ABSTRACT

We investigate the crustal structure in the Mississippi embayment and adjacent areas using teleseismic receiver functions. Specifically, we use three-component broadband observations from three seismic stations of the Cooperative New Madrid Seismic Network (CNMSN) in the central United States. Each site is located in a different near-surface geological environment: SLM is located in eastern Missouri on Paleozoic sedimentary rocks, BLO is located on the flank of the Illinois Basin, and MPH is located in the Mississippi embayment above a thick sequence of late Cretaceous and Cenozoic unconsolidated and poorly consolidated sediments. Crustal velocity models estimated using observations from the three stations are relatively simple, being characterized by smooth velocity variations through the middle and lower crust and by similar crustal thickness (40-43 km). Our analysis verified out differences in the shallow crustal structure at the three station sites. Large arrivals produced at the sharp transition from relatively fast bedrock to the thick coverage of low velocity sediments dominate the observations at station MPH. Beneath MPH, the average sediment shear-velocity is about 0.8 km/sec; a relatively sharp transition brings the shear velocity up to about 2.6-2.8 km/sec in the basement underneath. The upper crust at BLO consists of 500-m Pennsylvanian-age clastic sedimentary rocks with a shear-velocity of about 1.5-1.7 km/sec. The shallow underneath SLM crust consists of about 1000 m Ordovician and a Cambrian carbonate stratum with a shear-velocity of about 2.5-2.7 km/sec and it reaches 3.0-3.5 km/sec under the site. This study may be the first to use the receiver function technique to focus on very shallow structure, which is important for ground motion amplifications and strong-motion hazard estimation in the studied regions.

INTRODUCTION

Teleseismic receiver functions generally provide valuable information on the largest velocity contrasts in the local structure, i.e. the crust-mantle transition and the sediment-basement boundary [e.g. Langston, 1979; Owens et al., 1984; Ammon and Zandt, 1993] Often detail regarding intra-crustal variations can be extracted from the observations, but for stations resting on thick sediment sequences, information from the deeper structure may be masked by multiples within the sediments [Kind et al., 1995]. These multiples, however, carry shear-velocity information valuable to those interested in strong-motion estimation and seismic hazard analysis. The amplitude and frequency content of the ground motions from on earthquake can be greatly affected by properties (e.g. impedance contrasts) and configuration of the near surface materials via mechanisms such as damping and focusing. Street et al., [1997] observed that the impedance contrast at the soil/bedrock interface has a strong effect on the resulting ground motions at the surface. The problem of the site effects is a particular importance too much of the central United States because of the thick deposits of unconsolidated sediments in the embayment. Complicating the assessments of the site affects in the Mississippi embayment is the lack of detailed shear wave velocity profiles of the sediments down to the bedrock. We present the results of our use of receiver functions to delineate the variations in seismic velocity structure in the south-central conterminous United States (CUS). Specifically, we estimate and compare the structures beneath three seismic stations: SLM, BLO, and MPH, located at St. Louis, MO, Bloomington, IN, and Memphis, TN, respectively. We show how receiver functions can contribute useful information for strong-motion hazard estimation, particularly in regions with thick sediments but little deep borehole information.

Stations SLM and BLO are located on the mid-continent platform (Paleozoic sediment rest on Precambrian basement), and MPH is located on the Mississippi Embayment, a large sedimentary basin located in the south-central CUS. We focus our attention on the shallow crust, and use the platform stations to contrast site effects observed on records from MPH and caused by a thick package unconsolidated sediments. To minimize the effects of the receiver-function trade off between absolute velocity and depth to the impedance [Ammon et al., 1990], we include in the analysis some independent results from reflection and borehole data.

GEOLOGIC SETTING OF THE SOUTH_CENTRAL CONTERMINOUS U.S.

The Mississippi Embayment is the site of a late Precambrian Continental rift that was reactivated in the Mesozoic [e.g. Kane et al., 1981]. The southward plunging trough is filled with sediments that thicken from a thin margin in southern Illinois to about 1000 meters in southwestern Tennessee [Stearns and Marker, 1962] and grade smoothly into the Gulf Coastal Plain [Mooney et al., 1983]. These sediment deposits are unconsolidated and poorly consolidated, age from late Cretaceous to present, and uncomformably overlay Paleozoic carbonate and clastic rocks [Andrews et al., 1985; Braile, 1989]. Other important regional structures include the Illinois basin to the north, the Ozark uplift to the Northwest, and the Nashville dome and Cincinnati arch to the Northeast (Figure 1). The Ozark uplift is composed of Precambrian crystalline rocks and was a site of active faulting in the Devonian [Stewart, 1968; Braile, 1989]. The Illinois basin contains up to 1.5 km of Pennsylvanian age clastic sedimentary rocks and older Paleozoic carbonates. The Cincinnati Arch adjoins the basin on the east, has generally existed since Precambrian time. It is low and broad with several components, including the Nashville Dome.

