) sediment thickness and webservice—data comes from Oliver S. Boyd for both Gulf and Atlantic Coasts. This can be accessed from a web-service (https://earthquake.usgs.gov/ws/nshmp/data/gulf) as a part of the NSHMP Non-NSHM Services. Given latitude and longitude one can get sediment thickness. (2) GC19 implementation: function of period (T), magnitude (M), source-to-site distance (R), and sediment thickness (z). A preliminary implementation can be found on the Response Spectrum Plotter tool of USGS (https://earthquake.usgs.gov/nshmp/gmm/spectra), but it ignores V_S30. The tool temporarily uses Z_2.5 for z input. For deterministic response spectra, the model reduces SA at short period but increases SA at long period (>1 s). (3) Hazard curves were shown for example sites for three cases to illustrate the differences between hard rock, a site outside the Atlantic and Gulf Coastal Plains with estimated non-rock V_S30, and an Atlantic and Gulf Coastal Plains site: results for New Madrid, Missouri show that GC19 is similar to Central and Eastern United States. Site effects modeled at long period (5 s), are lower at 1 s, and cross over at 0.2 s. Results really depend on the return period. For probabilistic response spectra, in general, GC19 reduces SA more than the current Central and Eastern United States amplification at short period, but SA is almost the same for longer periods. Similar results were shown for other locations such as Miami, Florida, where GC19 reduces SA at short period but increases at long period. (4) Planned next steps: implement other models (for example, the unpublished results of Charles Mueller), look at aleatory variability, and address handling path properties for NGA-East.

2:20 p.m.–2:59 p.m. Discussion

Discussions started after Peter M. Power’s presentation (see above).

Oliver S. Boyd presented a few slides summarizing the issues for discussion that came up during the two days of workshop:

Jonathan P. Stewart chatted that he has a suggestion on a path forward. He added this to the chat window “At the time we generated the NGA-East GWG site models that are currently in use, we did not have access to Atlantic Coastal Plain depth maps as have been shown in this workshop. We looked for trends of residuals with respect to some older basin models (Grace Parker can speak to the details), which were inconclusive, but the maps of Atlantic Coastal Plain extent and depth should be used to recheck this. Trends that are identified could be modelled to capture Atlantic Coastal Plain-specific deviations from the general Central and Eastern North America model. This would provide a valuable check and would support the evaluation of phiS2S models specific to the Atlantic Coastal Plain.” In response, Youssef Hashash wrote “Per Jon's note, some of the work we have done can be extended to the Atlantic Coastal Plain depths and bring in natural period influence.”

Youssef Hashash put this note in the comments: “As I mentioned in my call, we have an update of Harmon's work that would be more suitable for the greater depths.”

2:59 p.m.–3:00 p.m. Next Steps

Oliver S. Boyd quickly summarized planned next steps for the Atlantic and Gulf Coastal Plain working group efforts. They plan to complete and publish the Atlantic and Gulf Coastal Plains sediment thickness model. They plan to compare observations from Morgan P. Moschetti and other’s ground motion database with the Guo and Chapman (2019) and Pratt and Schleicher (2021) site amplification models. The working group plans to compare these results with other published models, for example, Pezeshk and others (2018), Cramer (2018), Harmon and others (2019b) and NGA-East (Goulet and others, 2018). Lastly, they plan to determine logic tree weights for these various site amplification models. More sensitivity analyses would be useful, and the working group and larger earthquake engineering community are likely to benefit from an additional workshop, which was not discussed. Oliver S. Boyd and others will complete a workshop report detailing this meeting.

3:00 p.m. Meeting Adjourned