The 2003 event was a magnitude MbLg = 4.0 event located 1.42 km north of Blytheville, Arkansas. The 2004 event was a magnitude MbLg = 3.4 event located ...
Proceedings of the 8th U.S. National Conference on Earthquake Engineering April 18-22, 2006, San Francisco, California, USA Paper No. 1358
VALIDATION OF GROUND MOTIONS IN THE UPPER MISSISSIPPI EMBAYMENT B.K. Lawrence1, G.J. Rix2 and P.W. Mayne2
ABSTRACT The New Madrid Seismic Zone (NMSZ) is a seismically active region that dominates the seismic hazard of the central United States. A key geologic feature in the NMSZ is the Upper Mississippi Embayment (UME), which is composed of deep soil sediments (z < 1200 m). A field testing program consisting of surface wave and seismic cone penetration tests (SCPT) was conducted at several broadband stations in the UME. Shear wave velocity profiles for depths greater than the penetration depth of testing were taken from suspension-type logging tests at other sites in the UME. Shear wave velocity profiles are used as input into a linear site response program using two different small-strain damping models. Response spectra from small (approximately MbLg ≤ 4) earthquakes in the UME at these sites are compared with response spectra estimated from the two damping models to evaluate the ability of these models to predict ground motions given new site characterizations in the upper 30 to 50 meters. Given the lack of observed strong ground motions in the central U.S., these comparisons using weak motions are an important step in validating ground motion predictions for the NMSZ. Introduction The New Madrid Seismic Zone (NMSZ) is an area that has experienced large earthquakes up to M = 7.8 (Bakun and Hopper, 2004) on an infrequent basis, which makes validation of earthquake models for major events a very challenging task. Much of the surrounding region is underlain by thick sediments known as the Upper Mississippi Embayment (UME), which reach up to 1200 meters thick in the study area, although the southern end reaches up to 1.6 km in thickness. Validation of even small to moderate earthquake ground motions is of interest, as these events occur on a more frequent basis and allow for some measure of verification of predictive models. A field testing program was undertaken in 2004-2005 at a series of selected sites within the NMSZ. These sites were selected because they had broadband instruments located in the 1
Graduate Research Assistant, School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0355 2 Professor, School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 303320355
UME and were within a reasonable distance from recent small-to-moderate earthquake epicenters. These stations are operated by either the Center for Earthquake Research and Information (CERI) at the University of Memphis or Saint Louis University (SLU). Subsurface information obtained from field testing was used to develop input parameters for the site response program RASCALS (Response Spectrum and Accelerogram Scaling: Silva, 1998). A unique aspect of this study is that the upper portion of the shear wave velocity (Vs) profiles was characterized at the same sites where small-magnitude earthquake motions were recently recorded. Although the measurements are not from large earthquakes, this nonetheless provides significant new data for validation of ground motion predictions from small earthquakes in the NMSZ and UME. Station Descriptions The stations visited in this study were broadband stations in the CERI and SLU networks. The location, name and elevation of the stations are described in Table 1. The first eleven stations in Table 1 are part of the CERI network. The remaining three stations (UTMT, MPH and PLAL) are part of the SLU network. Broadband seismometers consist of three-component CMG40T sensors. Table 1.
Identification information for stations visited for Vs profiling Station Name GLAT GNAR HALT HBAR HENM LNXT LPAR PARM PEBM PENM HICK UTMT MPH PLAL
Longitude -89.2876 -90.0178 -89.3395 -90.6572 -89.4717 -89.4913 -90.3002 -89.7522 -89.8623 -89.6280 -89.2288 -88.8642 -89.9320 -88.0750
Latitude 36.2694 35.9652 35.9106 35.5550 36.7160 36.1014 35.6019 36.6635 36.1131 36.4502 36.5409 36.3423 35.1240 34.9822
Elevation (m) 120 71 85 74 88 143 67 85 76 85 141 120 93 165
Earthquake Motions Ground motions for this study resulted from earthquake events that occurred on April 30, 2003 and June 15, 2004. The 2003 event was a magnitude MbLg = 4.0 event located 1.42 km north of Blytheville, Arkansas. The 2004 event was a magnitude MbLg = 3.4 event located 1.56 km west of Big Ridge, Missouri.