Although several studies of regional structure exist, detailed information is limited. Stewart [1968] interpreted refraction profiles across Missouri and Baldwin [1980] reported the results of similar work in Southern Illinois and Indiana. Herrmann [1969] modeled fundamental and higher mode surface-wave group velocities for paths along the eastern flank of the basin and Cincinnati Arch, and constructed a three-layer model for the path near station BLO which included a 500 m thick top layer with a shear-wave velocity of 2.16 km/sec, overlying a 2 km thick layer with 3.14 km/sec shear-wave velocity. Woods et al., [1989] and Woods [1986] used portable arrays of analog stations to sample both the western shelf and the interior regions of the Illinois basin. They used fundamental mode group velocity and inter-station phase velocities obtained from the arrays, to estimate a Vs of 1.9 km/sec for the first 500 m depth, and 2.8 km/sec for the remaining kilometer of sediments in the Illinois basin. They found Vs of 2.5 km/sec was found for the upper kilometer the Ozark uplift, and beneath those, Vs of 3.5 km/sec. Langston [1994] used seismograms recorded at the IRIS station Cathedral Caves, Missouri (CCM) to study the nature of wave propagation along paths from NMSZ across the Ozark uplift. His results indicate a 40 km thick crust, characterized by a relatively smooth-velocity gradient at the Moho.

The Mississippi embayment has been the site of many reflection and refraction surveys [Braile et. al., 1989]. Sediment thickness within the New Madrid Seismic Zone (NMSZ) was estimated by linear extrapolation from well-log data [Dorman and Smalley, 1994], interpretation of seismic reflection profiles [Mooney et al., 1983; Andrews et al., 1985; Street et al., 1995] and travel-time differences between the direct and converted waves generated at the base of the sedimentary section [Hough 1990, Chen et al., 1996]. The depths from surface to Paleozoic bedrock obtained by Dart, [1990] from 35 drill holes where located inside the PANDA array. The sediment thickness beneath each PANDA stations was estimated by linear interpolation that varies from 405 m to 860 m. Street, [1995] obtained the maximum sedimentary thickness as 640 m and 767 m from 11 of the strong motion stations along a north-south axis in the New Madrid zone by means of P and SH wave seismic reflection and refraction techniques. These are good agreement between the thickness of the sediments based on seismic data and those shown Dorman and Smalley, [1994]. Chen et al. [1996], used data from PANDA stations in the Northern Mississippi embayment, obtained the time difference between the direct P and the Sp wave in the 0.57-0.88 sec range, which correspond to 460 to 715 m thickness, respectively, assuming the sediment velocity of Andrews et al., [1985]. Bodin & Horton [1999] observed 960 m sedimentary thickness with shear wave velocity of 834 m/s under the Memphis site using the Horizontal/Vertical spectral ratio of braodband microtremor measurements.

RECEIVER FUNCTION ANALYSIS AND CRUSTAL STRUCTURE

Radial and transverse receiver functions are computed by deconvolving the vertical component seismogram from the radial and the transverse. This process eliminates near-source and instrument effects and produces two time histories characterized by a series of pulses, each corresponding to the arrival of a converted wave or shear-wave reverberation from the local structure. The first 20-30 seconds of the receiver functions contains information about the crustal and uppermost mantle velocity structure beneath the receiver. Several investigators review the use of receiver functions to estimate earth structure beneath three component broadband seismic stations [e.g.: Langston, 1979; Owens et al., 1984; Ammon et al., 1990]. Receiver function analysis is most successful in regions of relative lateral homogeneity [e.g. Owens et al., 1984] but even in ideal situation, receiver function dependence on velocity structure is non-unique [Ammon et al., 1990]. We relied on refraction and borehole information to minimize the ambiguity.

