Martin C. Chapman

A seismotectonic model for the 300-kilometer-long eastern Tennessee seismic zone

Ten years of monitoring microearthquakes with a regional seismic network has revealed the presence of a well-defined, linear zone of seismic activity in eastern Tennessee. This zone produced the second highest release of seismic strain energy in the United States east of the Rocky Mountains during the last decade, when normalized by crustal area. The data indicate that seismicity produced by regional, intraplate stresses is now concentrating near the boundary between relatively strong and weak basement crustal blocks.

On the Use of Elastic Input Energy for Seismic Hazard Analysis

The disaggregation of probabilistic seismic hazard calculations based on elastic input energy may prove useful for the identification of scenario events because input energy is a convenient single-parameter descriptor of motion duration and amplitude. To investigate this application, regression models are derived for the absolute input energy equivalent velocity, V ea , and the elastic pseudo-relative velocity response, PSV, in the frequency range 0.5 to 10 Hz. Disaggregation of a general seismic hazard model using V ea indicates that the modal magnitudes for the higher frequency oscillators tend to be larger, and vary less with oscillator frequency, than those derived using PSV. Larger magnitude earthquakes contribute more to seismic hazard if V ea is used. The dependence of V ea and PSV upon site classification is virtually identical, and V ea can be predicted with slightly less uncertainty as a function of magnitude, distance and site classification.

The Mw 5.8 Mineral, Virginia, earthquake of August 2011 and aftershock sequence: constraints on earthquake source parameters and fault geometry

The Mw 5.8 earthquake of 23 August 2011 (17:51:04 UTC) (moment, M0 5:7 × 10 17 N·m) occurred near Mineral, Virginia, within the central Virginia seis- mic zone and was felt by more people than any other earthquake in United States history. The U.S. Geological Survey (USGS) received 148,638 felt reports from 31 states and 4 Canadian provinces. The USGS PAGER system estimates as many as 120,000 people were exposed to shaking intensity levels of IV and greater, with approximately 10,000 exposed to shaking as high as intensity VIII. Both regional and teleseismic moment tensor solutions characterize the earthquake as a northeast- striking reverse fault that nucleated at a depth of approximately 7 2 km. The distri- bution of reported macroseismic intensities is roughly ten times the area of a similarly sized earthquake in the western United States (Horton and Williams, 2012). Near- source and far-field damage reports, which extend as far away as Washington, D.C., (135 km away) and Baltimore, Maryland, (200 km away) are consistent with an earthquake of this size and depth in the eastern United States (EUS). Within the first few days following the earthquake, several government and aca- demic institutions installed 36 portable seismograph stations in the epicentral region, making this among the best-recorded aftershock sequences in the EUS. Based on modeling of these data, we provide a detailed description of the source parameters of the mainshock and analysis of the subsequent aftershock sequence for defining the fault geometry, area of rupture, and observations of the aftershock sequence mag- nitude-frequency and temporal distribution. The observed slope of the magnitude- frequency curve or b-value for the aftershock sequence is consistent with previous EUS studies (b 0:75), suggesting that most of the accumulated strain was released by the mainshock. The aftershocks define a rupture that extends between approxi- mately 2-8 km in depth and 8-10 km along the strike of the fault plane. Best-fit modeling of the geometry of the aftershock sequence defines a rupture plane that strikes N36°E and dips to the east-southeast at 49.5°. Moment tensor solutions of the mainshock and larger aftershocks are consistent with the distribution of aftershock locations, both indicating reverse slip along a northeast-southwest striking southeast- dipping fault plane.

Site-Response Models for Charleston, South Carolina, and Vicinity Developed from Shallow Geotechnical Investigations

The study models the response of near-surface materials in Charleston, South Carolina, and the adjacent area. Geotechnical investigations at 281 locations were made available by local engineering firms. The data used for dynamic site- response analysis were derived from shear-wave velocity measurements at 52 loca- tions. Site response was quantified as the ratio of surface motion to hypothetical hard-rock basement outcrop motion. Scenario earthquake motions were developed with the stochastic model. Acceleration response ratios for 5% critical oscillator damping were computed for 12 frequencies ranging from 0.1 to 30 Hz and for peak ground acceleration. Two features determine the general nature of site response in the study area: the impedance contrast between Mesozoic basement and Cretaceous sediments, and the shallow impedance contrast between Quaternary and Tertiary sediments. Average S- wave velocities in the Quaternary are relatively uniform and range from 150 to 250 m/sec. They are not strongly correlated with surface geology. The velocities of the immediately underlying Tertiary sediments range from 300 to 500 m/sec. Because of the uniformity of velocity in the Quaternary, depth to the Quaternary-Tertiary contact appears to be the most important variable leading to differences in calculated site response. This surface is irregular, and varies in depth from near surface at inland sites to approximately 30 m at sites near the coast. As a consequence, estimated site response in the frequency band 1-10 Hz varies by as much as a factor of 3. Site response at frequencies less than 1 Hz is dominated by the first few resonant har- monics of the entire sedimentary section, with fundamental frequency near 0.2 Hz.