The two sites chosen for site response analysis in this paper are GLAT, a site on 770 meters of Pleistocene deposits, and LPAR, a site on 867 meters of Holocene deposits. The epicentral distances to GLAT were 67.1 and 62.1 km for the 2003 and 2004 earthquakes, respectively. The epicentral distance to LPAR was 136.7 km for the 2004 earthquake. Data were not available at LPAR for the 2003 earthquake. Displacement time histories provided by CERI were corrected for station instrument responses. The original data represents ground displacement in microns over a frequency band of 0.1 Hz to close to the Nyquist frequency. Ground motions were sampled at either 20 or 100 samples per second (sps), which translates to Nyquist frequencies of 10 or 50 Hz, respectively. Since the peak response spectral amplitudes are near 10 Hz for most records, the stations sampled at 100 sps provide much better accuracy in the higher frequency portion of the record. The resultant of the two horizontal motions from each earthquake record was used for comparison to results of site response models. The resultant of the two horizontal components was calculated using the program “gmeanrot50” developed by Dr. David Boore at the U.S. Geological Society (USGS) and available at the website http://quake.wr.usgs.gov/~boore/. The algorithm implemented in the program iteratively computes the geometric mean of the two horizontal components through a total of ninety, 1-degree rotations. The computed resultant is the median value of these ninety geometric means. Field Testing Cone Penetration Testing (CPT) Cone penetration testing was performed using an instrumented electronic steel probe to collect detailed subsurface information. The tests were conducted by hydraulically pushing the penetrometer at a constant rate of 20 mm/s with stops at 1-m rod intervals. The CPT testing at the station sites was performed using a 10-cm2 cone outfitted to measure tip resistance (qt), side resistance (fs), porewater pressure at the u2 (shoulder) position and Vs. Penetrometer measurements are automatically collected at 25-mm increments, and downhole Vs obtained at 1meter increments. Shear wave velocities were calculated using a pseudo-interval method and a repetitive horizontal surface source. An example of a CPT sounding log for the LPAR site is shown in Fig. 1. Array-Based Surface Wave Testing Array-based surface wave testing requires a source, an array of receivers, and a data acquisition and processing system (Yoon, 2005). In this study, the source for active harmonic tests was an APS Dynamics, Inc. Model 400 Electro-Seis electromechanical device. A program controlled the shaker to operate at 76 frequencies between 3.125 to 100 Hz. Array spacing was non-uniform. For approximately half of the tests, the spacing of the linear array of 15 receivers was: 2.44, 3.05, 3.66, 4.57, 5.49, 6.71, 8.53, 10.36, 12.80, 15.24, 18.29, 21.34, 24.38, 28.96 and 33.53 meters. The remaining tests had the same receiver-toreceiver spacing, but the source-to-1st receiver spacing was increased from 2.44 to 8.53 meters.
Figure 1. CPT data from LPAR site in Lepanto, Arkansas In order to obtain a Vs profile, an inversion of the dispersion data obtained from the field measurements was performed. Inversion procedures are based on either fundamental mode or multiple mode surface waves, depending on the complexity of the field dispersion data. Shear Wave Velocity Profiles Field measurements were used to define Vs profiles down to the depth that was tested (typically the upper 30 to 50 meters). Fig. 2 shows field measurements of Vs in addition to the interpreted Vs profile. Our interpretation of the profile attempted to balance the benefits of the smaller-scale accuracy of the SCPT with the larger-scale trends of surface wave tests. Below the depth of field testing, each site used one of two generic soil profiles for the UME: lowlands (Holocene) or uplands (Pleistocene). These models were taken from Romero (2001) and have been updated based on data from borehole logging tests performed in Arkansas and Tennessee, as well as data from Street, Woolery, and Chiu (2004). The crustal model below the soil profile was interpreted from Catchings (1999). Shear wave velocities vary from 2040 to 4510 m/s for the crustal model. Calculation and Comparison of Response Spectra A goal of this research is to compare response spectra generated by different small strain damping profiles to response spectra calculated from earthquake motions measured in the UME.
