Mid-America Ground Motion Models

R. B. Herrmann and A. Akinci
Department of Earth and Atmospheric Sciences
Saint Louis University
3507 Laclede Avenue
St. Louis, MO 63103

INTRODUCTION

The purpose of this page is to provide software for generating synthetic time series of mid-west earthquakes for use by Mid-America Earthquake Center investigators. Because of the lack of strong motion data for M > 5 earthquakes in the region, synthetic time series are generated using band-limited white-noise stochastic simulations and random vibration theory. Tools are provided to investigators to generate the data files require to run Boore's SMSIM programs.

Boore's SMSIM package provides two useful programs, td_drvr to generate actual time series and rv_drvr which uses random vibration theory to estimate peak motions. Since the SMSIM package is a general purpose research tool, we have created a program dorvt180 to create a standard set of parameter files for use by investigators. We do not want center investigators to create arbitrary motions, since there must be an internal consistency with hazard mapping input and results

We thus provide a restricted set of permitted ground motion models:
Model Name Source Wave Propagation Site Effect
1 Atkinson-Boore 1995 (AB95) AB95 AB95 ENA Hard rock
2 USGS 1996 USGS 96 150 bar USGS 96
(same as AB95)
Generic B-C Boundary
3 USGS 1996
(modified)
USGS 96 150 bar USGS 96
(same as AB95)
Mid-Continent Deep Soil
(new)
4 Mid-America Deep Soil
AB95 source
(modified)
AB95 Mid-America
(new)
Mid-Continent Deep Soil
(new)
5 Mid-America Deep Soil
USGS96 source
(modified)
USGS 96 150 bar Mid-America
(new)
Mid-Continent Deep Soil
(new)

DISCUSSION OF MODELS

Rationale

Since we are tasked with updating the 1996 NEHRP probabilistic seismic hazard maps for Mid-America, all deviations from the USGS procedure must be justified and documented. There must a transition between current maps and the modified area. Thus the USGS 1996 ground motion model is included as Model 2.

The USGS 1996 model differs significantly from the Atkinson and Boore (1995) model in the source spectrum. The USGS models uses a 150 bar Brune spectral scaling, while the AB95 model uses an empirical scaling which differs from the USGS model for large moment magnitude earthquakes. Thus the AB95 model is offered as an alternative.

The deep soils in mid-America are not included in either the AB95 or USGS96 grounnd motion scaling. The prototype Mid-Continent Deep Soil model gives site amplification and kappa values that differ according to the thickness of the soils. Model 3 differs from the USGS96 model only in the amplification and kappa values. In effect this is done by stripping off the B-C boundary surface, and grafting the deep soil model onto it.

Recent, unpublished studies by Herrmann and others provides more information on crustal wave propagation in Mid-America centered on New Madrid ( Ground Motion Scaling from Earthquakes in the New Madrid Seismic Zone). Crustal wave propagation differs from that used in AB95 and USGS96 in both the Q(f) quality factor and g(r) geometrical spreading terms. In addition a constraint on the composite effects of site amplification and kappa is found for the region between New Madrid, MO, and Dyersburg, TN. However, no new source scaling information is available, so that either the AB95 source (Model 4) or the USGS96 source (Model 5) scaling is used.

Mid-America Deep Soil Model

The prototype Mid-America Deep Soil Model consists of an optional soil, over 500 meters of Pennsylvanian/Mississippian limestones, over 500 meters of Cambrian/Ordovician limestones etc, over a preCambrian basement.

The deep soil model is constrained by shallow shear-wave velocity measurements made by Paul Mayne and others in the Bootheel, Memphis and western Kentucky. The proposed shear wave velocity model has velocities, Q and density which increase with depth. A single power law relation is used. The power law for velocity is constrained by the surface shear-wave velocity measurements and the requirement that the vertical travel time from the rock/soil contact at New Madrid by 0.6 seconds. The Qs is not as well constrained, other than having some reasonable value near the surface at New Madrid, and a kappa of 0.048 through 600 m of sediments. The power law relation requires more study since the exponent permits extrapolation from New Madrid observations to the 1000 m thick Memphis site (the vertical shear-wave travel time may be appropriate).