Frequency-Dependent Seismic Attenuation in the Eastern United States as Observed from the 2011 Central Virginia Earthquake and Aftershock Sequence

Abstract Ground shaking due to earthquakes in the eastern United States (EUS) is felt at significantly greater distances than in the western United States (WUS) and for some earthquakes it has been shown to display a strong preferential direction. Shaking intensity variation can be due to propagation path effects, source directivity, and/or site amplification. In this paper, we use S and Lg waves recorded from the 2011 central Virginia earthquake and aftershock sequence, in the Central Virginia Seismic Zone, to quantify attenuation as frequency‐dependent Q ( f ). In support of observations based on shaking intensity, we observe high Q values in the EUS relative to previous studies in the WUS with especially efficient propagation along the structural trend of the Appalachian mountains. Our analysis of Q ( f ) quantifies the path effects of the northeast‐trending felt distribution previously inferred from the U.S. Geological Survey (USGS) “Did You Feel It” data, historic intensity data, and the asymmetrical distribution of rockfalls and landslides.

Upper mantle anisotropy and transition zone thickness beneath southeastern North America and implications for mantle dynamics

A variety of models for mantle flow beneath southeastern North America have been proposed, including those that invoke westward driven return flow from the sinking Farallon slab, small-scale convective downwelling at the edge of the continental root, or the upward advective transport of volatiles from the deep slab through the upper mantle. We use shear wave splitting observations and receiver function analysis at broadband seismic stations in the southeastern United States to test several of these proposed mantle flow geometries. Near the coast, stations exhibit well-resolved null (no splitting) behavior for SKS phases over a range of back azimuths, consistent with either isotropic upper mantle or with a vertical axis of anisotropic symmetry. Farther inland we identify splitting with mainly NE–SW fast directions, consistent with asthenospheric shear due to absolute plate motion (APM), lithospheric anisotropy aligned with Appalachian tectonic structure, or a combination of these. Phase-weighted stacking of individual receiver functions allows us to place constraints on the timing of arrivals from the 410 and 660 km discontinuities and on average transition zone thickness beneath individual stations. At most stations we find transition zone thicknesses that are consistent with the global average (∼240 km), with two stations showing evidence for a slightly thickened transition zone (∼250 km). Our results are relevant for testing different models for mantle dynamics beneath the southeastern United States, but due to the sparse station coverage, we are unable to uniquely constrain the pattern of mantle flow beneath the region. Our SKS splitting observations support a model in which mantle flow is primarily vertical (either upwelling or downwelling) beneath the southeastern edge of the North American continent, in contrast to the likely horizontal, APM-driven flow beneath the continental interior. However, our receiver function analysis does not provide unequivocal support either for widespread hydration of the transition zone or for widespread thickening due to the downwelling of relatively cold mantle material. We expect that the necessary data to constrain such models more tightly can be obtained from the operation of denser seismic networks, including the Transportable Array and Flexible Array components of USArray.

The 9 December 2003 Central Virginia Earthquake Sequence: A Compound Earthquake in the Central Virginia Seismic Zone

The 9 December 2003 central Virginia earthquake sequence was a compound earthquake consisting of two nearly identical events occurring about 12 sec apart. The second event is separated by about 300 m from the first event along the azimuth of approximately 195 (±10)°. The source mechanism determined from regional waveform inversion indicates predominantly thrust faulting at a depth of approximately 10 (±2) km. The sequence with two events attained the combined seismic moment of M 0 2.64 (±1.01) × 10 15 N m ( M w 4.3). The focal mechanism indicates a subhorizontal P axis trending 301° and plunging 19°. A regional stress model for the central Virginia seismic zone (cvsz) derived from the 9 December 2003 events and 11 previous earthquakes indicates a thrust-faulting stress regime with σ 1 trending 133 (±12)° and plunging 14 (±3)°. The least principal stress axis ( σ 3 ) trends 25 (±10)° and plunges 52 (±3)°. The σ 1 axis is rotated approximately 68° clockwise relative to the average maximum horizontal compressional stress (S Hmax ) direction for eastern North America (approximately 65°). The 9 December 2003 earthquake sequence occurred among the systems of Paleozoic and Mesozoic faults above the southern Appalachian decollement, which is at depths from 12 to 19 km in the Piedmont geologic province of central Virginia.

