GEOPHYSICS, VOL. 61, NO. 6 (NOVEMBER-DECEMBER 1996); P. 1871–1882, 10 FIGS., 2 TABLES. Shear-wave splitting in Quaternary sediments: Neotectonic implications in the central New Madrid seismic zone James B. Harris ABSTRACT 8 ms at a two-way traveltime of 375 ms produced an Determining the extent and location of surface/near- average azimuthal anisotropy of 2% between the tar- surface structural deformation in the New Madrid get reflector (top of Quaternary gravel at a depth of seismic zone (NMSZ) is very important for evaluating 35 m) and the surface. Based on the shear-wave polariza- earthquake hazards. A shallow shear-wave splitting ex- tion data, two explanations for the azimuthal anisotropy periment, located near the crest of the Lake County in the study area are (1) fractures/cracks aligned in re- uplift (LCU) in the central NMSZ, shows the presence sponse to near-surface tensional stress produced by uplift of near-surface azimuthal anisotropy believed to be as- of the LCU, and (2) faults/fractures oriented parallel to sociated with neotectonic deformation. A shallow four- the Kentucky Bend scarp, a recently identified surface component data set, recorded using a hammer and mass deformation feature believed to be associated with con- source, displayed abundant shallow reflection energy on temporary seismicity in the central NMSZ. In addition records made with orthogonal source-receiver orienta- to increased seismic resolution by the use of shear-wave tions, an indicator of shear-wave splitting. Following ro- methods in unconsolidated, water-saturated sediments, tation of the data matrix by 40 , the S1 and S2 sections measurement of near-surface directional polarizations, (principal components of the data matrix) were aligned produced by shear-wave splitting, may provide valuable with the natural coordinate system at orientations of information for identifying neotectonic deformation and N35W and N55E, respectively. A dynamic mis-tie of evaluating associated earthquake hazards. INTRODUCTION of tectonic deformation (e.g., Zoback et al., 1980; Sexton and Jones, 1986; Odum et al., 1994). Although recent high- The relationship between geologic structure and contempo- resolution P-wave (compressional wave) investigations in the rary seismicity is one of the most important topics of research region have successfully imaged faults in poorly consolidated related to earthquake hazard evaluation in the New Madrid Tertiary sediments (Luzietti et al., 1992; Schweig et al., 1992; seismic zone (NMSZ), the most active earthquake zone in the Sexton et al., 1992; Van Arsdale et al., 1992), the Quaternary central and eastern United States. Determining the association section has been generally unresolvable. Consequently, studies between seismicity and structural deformation in the NMSZ is emphasizing the characteristics of S-wave (shear wave) prop- hindered by the presence of thick, unconsolidated sediments agation, which include increased seismic resolution in water- that cover the region. The upward continuation of basement saturated sediment sequences (i.e., Woolery et al., 1993) and faults into Quaternary strata is often masked by the inability of the phenomenon of shear-wave splitting (discussed in this pa- soft sediments to propagate large fractures, making identifica- per), may provide more suitable methods of investigating near- tion and age determination of near-surface geologic structures surface structure in the NMSZ and areas with similar geology. a difficult problem. Shear-wave splitting, induced by stress-aligned inclusions Seismic reflection methods have been used for many years (fractures, cracks/microcracks, pore spaces), causes shear within the NMSZ to determine the style, extent, and age waves to exhibit directional polarizations in response to Manuscript received by the Editor February 13, 1995; revised manuscript received January 31, 1996. Formerly Kentucky Geological Survey, University of Kentucky, Lexington, KY 40506-0107; presently Millsaps College, Department of Geology, 1701 N. State St., Jackson, MS 39210. c 1996 Society of Exploration Geophysicists. All rights reserved. 