For the Paleozoic layers, we use determinations of surface rock shear-wave velocities by Dr. Ron L. Street of the University of Kentucky mailto:GEO151@UKCC.UKY.EDU

     Date:         Sat, 13 Mar 99 08:03:30 EST
     From:  
     To: rbh@EAS.SLU.EDU
     
     Bob,
     
     My closest SH-wave measurements to St. Louis are at the strong-motion
     stations operated by Lamont:
     
           @ OLAP  bedrock is 1 m below the surface and Vs = 2540 m/s
           @ VIEN     "     " 4 m  "     "    "      "  Vs = 2095 m/s
     
     I found the original data, and these were good sites.
     
                                                          Ron

These values are simlar to results from surface wave dispersion studies in the area by Woods et al (1989) and Hutchensen (1994)

Prototype Model

Layer Material Thickness P-vel (m/s) S-vel (m/s) Density (gm/cc) Qp Qs
1 Soil h = 0-2000 m 1800 250 h0.18 0.8 log10 Vs - 0.1 6 h0.24 6 h0.24
2 Upper Paleozoic 500 m 4500 2500 2.5 500 500
3 Lower Paleozoic 500 m 5000 3000 2.6 500 500
4 PreCambrian - 6000 3500 2.7 500 500

Limitations

The present model of the deep soil should not change drastically as they evolve. The variation of parameters with depth into deep soils will be refined as new research is incorporated.

Walt Silva (personal communication) cautioned that modifications to the point source model used in the current simulations may be required close a large earthquake. In addition some of the sharp changes in the geometrical spreading terms seen in studies of small earthquakes will be smoothed. The net effect is that current simulations of large New Madrid earthquakes may overestimate ground motions at short distances.

The current choice of models does not use the latest EPRI model for central and eastern U. S. Since that model is the result of significant research, it should also be included as an option.

Future Directions

Investigate ground motions near a finite fault

Acquire more data on shallow shear-wave velocities and attempt to get Qs information.

REFERENCES

Atkinson, G. M., and D. B. Boore (1995). Ground-motion relations for eastern North America, Bull. Seism. Soc. Am. 85 17-30.

Boore, D. B. (1996). SMSIM - Fortran Program for Simulating Ground motions from Earthquakes: Version 1.0, USGS Open File Report 96-80-A

Frankel, A., C. Mueller, T. Barnhard, D. Perkins, E. V. Leyendecker, N. Dickman, S. Hanson and M. Hopper (1996). National Seismic-Hazard Maps: Documentation June 1996, USGS Open-File Report 96-532

Hutchensen, K. D. (1994). Shallow structure of the Illinois Basin from fundamental and higher-mode regional surface wave dispersion, Ph. D. Dissertation , Saint Louis University, 243pp.

Schneider, J. A., and P. W. Mayne (1998). Results of seismic piezocone and flat plate dilatometer tests performed in Blytheville, AR, Steele, Mo, and Shelby county, TN. Interim Report MAEC Project No. GT-3, GTRC Project No. E20-677, Geosystems Engineering Group, Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0355.

Woods, M. T., D. R. Russell, and R. B. Herrmann (1989). Dispersion of short period Rayleight waves with the Ozard Uplift and Illinois Basin, Seism. Res. Let. 60, No. 3., 111-118.

APPENDIX

Bob Bauer (Email: bauer@isgs.uiuc.edu) (personal communication), Illinois Geological Survey, gave me the following tabulations. For more details, get in touch with him. These are average values compiled from various sources which used different shearwave determination methods. As new sites/values are aquired average values may change. Question marks are where a lack of data exists and values are best estimates.

<<<<<<<<<<<<
ID Material Vs (m/sec)
ACahokia Alluvium230
DParkland Sand - dunes330 calf
ECarmi Member of Equality170
PPeoria Loess and Roxanna Silt200?
GGlasford Till365
OSand and Gravel of Glasford200?
MMounds Gravel360
TTertiary clay and sand324
KCretaceous sand and gravel280
CCherty residuum 200?
SSandy residuum 200?
RSandy cherty residuum 200?
ZTertiary, Cretaceous or Miss.2000
1Penn. shale1500
2Penn. sandstone2000
3Miss. shale2000
4Miss. limestone2900
5Devonian limestone2900
6Silurian limestone2700
7Devonian/Silurian dolomite/limestone2900
8Ordovician shale 2000
9Ordovician sandstone,dolonmite2000
10Ordovician dolomite2900
11Ordovician sandstone2000
12Ordovician limestone/dolomite2900

SOFTWARE

Herrmann's Distribution:

rbhsm18.exe self-extracting zip archive

Create a new directory in and MS-DOS window. Go to that directory. Copy rbhsm18.exe to that directory, and execute the program. You will find dorvt180.f, dorvt180.exe, rv_drvr.exe and td_drvr.exe.