We calculated the receiver functions using a time-domain iterative deconvolution technique described by Ligorria and Ammon [1999] and based on the earthquake modeling approach of Kikuchi and Kanamori [1982]. Iterative time-domain inversion procedure solves directly for the velocity model at each iteration. It is faster and enables smoothness constraints to be placed on the resulting velocity models. We invert velocity structures consisting of horizontal layers with fixed thickness. 2-km thin layer is used throughout the model, which provides some freedom in where precisely the larger velocity contrasts will occur. Using the shallow and deep borehole velocity information, we add finer detail to the surface about 1 km of depth with 100-200 m thin layers. Number of iteration is taken as ten, which is convenient to allow the calculations converge enough to the observations. We have used smoothness constraint of 0.01. Therefore, inversion models would not display oscillatory behavior with depth. Since a large bandwidth is needed to follow velocity changes in the thin layers at shallow depths, we model our observations in two bandwidths corresponding to Gaussian filter width factors of 2.5 and 5.0, roughly equivalent low-pass cutoff frequencies of about 1.2 and 2.4 Hz, respectively.

We assumed a Poisson`s ratio of 0.25 for the crust beneath BLO and SLM but use a value of about 0.40-0.45 for the first km in the MPH model. We fixed the model densities using the relation density = 0.32Vp+0.77 for lower part of crust and Paul Mayne's density relation, density = 0.8*log10 (Vs)-0.1 for shallow part of the Memphis site. Each seismogram was analyzed independently without stacking and the estimated crustal structures were averaged to obtain a final crustal structure for each site. The initial velocity models used in the inversion (Figure 2) were modified version of the HAMBURG model of Herrmann and Ammon [1996], obtained from regional waveform modeling of seismograms recorded at station CCM (Cathedral Caves, MO). For the inversion of BLO and SLM receiver functions, we modified the upper layer of the HAMBURG model guided by the fundamental and higher mode group velocity observations of [Herrmann, 1969]. For MPH, the upper km of the HAMBURG velocity model was modified using constraints from shallow [NUREG/CR-0985, 1980] and deep borehole [USGS, Linda well] studies of the MPH site.

Information on the twelve earthquakes is listed in Table 1. Twenty seconds of the radial receiver functions are shown in Figures 3, 4, and 5. The BLO and SLM radial receiver functions are simple, characterized by small Ps conversions, indicative of a crust and uppermost mantle with no sharp transitions. Shallow reverberations between the surface and the sediment-basement contact dominate the MPH receiver function.

Inverted velocities from the receiver functions were averaged at each depth and site to obtain a final velocity structure. The results of our receiver function inversions are summarized in Figure 6. The deep structure is relatively simple, with a crust-to-mantle transition at depths of about 40-45 km. The deeper value corresponds to MPH, which may be less reliable because of the dominance of the near-surface structure in the observed receiver functions. Because of the broad pronounced pulse at the beginning of the records is a consequence of thick low-velocity sedimentary layer, we can not see clear conversion from Moho. Since both our data and scope are limited, we focus attention on the near-surface structures, which vary substantially. Most of the differences in the first several seconds of the MPH radial receiver functions are due to variations in the velocity structure of the sedimentary layer, which lead to variations in the constructive/destructive interference pattern between the near surface multiples. Station SLM is underlain by relatively high velocity sequence with a thickness of about one-km. The velocity of the material beneath BLO is similar to that beneath SLM, but half as thick, about 0.5 km. The crust beneath MPH has a near-surface shear velocity of about 0.8 km/sec, underlain by a sharp, 2.0-km/sec velocity contrast located about one km beneath the surface. This transition undoubtedly represents the transition from poorly or unconsolidated sediments to Paleozoic sedimentary basement rocks.

