- © 2005 by the Seismological Society of America
A geologic investigation of the city of Memphis and southern Shelby County, Tennessee reveals Quaternary faulting. Structure contour maps and cross-sections of the top of the Pliocene-Pleistocene Upland Complex (Lafayette gravel), Eocene Upper Claiborne Group, and Eocene Lower Claiborne Group reveal two 20-m down-to-the-northwest faults that strike approximately N30°E The western fault is herein called the Memphis Fault and the eastern, the Ellendale Fault.
The Ellendale Fault is projected to pass beneath the Wolf River flood plain in east Memphis. At this location, a cut bank exposes flood-plain sands that are folded into an anticline with a wavelength of ∼ 100 m, an amplitude of ∼4 m, and a hinge oriented 10°N7°W. The folded sands are truncated by overlying, flat-lying, clayey silt overbank sediments. A 1-m-wide earthquake liquefaction sand dike was found above the crest of the anticline, in the overbank sediments. 14C dates reveal that the folding occurred between A.D. 390 and 450, and the liquefaction occurred post A.D. 450. Seismic-reflection lines indicate that the fold extends to more than 60 m depth into the Lower Claiborne Group, has a length of more than 1 km, and is thus tectonic in origin. The reflection lines also reveal a N25°E down-to-the-west fault with 5 m of displacement in the Upper Claiborne Group that we believe is within the Ellendale Fault zone. We believe the anticline formed during ∼5 m of Quaternary right-lateral strike-slip movement on the N25°E-trending fault. Based on these observations, we also believe that the Ellendale Fault, and perhaps the Memphis Fault, may pose seismic threats to the city of Memphis and Shelby County, Tennessee.
Several research projects (Broughton et al., 2001; Romero and Rix, 2001; Schneider et al., 2001; Cramer et al., 2003; Gomberg et al., 2003; Van Arsdale et al., 2003; Williams et al., 2003), currently focused on the city of Memphis and Shelby County, Tennessee, address hazards related to a New Madrid seismic zone (NMSZ) earthquake (Figure 1). The possible presence of Quaternary faults beneath Shelby County has not been addressed, however, The only detailed map of subsurface Shelby County is a U.S. Geological Survey Water-Resources Investigations map by Kingsbury and Parks (1993) that shows several faults displacing Eocene strata. In this current project, we geologically interpreted more than 600 geotechnical borings and water wells of southern Shelby County and made structure contour maps of the tops of the Pliocene-Pleistocene Upland Complex (Lafayette gravel), Eocene Upper Claiborne Group, and Eocene Lower Claiborne Group to determine if they have been displaced by faulting. In addition, we present drainage and paleodrainage analyses, a logged portion of the Wolf River bank, and seismic-reflection lines to interpret tectonic structures in Shelby County.
GEOLOGY OF THE NORTHERN MISSISSIPPI EMBAYMENT AND SHELBY COUNTY, TENNESSEE
The Mississippi embayment (Figure 1 inset) is a broad, south-southwest-plunging shallow trough filled with poorly consolidated Late Cretaceous and Tertiary shallow marine and fluvial sediments (Buschbach and Schwalb, 1984; Van Arsdale and TenBrink, 2000). Beneath the northwestern portion of the Mississippi embayment is the Reelfoot Rift (Figure 1), a southwest-trending graben approximately 70 km wide and 320 km long. This rift is a late Precambrian-Early Paleozoic aulacogen (Ervin and McGinnis, 1975, Braile et al., 1982). The rift geometry has been defined primarily through gravity and magnetic surveys (Hildenbrand et al., 1977; Hildenbrand and Hendricks, 1995). During the winter months of 1811-1812 at least three great earthquakes occurred within the Reelfoot Rift along the New Madrid seismic zone (NMSZ) (Johnston and Schweig, 1996) (Figure 1). The NMSZ is a right-lateral strike-slip system with a left compressional stepover that extends from Marked Tree, Arkansas into southeastern Missouri (Russ, 1982; Schweig and Ellis, 1994). Although NMSZ earthquakes occur primarily within the center of the rift (Figure 1), the southeastern margin of the Reelfoot Rift (Parrish and Van Arsdale, 2004) has been tectonically active in Quaternary time along the Crittenden County fault zone, 25 km northwest of Memphis (Crone, 1992; Luzietti et al., 1992; Williams et al., 1995). Additional Quaternary faulting has been identified within the southeastern margin of the Reelfoot Rift at Porters Gap, Tennessee (Cox et al., 2001) and 20 km north of Memphis (Williams et al., 2001; Cherryhomes, 2003). Cox et al. (2001) have proposed that this rift margin, which passes near Memphis, may currently be accumulating strain.