1871 1872 Harris propagation through azimuthally anisotropic media (Crampin, image near-surface geologic structure in the central NMSZ 1985; Thomsen, 1988; Tatham and McCormack, 1991). This (Harris et al., 1994). A four-component reflection test was phenomenon is manifested by differences in shear-wave veloc- completed using a hammer and mass energy source (i.e., ity between waves traveling parallel (S1—fast shear wave) and Hasbrouck, 1987). The objectives of the experiment addressed perpendicular (S2—slow shear wave) to the trend of the inclu- two questions: (1) could shear-wave splitting be identified on sions. The difference in arrival time between the two waves, reflection records in near-surface unconsolidated sediments? measured for a particular event, can be used to estimate the and (2) if shear-wave splitting was identifiable, could its pres- amount of anisotropy that exists along the path of the wave. ence be used as an additional tool in the evaluation of shallow When shear-wave splitting observations are made at the surface deformation associated with NMSZ seismicity? (i.e., reflection surveys), the last anisotropic horizon encoun- tered is responsible for the recorded polarizations. Reflection GEOLOGIC SETTING AND SEISMICITY experiments designed to evaluate shear-wave splitting typically record an orthogonal pair of S-waves, generally designated The location of the experiment was near the center of a SH-waves (horizontally polarized shear waves) and SV-waves region called Kentucky Bend, which lies inside a meander (shear waves polarized in a vertical plane). However, unless the loop of the Mississippi River in extreme western Kentucky seismic line is exactly parallel or perpendicular to the trend of (Figure 1). Geologically, the study area is situated in the upper the inclusions, the S-wave designations are only useful in de- Mississippi embayment, where Paleozoic bedrock is overlain scribing the acquisition geometry. Most commonly, the S-wave by approximately 600 m of Cretaceous to Recent unconsol- components must be rotated from the acquisition coordinate idated sediments, with local accumulations of 50 to 60 m of system into the natural coordinate system. Quaternary sand, silt, and gravel. The area is located in the A shear-wave splitting experiment was carried out as part central NMSZ near the crest of the Lake County uplift (LCU), of a shallow SH-wave seismic reflection program designed to the most prominent known example of surficial deformation in FIG. 1. The pattern of contemporary seismicity and regional tectonic setting of the New Madrid seismic zone (modified after Luzietti et al. 1992). Inset shows the location of the study area in relation to the Lake County uplift (modified after Russ 1982). Shallow Shear-wave Splitting: NMSZ 1873 the NMSZ. Russ (1982, 97) described the LCU as “...a gently and radial source-radial receiver (SV-SV) reflection records sloping, irregularly shaped topographic bulge whose surface taken along a 75-m-long test line oriented N85W (see Figure 1 has been upwarped as much as 10 m above the general level for location). The data were recorded on a 12-channel engi- of the Mississippi River Valley.” Along the eastern edge of the neering seismograph using a 5.4-kg sledgehammer/10-kg steel LCU lies the Reelfoot Lake Basin. Separating the LCU from I-beam energy source and single-component 30-Hz horizontal the basin is the north-south–trending Reelfoot scarp, which geophones. The geophones were first oriented transverse to the Russ (1979) described as a complex monoclinal fold of 3 to 9 m seismic line (SH-mode), and recordings were made of horizon- in height. Recent paleoseismologic trenching on the central tally directed transverse and radial hammer blows (five blows Reelfoot scarp (Kelson et al., 1992; 1994) has found evidence on each side of the I-beam were stacked using the polarity re- for at least three moderate to large earthquakes within the past versal feature of the seismograph). The line was then reshot 2200 years, including the great earthquakes of 1811–1812. with radial geophone orientation (SV-mode), and both trans- The four major earthquakes that occurred in the cen- verse and radial hammer blows were repeated. Table 1 details tral Mississippi River Valley between December 1811 and the seismic data acquisition parameters used in the experiment. February 1812 constitute the largest sequence of earthquakes Examination of the field records showed that records made in recorded North American history. The magnitudes of the when the receivers were oriented perpendicular to the source events were estimated to range from mb,Lg 7.0 to mb,Lg 7.3 motion contained as much or more S-wave reflection en- (Nuttli, 1973; Street, 1982). The earthquakes, centered near ergy than the recordings made with coincident source-receiver New Madrid, Missouri, caused ground failure over an area of orientations (Figure 3). This observation is consistent with in- 48 000 km2 (Fuller, 1912) and damaged structures as far away terpreted shear-wave splitting from previous multicomponent as Cincinnati, Ohio, and St. Louis, Missouri. In the 3 1/2-month reflection surveys (Alford, 1986; Willis et al., 1986; Squires period following the first earthquake, more than 200 after- et al., 1989; Mueller, 1991), and suggested that the near-surface shocks greater than mb,Lg 5.0 (with at least six greater than structure in the study area was azimuthally