GLAT Test Site Obion, TN
LPAR Test Site Lepanto, AR
Vs (m/s) 200
Vs (m/s)
300
400
100 0
10
10
20
20
SCPT 30
Surface Wave Combined
Depth (m)
Depth (m)
100 0
200
300
400
SCPT 30
SurfaceWave Combined
40
40
50
50
60
60
Figure 2. Vs profiles from field tests with interpreted profiles The program used for these calculations was RASCALS, a 1-D equivalent-linear program that computes the Fourier amplitude spectrum at the earthquake source based on the method suggested by Brune (1970) and computes peak ground motion parameters and response spectra based on the method suggested by Boore (1983) using random vibration theory (RVT) techniques. Although RASCALS is able to perform equivalent-linear analyses, a linear analysis was used because the observed ground motions were so low. Two different small strain damping profiles were used in the analyses: EPRI (1993) and Park and Hashash (2005). Small Strain Damping Profiles The two small strain damping profiles are listed in Table 2. Although damping values are listed for depths below 150 meters for completeness, the calculation scheme used for this research only includes damping for the upper 150 meters of soil. Park and Hashash (2005) developed their small strain damping values by increasing the EPRI damping profile until observed and predicted ground motions for the Enola earthquake (Mw = 4.5) were in close agreement. Park and Hashash (2005) performed a similar comparison of observed to predicted response spectra for several sites. Of these sites, GLAT was the only site where these analyses were performed both in this study and in Park and Hashash (2005). Attenuation was calculated using a hybrid approach described in Romero (2001), whereby the spectral decay parameter, κtotal, is accounted for in part through a kappa high pass filter (κhp) and in part through soil damping in the upper 150 meters of the soil profile (κ150). κtotal was kept constant for each damping model (κtotal = 0.048 sec for EPRI and κtotal = 0.054 sec for Park and Hashash (2005) (PH)). The values of κtotal were taken from Herrmann and Akinci
(2000) for EPRI damping and Park and Hashash (2005) for PH damping. The Herrmann and Akinci (2000) κtotal is for an embayment thickness of 600 meters, and the Park and Hashash (2005) κtotal is for an embayment thickness of 1000 meters. Although in reality the values of κtotal are expected to vary between sites due to the different thicknesses of soil, they are kept constant between sites within each damping model. Although the sites in this study have 770 and 867 meters of soil, they could be considered roughly bounded by the 600 to 1000 meters of soil in the two damping profiles. Table 2.
Small strain damping profiles from EPRI and Park and Hashash. Depth Range (m) 0–6 6 – 15 15 – 37 37 – 76 76 – 152 152 – 250 250 – 350 350 – 450 450 – 550 550 – 650 650 – 1000
Small Strain Damping (%) EPRI PH 1.4 3.5 1.2 3.5 1.0 3.5 0.8 3.5 0.7 3.5 0.5 3.0 0.5 2.5 0.5 2.0 0.5 1.5 0.5 1.0 0.5 0.4
Source Models Three source models for the Central and Eastern United States (CEUS) were used for comparison in these analyses: Atkinson and Boore (1995), Frankel, et al (1996) and Silva (2003). The ground surface response spectra of these models were averaged logarithmically and compared to the observed ground motions. Comparison of Spectra Calculated spectra for the two damping profiles were compared to recorded spectra for GLAT and LPAR (Figs. 3 through 5). The ratio of predicted to observed response spectra are plotted versus period in Fig. 6. For GLAT 2003 (Fig. 3), the peak amplitude of observed spectral acceleration is approximately 45 percent and 75 percent higher than that of EPRI and PH, respectively. The peaks of EPRI and PH are shifted to a slightly lower period relative to the observed motion. The three motions are generally in better agreement at periods greater than 0.25 seconds. The response spectrum is overpredicted by the two damping models for GLAT 2004 (Fig. 4). The peak amplitude is overpredicted by approximately 55 percent for EPRI and 15 percent by PH. The shape of the predicted response spectra are very similar to the observed response spectra. The LPAR 2004 event (Fig. 5) is fairly well predicted by both EPRI and PH damping models.
GLAT 2003 0.008
Recorded
Spectral Acceleration (g)
EPRI (1993) damping Hashash and Park (2005) damping
0.006
0.004
0.002
0.000 0.01
0.1
1
10
Period (sec)
Figure 3. Response spectra comparison for 2003 earthquake at GLAT.
GLAT 2004 0.0025
Recorded EPRI (1993) damping Park and Hashash (2005) damping
Spectral Acceleration (g)
0.0020
0.0015
0.0010
0.0005
0.0000 0.01
0.1
1 Period (sec)
Figure 4. Response spectra comparison for 2004 earthquake at GLAT.
10
LPAR 2004 0.0010
Recorded EPRI (1993) damping
Spectral Acceleration (g)
0.0008
Park and Hashash (2005) damping
0.0006
0.0004
0.0002
0.0000 0.01
0.1
1
10
Period (sec)
Figure 5. Response spectra comparison for 2004 earthquake at LPAR.