The stratigraphy of the upper 1 km in the northern Mississippi embayment (Figure 2) consists of Cambrian carbonate rocks, unconformably overlain by Upper Cretaceous marine and terrestrial sands, silts, and clays that are overlain by Tertiary marine and terrestrial sands, silts, and clays (Van Arsdale and tenBrink, 2000). The stratigraphy of the upper 360 m beneath the city of Memphis and Shelby County, Tennessee consists of the Eocene Claiborne Group, which is subdivided in ascending order into Memphis Sand, Cook Mountain Formation (clay, silt, and minor sand and lignite), and Cockfield Formation (sand, silt, and clay) (Kingsbury and Parks, 1993). The Eocene/Oligocene Jackson Formation exists beneath northwestern Shelby County and is included in the Upper Claiborne in this study. These Eocene to Oligocene stratigraphic units consist of near-shore marine and deltaic sequences (Cushing et al., 1964; Russell and Parks, 1975). In the Mississippi River valley, the Upper Claiborne is overlain by Mississippi River alluvium, but beneath Memphis and Shelby County the Upper Claiborne is overlain by the Pliocene-Pleistocene Upland Complex (Lafayette gravel) (Autin et al., 1991), a fluvial sand and gravel deposit that has erosional upper and lower contacts. The Upland Complex is overlain by up to three Pleistocene loess deposits and locally by Late Pleistocene and Holocene alluvium of the Wolf River and Nonconnah Creek in southern Shelby County (Figure 3).
NEW MADRID SEISMIC ZONE AND SHELBY COUNTY
Geological and geophysical evidence for neotectonic activity in the NMSZ region has been accumulating rapidly over the past two decades (Schweig and Van Arsdale, 1996; Cox et al., 200la, 200lb). Geological features indicative of earthquakes include liquefaction, faulting, and warping of the ground surface (Schweig and Van Arsdale, 1996). Earthquake recurrence intervals in the NMSZ are estimated to be approximately 500 years (Kelson et al., 1996; Tuttle et al., 2002), with documented earthquakes at A.D. 1811-1812, A.D. 1450 ± 150 years, A.D. 900 ± 100 years, and possible earlier events at approximately A.D. 300 and 500 B.C. (Tuttle et al., 2002).
Shelby County, Tennessee and the city of Memphis are located near the southeastern margin of the Reelfoot Rift and the NMSZ (Figure 1), previous regional studies have proposed faults beneath Shelby County (Kingsbury and Parks, 1993; Johnson et al., 1994). Johnson et al. (1994) derived a basement fault map from geophysical data, suggesting that Shelby County is transected by several basement tectonic structures. Kingsbury and Parks (1993) inferred faults from displaced Eocene Upper Claiborne Group strata in water wells. One of their inferred faults corresponds with a recently identified, northeast-striking, down-to-the-west fault called the Ellendale Fault which displaces the Upland Complex by as much as 20 m (Yates et al., 2001; Velasco et al., 2002).
Numerous sand dikes, formed by earthquake liquefaction, have been found in cut banks along the Wolf and Loosahatchie Rivers in Shelby County (Figure 3) (Van Arsdale et al., 1998; Broughton et al., 2001). The dikes probably formed during the 1811-1812 New Madrid earthquakes, as the dikes show no weathering effects, and some extrude onto the Wolf River flood plain (Broughton et al., 2001). Broughton et al. (2001) speculated that the liquefaction may have been caused by more local earthquakes, however, perhaps in Shelby County. This speculation by Broughton et al. (2001) is supported in recent work by Hough and Martin (2002), who have proposed that an estimated Mw 6.1 earthquake may have occurred in southern Shelby County during the New Madrid sequence of 1811-1812.
The purpose of our study was to determine if the Ellendale Fault displaces Tertiary and Quaternary sediments as proposed by Yates et al. (2001) and Velasco et al. (2002). We used more than 600 geotechnical boring and water-well logs to develop structure contour maps and geologic cross-sections of subsurface geologic units. In addition, we analyzed modern and paleostream patterns, logged a portion of the Wolf River bank containing a sand dike and anticline, and acquired and analyzed seismic-reflection profiles.