CONCLUSIONS AND DISCUSSIONS

The dominant feature MPH receiver functions are the reverberations between the surface and the bottom of a very low-velocity sediment layer. Seismic hazard calculations utilize attenuation and soil amplification factors, which includes the effects of crustal structure and the surface geology, soil column thickness. Therefore, the shear wave velocities of the bedrock and overlying soils must be known for successfully modeling the ground motions at a site. In this study, receiver functions used to constrain shallow shear wave velocities, which are important for site amplification and help since we have no deep well logs. We demonstrate the sensitivity of receiver functions to the low velocity surface layers, which underlay the bedrocks and how receiver functions carry useful information for seismic hazard estimation in a region. Crustal velocity models estimated using receiver functions from seismic stations BLO, SLM and MPH all show a 40-45 km thick crust with a relatively smooth velocity transition. These results are consistent with previous work refraction profile analysis described by Stewart [1968], Barille, [1989] and receiver function analysis of CCM Langston [1994]. The average crustal structures in the embayment consists of a one km thick 0.8 km/sec loosely consolidated Tertiary and cretaceous sedimentary deposits, five km thick ~ 2.5-3.0 km/sec Paleozoic carbonate and clastic sedimentary rocks, 10-15 km thick ~3.4-3.6 km/sec, granitic upper-crust and 12 km thick ~3.7-3.9 km/sec metamorphic lower crust and 10 km thick ~4.2-4.3 km/sec modified lower crust. Modified lower crust, which defined by shear wave velocity of 4.2-4.3 km/sec, is observed at three sites with a smooth gradient at the Moho. The presence of high velocity basal crust with a ~4.3 km/sec shear velocity underlain beneath the embayment indicates that the lower crust has been altered by the injection of mantle material, consistent with the hypothesis of Mooney et al. [1983].

New information can be provided by the receiver function method for the shear-velocity structure in the crust, which is only rarely obtained from other studies. Large data set, which is recorded by denser broadband array in/around Mississippi embayment in the future, could improve these results.

ACKNOLEDGEMENTS

We would like to the thank Eric Haug for his support and efforts installing the seismic stations used in this analysis. This work was supported by Saint Louis University and NSF under the Mid-America Earthquake Center Contract ___________. Instrumentation support was derived from USGS contract ____________.

REFERENCES

Andrews, M. C., W. D. Mooney, and R. P. Meyer, The relocation of microearthquakes in the northern Mississippi embayment, J. Geophys. Res. 90, 10223-10236, 1985.

Ammon, C. J., G. E,. Randall and G. Zandt , On the Nonuniqueness of Receiver Function Inversions, J. Geophys. Res., 95, 153003-15318, 1990.

Ammon, C. J., and G. Zandt, The receiver structure beneath the southern Mojave Block, Bull. Seism. Soc. Am., 83, 737-755, 1993.

Baldwin, J. L., A crustal seismic refraction study in Southwest Indiana and Southern Illinois, M.S. thesis, Purdue University, Lafayette, IN, 1980.

Bodin, P. and Horton, S.,  Broadband microtremor observation of basin resonance in the Mississippi embayment, Central US, Goephy. Res. Lett., 26,7, 903-906, 1999

Braile, L. W., G. R. Keller, W. J. Hinze, and E. G. Lidiak, An ancient rift complex and its relation to contemporary seismicity in the New Madrid seismic zone, Tectonics, 225-237, 1982.

Braile, L.W., Crustal structure of the continental interior, in L. C. Pakiser and W. D. Mooney (Editors). Geophysical Framework of the Continental United States, Geological Society of America Memoir 172, Boulder, Colo.

Braile, L.W., W. J.,Hinze, R.R.B. von Frese, and G.R. Keller, Seismic properties of the crust and uppermost mantle of the conterminous United States and adjacent Canada, in L.C. Pakiser and W.D. Mooney, (Editors), Geophysical Framework of the Continental United States, Geological Society of America Memoir 172, Boulder, Colo.

Chen, K. C., J. M. Chiu, and Y. T. Yang, Shear-wave velocity of the sedimentary Basin in the upper Mississippi Embayment using S _to P converted waves, Bull. Seis. Soc. Am., 86, 848-856, 1996.

Dorman J., and Smalley, R., Low frequency seismic surface waves in the New Madrid zone, Seism. Res. Lett. 65, 137-148, 1994.

Hough, S. E., Constraining sediment thickness in the San Francisco Bay area using observed resonances and Ps conversions, Geophys. Res. Lett., 17, 1469-1472, 1990.

Herrmann, R. B., The structure of Cincinnati Arch as determined by short-period Rayleigh waves within the Ozark Uplift and Illinois Basin. Seism. Res. Let. 60, No:3, 111-118, 1969.

Herrmann, R. B., Broad band seismology and small regional seismic networks, in K.M. Shedlock and A.C. Johnston (Editors), Investigations of the New Madrid Seismic zone, US Geological Survey Prof. Paper 1538 (in draft), 1995.

Herrmann, R. B. and Ammon, C. J., Faulting parameters of earthquakes in the New Madrid, Missouri region, Engineering Geology, 46, 299-311, 1996.