Joint hypocenter-velocity inversion for the eastern Tennessee seismic zone

A joint hypocenter-velocity inversion for the eastern Tennessee seismic zone (ETSZ) has resolved velocity features in basement rock below detached Appalachian thrust sheets. P and S wave arrival times from 492 earthquakes have been inverted for one-(1-D) and three-dimensional (3-D) velocity models to midcrustal depths. The 3-D P and S wave velocity solutions are computed independly and are very similar. In relation to the 1-D model, velocity anomalies range from −8% to +16% in the first layer (upper 5 km) and between ±7% in deeper layers. Prominent velocity anomalies parallel the seismic zone and are consistent from layer to layer. The most persistent anomaly is a low-velocity region that borders the seismic zone to the northwest and is flanked on either side by regions of anomalously high velocity. The New York-Alabama (NY-AL) magnetic lineament coincides with or lies close to the southeast boundary of the prominent velocity low in both the P and S wave velocity images. The spatial coincidence between velocity, gravity, and magnetic gradients suggests that major discontinuities are present in the basement. Relocation in the 3-D velocity model reduced the number of very deep earthquakes (below 20 km) and further accentuated differences in seismogenic properties on either side of the NY-AL lineament. After relocation, most earthquakes occur in a vertically bounded region roughly 30 km wide extending from 4 to 22 km in depth. Most earthquakes occur in regions characterized by either average velocity or small velocity anomalies.

Ground-Motion Attenuation in the Atlantic Coastal Plain near Charleston, South Carolina

Charleston, South Carolina, lies on approximately 1 km of Cretaceous and Cenozoic sediments of the Atlantic Coastal Plain. Estimation of high-frequency absorption due to these sediments is important for strong motion prediction. We attempt the measurement using microearthquake data recorded at small distances by surface and shallow subsurface short-period stations in the Middleton Place––Summerville seismic zone. The problem is difficult because it involves potential bias due to the seismic source, propagation through basement, and strong site-specific spectral modulation. Previous studies involving drilling and seismic reflection profiling indicate a 775-m thickness of sediments, with average vertical P - and S -wave velocities of 2.14 and 0.700 km/sec for the network area. The attenuation parameter kappa ( κ s and κ p) for S and P waves is estimated from spectral analysis of the direct S and converted Sp phases. The ratio of S to Sp provides a useful check for bias. Multiple linear regression using all stations yields κ s = 0.049 and κ p = 0.024. The regression results are interpreted as upper-bound estimates because they assume source corner frequencies in excess of 25 Hz. A similar analysis is carried out for a hard-rock environment using reservoir-induced microearthquakes at Lake Monticello, South Carolina. From that, we estimate a maximum potential bias of 0.014 sec–1, yielding 0.035 < κ s < 0.049 and 0.010 < κ p < 0.024 as likely values near Charleston. We favor the lower limits of these ranges because they imply numerically similar values for the path-average quality factors ( Q s = 32, Q p = 36), whereas the upper range values imply that Q s is substantially larger than Q p ( Q s = 22, Q p = 15).

The Influence of the Coastal Plain Sedimentary Wedge on Strong Ground Motions from the 1886 Charleston, South Carolina, Earthquake

The effect of Atlantic Coastal Plain sediments on ground motion from the 1886 Charleston, South Carolina, earthquake was studied using linear regression analysis and ground motion modeling. Statistical tests applied to 264 Modified Mercalli intensity reports at epicentral distances less than 400 km showed that intensities within the Lower Coastal Plain (near the coast) averaged from 0.5 to 1.0 units less than intensities within other areas over comparable distances. Simulated ground acceleration time histories were generated along two profiles: one trending northeast from the epicenter along the coast and another trending northwest toward the Fall Line. The modeling results (which assume linear behavior) indicate that the thick coastal sediments attenuate high frequency motions (>3-10 Hz), while amplifying lower frequency motions. In contrast, locations on thinner sediments to the northwest experience amplification of high frequency motions. The overall effect is that peak acceleration decays more slowly with distance in the thinning sediments to the northwest of the epicenter than in the thick sediments along the coast.