Structure Contour Maps and Geologic Cross-sections
We interpreted geotechnical boring and water-well logs from private geotechnical companies (Hall Blake and Associates, PSI, and Tristate), government organizations (U.S. Geological Survey, Tennessee Department of Transportation, Shelby County Health Department, U.S. Army Corps of Engineers, and the Tennessee Department of Environment and Conservation), and Ng et al. (1989). These well and boring data are available at http://gwidc.gwi.memphis.edu/website/introduction/. We made structure contour maps of the top of the Upland Complex, Upper Claiborne Group, and Lower Claiborne Group using ArcGIS and Arc/Info. The interpolation method used was inverse distance-weighted (Velasco, 2002). Inverse distance-weighted interpolation is a technique for estimating values of a locally dependent variable by determining cell values using a linearly weighted combination of a set of sample points. The weight is a function of the inverse distance between the estimation point and a known data point. We also constructed geologic cross-sections from the structure contour maps by exporting the ArcGIS format files into Microsoft Excel® for plotting.
Southern Shelby County Drainage and Paleodrainage Analysis
We interpreted the surface topography and drainage pattern of southern Shelby County using the National Elevation Data Set (NED) and the drainage patterns from the National Hydrography Data Set (NHD). These data were converted to an ArcGIS format and displayed for analyses in ArcMAP. The river patterns during Upland Complex deposition were determined from a structure contour map of the top of the Upper Claiborne Group. The paleodrainage networks were created from hydrology algorithms in ArcGIS that link progressive (ascending) grid cells (400 ft2) of the greatest calculated accumulation of surface flow. The resulting paleodrainage pattern represents accumulation grid cells that have 500 or greater contributors of flow.
Wolf River Bank Log
During early work on this project we found and logged a sand dike in a Wolf River cut bank in east Memphis that had been emplaced by earthquake liquefaction. Subsequent erosion removed the near-surface sand dike and exposed an underlying anticline which was also logged. Field relationships reveal that the sand dike emanated from the crest of the anticline. The flood plain sediments in and adjacent to the fold were 14C dated.
Shallow Seismic-reflection Surveys
Three shallow S-wave seismic-reflection lines were acquired in eastern Shelby County. Two of the lines were recorded on a 24-channel Geometrics Strataview seismograph, and one was recorded on a Bison 12-channel engineering seismograph (Table 1). In both cases we used a 10-kg sledgehammer energy source and single-component 4.5-Hz horizontal geophones. Figure 4 shows example shot gathers from S-wave profile Wolf1. We also acquired three shallow P-wave seismic-reflection lines at the same location using a 24-channel Geometrics Strataview seismograph, a 40-kg propelled energy source, and 30-Hz P-wave geophones.
Seismic-reflection field data were processed using WinSeis Turbo for Windows® (Kansas Geological Survey, 1994). Because seismic data quality (signal to noise ratio) was low, we restricted our seismic processing to a basic sequence to avoid processing artifacts that might lead to false geologic interpretations (Steeples and Miller, 1998). Table 2 shows the processing steps used for the S-wave seismic profiles. The P-wave and S-wave data had very consistent near-surface velocities, calculated from first arrivals (direct wave and refractions), and there were no elevation changes along the lines, therefore statics problems were not anticipated. Two-way travel times and approximate reflector depths are presented in the discussion of the seismic lines below. Depths were calculated from normal moveout velocities, and for the S-wave profiles using velocity data from a nearby refraction survey (http://www.ceri.memphis.edu/usgs/urban_map/geotech/williams12.shtml), and supported by depths calculated from our structure maps.
Structure Contour Maps and Geologic Cross-sections
Structure contour maps of the tops of the Upland Complex, Upper Claiborne Group, and Lower Claiborne Group for southern Shelby County (Figures 5, 6, 7, 8) show surfaces that slope from southeast to northwest. Additionally, the surfaces have two 20-m down-to-the-west steps striking approximately N30°E that we interpret to be the Ellendale and Memphis Faults. These faults are particularly well illustrated in the geologic cross-sections (Figures 9 and 10). The faults are prominent in the cross-sections because of the high vertical exaggeration, but this vertical exaggeration illustrates that the three contacts, separated in time by approximately 40 million years, are nearly parallel. Cross-sections 1-3 (Figure 10) illustrate that the Upland Complex is thin on the upthrown block east of the Ellendale Fault.