Kane, M. F., T. G. Hildebrand and J. D. Hendricks, A model for the tectonic evolution of the Mississippi embayment and its contemporary seismicity, Geology, 9, 563-567, 1981.

Kind, R., Kosarev, G. L. and Peterson, N. V., Receiver functions at the stations of the German Regional Seismic Network (GRSN). Geophys. J. Int., 121, 191-202, 1995.

Kikuchi, M. and H. Kanamori, Inversion of complex body waves. Bull. Seis. Soc. Am., 72, 491-506, 1982.

Langston, C. A., Structure under Mount Rainier, Washington, inferred from teleseismic body waves, J. Geophys. Res., 84, 4749-4762, 1979.

Langston, A.C., An integrated study of crustal Structure and Regional wave propagation for Southern Missiouri, Bull. Seis. Soc. Am., 84, 105-118, 1994.

Ligorria, J. P., and C. J. Ammon, Iterative deconvolution of teleseismic seismograms and receiver function estimation, Bull. Seism. Soc. Am., in revision, 1999.

Mooney, W. D., M. C. Andrews, A. Ginzburg, D. A. Peters and R. M. Hamilton, Crustal structure of the northern Mississippi embayment and a comparison with other continental rift zones, Tectonophysics, 94, 327-348, 1983.

Owens, T. J., G. Zandt, and S. R. Taylor, Seismic evidence for an ancient rift beneath the Cumberland Plateau, TN: A detailed analysis of broadband teleseismic P-waveforms, J. Geophys. Res., 7783-7795, 1984.

Street, R., Wang, Z., Woolery, E., Hunt, J., Harris, J., Site effects at vertical accelerometer array near Paducah, Kentucy, Engineering Geology, 46, 349-367, 1997.

Street, R., Woolery, E., Wang, Z., Harris, J., A short note on shear-wave velocities and other site conditions at selected strong-motion stations in the New-Madrid seismic zone, Seis. Res. Lett., 66, 56-63, 1995.

Stearns, R.G. and M. V. Marcher, Late Cretaceous and subsequent structural development of the northern Mississippi embayment area., Geol. Soc. Am. Bull., 73, 1387-1394, 1962.

Stewart, S. W., Crustal structure in Missouri by seismic refraction methods, Bull. Seism. Soc. Am., 58, 291-323, 1968.

Woods, M. T., Shallow crustal structure as inferred from short period Rayleigh wave dispersion, M.S. Thesis, Saint Louis University, St Louis MO, 1986.

Woods, M. T., Russell, D. R., Herrmann, R. H., Dispersion of short period Rayleigh waves within the Ozark Uplift and Illinois Basin, Seis., Res., Let., 60, No:3, 111-118, 1989.

Figure Captions

Figure-1 Overall view of the main units of the north central United States showing the major basins, arches and domes, location of the three stations-sites in the north Central Unites States beneath which the crustal structure has been determined using teleseismic receiver functions.

Figure-2 Initial velocity models used in the inversion for SLM, BLO and MPH sites (a) with a detailed upper 2 km crustal section (b).

Figure-3 SLM inversion results. Radial receiver functions at different frequency contents. Traces in each plot from left to right ranges from a=2.5 to a=5.0. Solid lines refer to observed receiver function and dashed lines to the synthetic. The inversion velocity models are shown on the right as dashed lines and the solid line indicates starting model. The inversion velocity models are also given with the shallow part of the crustal model.

Figure-4 BLO inversion results. Radial receiver functions at different frequency contents. Traces in each plot from left to right ranges from a=2.5 to a=5.0. Solid lines refer to observed receiver function and dashed lines to the synthetic. The inversion velocity models are shown on the right as dashed lines and the solid line indicates starting model. The inversion velocity models are also given with the shallow part of the crustal model.

Figure-5 MPH inversion results. Radial receiver functions at different frequency contents. Traces in each plot from left to right ranges from a=2.5 to a=5.0. Solid lines refer to observed receiver function and dashed lines to the synthetic. The inversion velocity models are shown on the right as dashed lines and the solid line indicates starting model. The inversion velocity models are also given with the shallow part of the crustal model.

Figure-6 Final version of velocity structures obtained and averaged from inversion processes of receiver functions at three sites. Inverted velocity models are shown by dashed, solid line show starting models. . In this work, we compare the shallow and deeper part of the crustal structures of three station sites, which are located in geologically different area