The Herrmann executables were compiled using g77 MINGW32 Distribution of G77, GCC, G++ For related pages see GNU Win32 related projects, and EGCS Development Toolchain for x86-win32 targets These compilers are free and work under W95/98/NT. It is only necessary to find a suitable version of make

Purpose and use of the program dorvt180. The purpose of dorvt18X is to create the data files for working with Boore's SMSIM18 package. The 'X', now currently '0' permits us to change the ground motion models, but still maintain compatibility with the SMSIM18 formats.

On Windows 95/98/NT go to an MS-DOS window. Then CD to the directory for working with these programs, perhaps

	cd c:\usr\boore.99
You may wish to place your executables into c:\usr\boore.99\bin
Go to a work directory for this simulation, e.g., something like
	cd work
Now run the program dorvt180 to compute the data files (assuming that its relative location in a neighboring bin18 directory):
	..\bin18\dorvt180
You will see the following:

                           DORVT
 This program creates control files for use with Dr.
 David Boore s programs (SMSIM1.8) for simulating earthquake
 ground motions. The choices permit only certain Eastern
 North America ground motion models
  
 (SMSIM - Fortran Program for Simulating Ground motions from
 Earthquakes: Version 1.0, USGS Open File Report 96-80-A)
  
 Enter Ground Motion model:
     1    Atkinson-Boore 1995 ENA
     2    Frankel  NEHRP 1996 ENA
     3    Frankel (modified)  Deep Soil
     4    Mid-America   Deep Soil - AB src
     5    Mid-America   Deep Soil - F  src
1
 Enter deep soil thickness meters
     (0-2000) meters
     Ignored for models 1 and 2
0
 Enter moment magnitude (3.0-8.5):
6
 Enter Hypocentral Distance km(1.0-1000):
10
 Enter oscillator damping (0.0-1.0):
0.05
 Enter Nper, Tlow, Thigh :
     Nper  - number of oscillator periods
       0   - do not use oscillator
     Tlow  - lowest oscillator period > 0
     Thigh - highest period > 0
20 .1 10
 Enter number of runs (1-200):
      number of time domain simulations
      to be run for average spectra.
      For a time history output, enter 1
100
 Enter the random number seed (1-32000):
123

If you execute the command DIR, then you will see the following files in the work directory

	rv.dat
	td.dat
	model.dat

To perform a time domain simulation, just run

	..\bin18\td_drvr < td.dat > td.out

The Boore time domain simulation is run, creating the following files

	tdsum.out
	tdcol.out
	tddva.out
	tavd.out

The file tavd.out is the time series outout, and will be very large. For this example it is 16385 lines long. The first few lines are

         T           A           V           D
    0.000000  5.9605E-06  0.0000E+00  0.0000E+00
    0.005000 -4.4703E-06  3.7253E-09  3.1044E-11
    0.010000  3.6318E-06  1.6289E-09  2.7550E-11
    0.015000 -1.0626E-05 -1.5857E-08  2.1684E-11
    0.020000 -6.1726E-06 -5.7854E-08 -1.7187E-10
    0.025000 -5.9848E-06 -8.8247E-08 -5.3752E-10
    0.030000  4.8251E-06 -9.1147E-08 -1.0085E-09
    0.035000 -1.6128E-05 -1.1940E-07 -1.4912E-09
    0.040000  5.0532E-06 -1.4709E-07 -2.2016E-09
where T is the time in seconds, A, the acceleration in (cm/s/s), V, the velocity in (cm/s) and D the displacement in (cm) for this simulation. These computations take some time since 100 simulations were requested in the dorvt180 responses. In any event, only the time series from the first simulation is output. If the long time series is inconvenient, the user can change the sampling interval from 0.005 to 0.01 seconds to reduce the number of points. The long time series result from the desire to generate synthetics that also have realistic displacements.