Southern Shelby County Topographic, Drainage, and Paleodrainage Maps
The topography of southern Shelby County reveals an increase in elevation across the Ellendale Fault (Figure 11A). The modern Wolf River and Nonconnah Creek flow from southeast to northwest across the county. Upon closer inspection, it is apparent that Wolf River and, to a lesser degree, Nonconnah Creek have three differently oriented reaches. Both streams flow westerly, turn and flow northwesterly, and then turn again to resume a western course to the Mississippi River. The northwest-oriented reaches of Nonconnah Creek and Wolf River flow across and at right angles to the Ellendale Fault.
Paleodrainage interpreted on the top of the Upper Claiborne Group reflects the drainage pattern during Pliocene-Pleistocene Upland Complex deposition (Figure 11B). Thus, we interpret the two major west-flowing paleodrainages in Figure 11B to be the ancestral Nonconnah Creek and the ancestral Wolf River. During Upland Complex time these streams apparently flowed northwest and merged near the location of the Memphis Fault.
Anticline and Liquefaction Sand Dike Exposed in the Northern Bank of the Wolf River
Field investigations conducted along the Wolf River identified flood-plain sediments consisting of point-bar sands overlain by an overbank clayey silt (Broughton et al., 2001). We found only two anticlines when canoeing the Wolf River across all of Shelby County. The eastern anticline is located in the boxed area of Figure 3 and is shown in Figure 12. The basal point-bar sand is folded into an anticline that has a wavelength of approximately 100 m, a minimum amplitude of 4 m (Figure 12A), and a hinge oriented 10°N7°W (Figure 12B). An overlying horizontal clayey silt overbank unit truncates the anticline (Figure 13B). A 1-m-wide sand dike was found in the crest of the anticline. The dike originated in the point-bar sand and intruded into the overlying flat-lying clayey silt to within 89 cm of the ground surface. Radiometric dates obtained from the flood-plain sediments reveal that the youngest folded-point bar sand is 1610 ± 60 BP and the flat-lying clayey silt is 1550 ± 40 BP (Figure 13A).
Six seismic-reflection profiles were acquired to determine the extent of the anticline exposed in the Wolf River bank and to reveal possible faults beneath or adjacent to the anticline (Figure 12). Our data quality varied because the site is adjacent to a highway, cultural noise is high, and the loose flood-plain sediments significantly attenuated the source. Only the three highest-quality lines are presented here (Figures 14, 15, 16).
Wolf1 is a 450-m-long S-wave line acquired on the south side of the Wolf River within 100 m of the river bank anticline (Figures 12B and 14). The earliest highlighted reflector is at 120 msec TWTT (∼ 15 m depth) and is interpreted to be the top of the Upper Claiborne Group. The latest highlighted reflector is at 300 msec TWTT (∼40 m depth) and is within the Upper Claiborne Group. An anticline is visible at CMP 175; the fold increases in amplitude and wavelength with depth. At CMP 210 the deeper reflector appears to be truncated by a fault. The anticline exposed in Wolf1 is not on strike with the N7°W anticline exposed in the northern bank of the Wolf River, therefore the Wolfl anticline may be a second anticline.
Shelby01 is a 720-m-long S-wave line (Figures 12B and 15). In this northern line, we highlight reflectors at 120 msec and 240 msec TWTT (∼15 m and ∼30 m). The 15-m-deep reflector is interpreted to be the top of the Upper Claiborne Group, and the 30-m-deep reflector is within the Upper Claiborne Group. A prominent anticline with a wavelength of 100 m and minimum amplitude of 5 m is evident from CMP 350 to the western end of the line. At CMP 145, the 30-m-deep reflector is displaced about 5 m down to the west, and the overlying reflector at 15-m depth is anticlinally folded.
The Agricenter line is a shallow P-wave reflection profile that was shot along the same route as Shelby01 to achieve a greater depth of penetration (Figures 12B and 16). An anticline is evident on the western end of the line that can be identified to a minimum depth of 60 m, within the Lower Claiborne Group. The fault identified on Shelby01 is not apparent on this line, probably because of the weaker quality data of the P-wave line.
DISCUSSION AND CONCLUSIONS
Structure contour maps of the top of the Upland Complex, Upper Claiborne Group, and Lower Claiborne Group reveal two down-to-the-west 20-m faults, called the Ellendale (eastern) and Memphis (western) Faults (Figure 9). The Ellendale Fault corresponds with one of the principal unnamed down-to-the-west faults mapped by Kingsbury and Parks (1993). Although it is possible that these steps are ancestral Mississippi/Ohio River terraces, we believe they are not terraces because the deepest displacements are on the top of the Lower Claiborne and, if erosional, would require spatially coincident Pliocene (∼5 Ma) and Eocene (∼40 Ma) terracing. Furthermore, the Claiborne Group is a near-shore marine and deltaic sequence that should not have northeast-trending terraces. The Memphis and Ellendale Faults appear to have been active intermittently since the Eocene because displacement on the top of the Lower Claiborne is generally greater than on the overlying Upland Complex across both faults and the Upland Complex is thinner east of the Ellendale Fault (Figure 10). The structure contour maps and cross-sections thus suggest post-Upland Complex Quaternary fault reactivation in southern Shelby County.