3 GLAT 2003 EPRI GLAT 2003 PH GLAT 2004 EPRI GLAT 2004 PH LPAR 2004 EPRI
2 Sa,pred/Sa,obs
LPAR 2004 PH
1
0 0.01
0.1
1 Period (sec)
Figure 6. Spectral ratios for all events
10
The peak amplitude is overpredicted by about 16 percent for EPRI and is underpredicted by about 5 percent for PH. Spectral ratios (Fig. 6) show that nearly all predictions are within a factor of two (ratios between 0.5 and 2.0) up to periods of 1 second. Above 1 second, each site is underpredicted, typically by a factor of 2 or more. LPAR 2004 event appears to be the best-predicted response spectra. The spectral ratio is within a factor of +/-2 for nearly all periods. The least difference between the EPRI and PH predictions is for GLAT 2003, which is the largest earthquake. Park and Hashash (2005) state that the EPRI values significantly overpredicted the observed ground motions. In our study, the EPRI damping result was closer to the observed spectrum for one event (GLAT 2003). Some differences between this study and Park and Hashash (2005) that may affect site response predictions are that this study included site-specific Vs in the upper 30 to 50 meters measured at the site, a varying crustal profile, and resultant horizontal ground motions. Park and Hashash (2005) used generic embayment soil profiles, constant bedrock Vs, and individual components of horizontal ground motion. Conclusions At this time, it appears that the average of three source models and both damping profiles reasonably predict earthquake spectra measured in the UME for magnitudes of MbLg ≤ 4. Both damping models appear to predict observed ground motions well, though the PH model appears to predict the 2004 events (GLAT and LPAR) better than the EPRI model. However, the EPRI model is a better predictor for the GLAT 2003 event. Although the underlying cause for these relationships is not known at this time, it is interesting to note that the largest underprediction occurred for the largest earthquake magnitude, the 2003 event. Some of the differences in input parameters, such as Vs profiles for shallow soil and bedrock, as well as comparison of component observed motions to resultant ground motions, may account for some differences in results. As the remaining sites are analyzed and compared to predicted response spectra, trends may become apparent relative to geology, soil thickness, location and earthquake magnitude. Acknowledgments This work was funded by the Mid-America Earthquake Center under a grant from the Earthquake Engineering Research Center’s program of the National Science Foundation under Award No. EEC-9701785. The authors thank Dr. Chuck Langston and Christy Chiu for providing earthquake ground motions for these sites. References Atkinson, G. and Boore, D., 1995. New ground motion relations for eastern North America, Bulletin of the Seismological Society of America 85, 17-30. Boore, D.M., 1983. Stochastic simulation of high-frequency ground motions based on seismological models of the radiated spectra, Bulletin of the Seismological Society of
America 73 (6), 1865-1894. Bakun, W.H. and Hopper, M.G., 2004. Magnitudes and Locations of the 1811-1812 New Madrid, Missouri, and the 1886 Charleston, South Carolina, Earthquakes, Bulletin of the Seismological Society of America 94 (1), 64-75. Brune, J.N., 1970. Tectonic stress and the spectra of seismic shear waves from earthquakes, Journal of Geophysical Research 75, 4997-5009. Catchings, R.D., 1999. Regional Vp, Vs, Vp/Vs, and Poisson’s ratios across earthquake source zones from Memphis, Tennessee, to St. Louis, Missouri, Bulletin of the Seismological Society of America 89 (6), 1591-1605. EPRI, 1996. Guidelines for Determining Design Basis Ground Motions, EPRI TR-102293, Palo Alto, CA. Frankel, A., Mueller, C., Barnhard, T., Perkins, D., Leyendecker, E., Dickman, N., Hanson, S., and Hopper, M., 1996. National seismic hazard maps, June 1996, U.S. Geological Survey: Open-File Report 96-532. Herrmann, R.B. and A. Akinci (2000), Mid-America Ground Motion Models, http://www.eas.slu.edu/People/RBHerrmann/GroundMotion/. Kramer, S.L., 1996. Geotechnical Earthquake Engineering, Prentice Hall, Upper Saddle River, New Jersey. Park, D. and Hashash, Y.M.A., 2005. Evaluation of seismic site factors in the Mississippi Embayment. I. Estimation of dynamic properties, Soil Dynamics and Earthquake Engineering 25, 133-144. Romero, S.M., 2001. Ground motion amplification of soils in the upper Mississippi Embayment, Ph.D. Thesis, Georgia Institute of Technology. Silva, W., 1998. RASCALS V 5.4. Silva, W., Gregor, N. and Darragh, R., 2003. Development of regional hard rock attenuation relations for central and eastern North America, mid-continent and gulf coast areas, Pacific Engineering and Analysis, El Cerrito, California. Street, R., Woolery, E., and Chiu, J-M., 2004. Shear-wave velocities of the post-Paleozoic sediments across the upper Mississippi embayment, Seismological Research Letters, 75, 390-405. Yoon, S., 2005. Array-Based Measurements of Surface Wave Dispersion and Attenuation Using Frequency-Wavenumber Analysis, Ph.D. Thesis, Georgia Institute of Technology.