The file tddva.out has (period,amplitude) pairs for plotting the linear asymptotic segments for peak velocity, displacment and acceleration on tripartite plots.

The file tdcol.out summaries the simulation results, by giving the filter period, natural frequency, the SD, PSV, and PSA of the simulations and the ratio (STD/PSV), which is an indication of the variability of any one simulation. The tdcol.out file so created is

   per     freq sd:tdcol___ psv:tdcol___ psa:tdcol___ std/psv:tdcol___
  0.100   10.000   2.644E-01    1.661E+01    1.044E+03        1.904E-01
  0.127    7.848   3.680E-01    1.814E+01    8.947E+02        2.088E-01
  0.162    6.158   4.603E-01    1.781E+01    6.892E+02        2.469E-01
  0.207    4.833   6.084E-01    1.848E+01    5.610E+02        2.254E-01
  0.264    3.793   7.244E-01    1.726E+01    4.114E+02        2.487E-01
  0.336    2.976   9.119E-01    1.705E+01    3.189E+02        2.603E-01
  0.428    2.336   1.059E+00    1.554E+01    2.281E+02        2.908E-01
  0.546    1.833   1.248E+00    1.437E+01    1.656E+02        3.218E-01
  0.695    1.438   1.301E+00    1.176E+01    1.063E+02        3.162E-01
  0.886    1.129   1.396E+00    9.899E+00    7.021E+01        3.361E-01
  1.129    0.886   1.469E+00    8.175E+00    4.550E+01        3.227E-01
  1.438    0.695   1.561E+00    6.818E+00    2.978E+01        3.488E-01
  1.833    0.546   1.609E+00    5.516E+00    1.891E+01        3.590E-01
  2.336    0.428   1.696E+00    4.562E+00    1.227E+01        3.542E-01
  2.976    0.336   1.776E+00    3.749E+00    7.914E+00        3.626E-01
  3.793    0.264   1.926E+00    3.191E+00    5.287E+00        3.896E-01
  4.833    0.207   2.012E+00    2.616E+00    3.401E+00        4.155E-01
  6.158    0.162   2.036E+00    2.077E+00    2.119E+00        4.122E-01
  7.848    0.127   1.979E+00    1.584E+00    1.268E+00        4.090E-01
 10.000    0.100   1.883E+00    1.183E+00    7.432E-01        3.859E-01

The tdsum.out lists all input parameters from the model.dat file, and also the mean peak motions from the simulations. An extract of the output is

 r, amag, kappa =   1.000E+01  6.000E+00  0.000E+00
   Time Start: 12:30:20.00
   Column file: tdcol.out
   const=   4.028E-24
   amag, stress, fa, fb, durex=  6.000 0.00E+00 1.629E-01 2.004E+00 3.07E+00
   am0, am0b_m0fa=   1.122E+25  4.989E-02
   npts, dt, total duration =  16384 0.00500  81.9
    pgd(cm)    std/pgd  pgv(cm/s)    std/pgv pga(cm/s2)    std/pga
   1.64E+00   3.70E-01   1.07E+01   1.74E-01   7.14E+02   1.20E-01
 Arias intensity(cm/s)  std/Arias
              1.70E+02   5.65E-02
   Fractional oscillator damping =  0.050
    per(s)      freq psv(cm/s)    std/psv psa(cm/s2)    psd(cm)
     0.100    10.000  1.66E+01   1.90E-01   1.04E+03  2.644E-01
     0.127     7.848  1.81E+01   2.09E-01   8.95E+02  3.680E-01
     0.162     6.158  1.78E+01   2.47E-01   6.89E+02  4.603E-01

David Boore's Distribution:


USGS ftp site
READ.ME for Version 18
SMSIM Version 1.84 self-extracting zip archive


Local SLU mirror for this software
READ.ME for Version 1.81
SMSIM Version 1.81 self-extracting zip archive

Boore provides all sources, except for Numerical Recipes routines. His executables are compiled with Lahey Fortran and use the PharLap DOS extender. These work under W95/98/NT.

Last Changed April 29, 1999