Topographic, drainage, and paleodrainage analyses of southern Shelby County also suggest Quaternary activity on the Hlendale and Memphis Faults. Surface elevation rises east-ward across the Ellendale Fault zone. Comparison of the modern surface drainage and the paleodrainage on the top of the Upper Claiborne Group indicates that the drainage pattern has changed since Pliocene-Pleistocene Upland Complex deposition. The paleodrainage pattern reveals that Wolf River and Nonconnah Creek merged in the proximity of the Memphis Fault during Upland Complex time and that the streams now follow two distinct but parallel courses. It appears that Quaternary displacement on the Memphis Fault is responsible for the paleodrainage alteration due to stream piracy upon profile steepening.
A Wolf River flood-plain bank exposure, near the surface projection of the Ellendale Fault, revealed a 10°N7°W-trending anticline with a wavelength of about 100 m and amplitude greater than or equal to 4 m. A 1-m-wide liquefaction sand dike emanated from the crest of the fold. 14C dates of the flood-plain sediments (Figure 13) reveal that the folding occurred between A.D. 390 and 450 and that the liquefaction dike was injected after A.D. 450.
East-west-oriented shallow seismic-reflection profiles were obtained to image the Ellendale Fault and to determine if the fold exposed in the Wolf River bank extended below the flood-plain sediments into the Eocene strata. We interpret folding that extends to a minimum depth of 40 m in the southern (Wolf1) seismic profile and to at least 30 m in the northern (Shelby01) profile. The P-wave Agricenter line, which was shot at the same location as Shelby01, reveals folding to a depth of at least 60 m. The Agricenter fold extends down through the Holocene WolfRiver flood-plain alluvium, Pliocene-Pleistocene Upland Complex, and Eocene Upper Claiborne Group and into the Eocene Lower Claiborne Group. The fold hinge trends N7°W between the Agricenter line and the riverbank fold, and the hinge is essentially perpendicular to the Wolf River flood plain. Considering the fold's wavelength, amplitude, greater than 1-km length, depth of folding, and trend, the fold could not be a Holocene sand bar, liquefaction deposit, or landslide feature. The best interpretation is that the 10°N7°W fold is tectonic in origin. Because the Wolfl fold extends to a depth of at least 40 m and appears to be bounded by a fault, we believe it too is of tectonic origin.
Faulting is revealed in seismic lines Wolfl and Shelby01. Connecting these displacements suggests a down-to-the-west fault that strikes N25°E (Figure 12B). The 5-m maximum displacement observed in the Shelby01 reflection line is less than the 20 m of down-to-the-west displacement across the Ellendale Fault observed in the structure contour maps. Thus, we believe that our seismic lines are on the eastern flank of the Ellendale Fault zone. The anticline exposed in the Wolf River bank and projected for more than 1 km has a hinge oriented 32° counterclockwise from the fault trace (Figure 12). This angular relationship suggests that the anticline may have been formed by right-lateral strike-slip motion (Harding, 1974; Lowell, 1985) on the underlying Ellendale Fault. To produce an anticline with a wavelength of 100 m and amplitude of 4 m requires approximately 3 m of shortening (Figure 12C). If the folding is due to strike-slip movement on the Ellendale Fault, which is compatible with the regional Quaternary stress field (Schweig and Ellis, 1994), then the Ellendale Fault has undergone 20 m of Quaternary dip-slip displacement and, at approximately A.D. 400, 5 m of right-lateral offset (Figure 12C).
We have not identified fault scarps or other tectonic landforms in Shelby County beyond those cited above, because the area is urbanized with extensive cut and fill. Although the Memphis and Ellendale Faults are aseismic, we believe that the data presented here support the conclusion that these faults may represent a significant seismic hazard to the city of Memphis and Shelby County.
This work was supported by the U.S. Geological Survey National Earthquake Hazards Reduction Program (NEHRP) under Contract No. 02HQGR0053. The research was also funded in part by the Earthquake Engineering Research Centers Program of the National Science Foundation under award number EEC-97O1785.