FEMA-273 SITE EFFECTS

Introduction

Research efforts of the MAE Center Seismic Hazard Maps effort have focused on those unique aspects of the central United States which would cause major perturbations to the current 1997 NEHRP Probabilistic Seismic Hazard maps. A survey of shallow surface materials and shear-wave velocities (robert Bauer, Illinois Geological Survey) indicates that much of the region would be described as a FEMA 273 Site Category B or C. However, the great sediment thicknesses (Figure 1) of the lower Mississippi Valley significantly perturb expected ground motions in such a way that sole use of surface material properties to modity hard rock spectra is wrong. The purpose of this presentation is to compare predicted linear motions to FEMA 273 site categores for three sites in the region: Sikeston, MO, Portageville, Mo and Memphis, TN. The nominal deep soil thicknesses are 100 m, 600m and 1000 m, respectively.

Figure 1. Prototype Thickness Model for the Mississippi Embayment.

FEMA 273 Methodology

FEMA-273 contains the methodology for defining site specific spectrum. This is accomplished byusing the 1996 NEHRP probabilistic hazard assessment maps and FEMA site categories. The following steps are performed:

Uses 1.0 sec Spectral Acceleration from maps
Uses 0.2 sec Spectral Acceleration form maps .
Uses A, B, C, D or E Sites terms
Scaling accounts for non-linearity of motion
Shapes spectral according to these parameters

Why Examine FEMA-273 Methodology

Understand methodology
Test whether appropriate for our region
If not appropriate, focus on getting the best numbers for our region
One step on focusing limited resources toward maximum payback .

Examination of Mid-America Deep Soil Sites

Memphis, TN (1000 m)
Portageville, MO (600 m)
Sikeston, MO (100 m )

Compare

NEHRP estimates for all sites, to
MAEC Deep Soil (Model X) estimates but
Model X has only deep soil attenuation, no non-linear surface effects

Procedure

FEMA 273 relies only on the 0.2 and 1.0 sec SA values from the 1997 NEHRP Probabilistic Seismic Hazard Maps. Using both the FEMA site categories and the spectral acceleration factors, spectral acceleration factors are obtained. The combination of site categories and ground motion level permit building in generic non-linear site amplifications. The resulting response spectra are plotted in the figures shown in Table 1, with the colors representing the different FEMA 273 site categories. In addition the 0.3 second spectral acceleration and peak acceleration obtained from the probabilistic analysis are also plotted as yellow shaded squares. The purpose of adding these two extra points is to test the consistency of the FEMA 273 procedure for defining response spectra shape on the basis of two parameters.

The 1997 NEHRP maps are based on the sue of two ground motions models: the Frankel USGS model and the EPRI model. For our seisitivity test of the implications of the deep soils on expected surface motions, we worked with the Frankel USGS model only since it was very easy to strip off the B-C shallow structure and then to overlay the MAE Center Model X deep soil shear-wave velocity and Q model. Thus there may be an offset in expected motions depending upon which of the USGS or EPRI models were larger.

For each geographical site [ Sikeston, 37.0N, 89.6W; Portageville, 36.425N, 89.700W; Memphis, 35.123N, 89.932W] appropriate deep soil ground motions tables (function of distance and magnitude) were used. The USGS Probabilistic Seismic Hazard Mapping program was run, and the expected peak motions for peak ground acceleration, 0.2s, 0.3s and 1.0s periods were extracted for probability levels of 10% in 50 years and 2% in 50 years. These deep soil estimates are plotted in the figures of Table 1 as black dots.

Besides the difference in bedrock input, another major difference in the presentation is that the deep-soil estimates are purely linear. The FEMA 273 site amplifications depend on the level of ground motion. It is our hope that it is valid to bring large motions up through the deep soil column in a linear manner and that the non-linear effects are still confined to the surface, within the realm of geotechnical engineering procedures. The non-linear effect built into the tables is obvious in the 2% 50 year figures for which a maximum level of high frequency motion is seen irrespective of the site classification. Thus we would expect that the high 0.2 s and 0.3 s levels at Sikeston for the 2%50 year plot would be reduced because of surface non-linear effects.

The last caveat is that in applying the FEMA 273 methodology the 10% 50 year and 2% 50 year procedures are to be used together to ensure a minimum bound on the motions.

Results

Table 1. Comparison of FEMA 273 Spectra to Deep Soil Spectra
Sikeston 10% 50 years	Portageville 10% 50 years	Memphis 10% 50 years
Sikeston 2% 50 years	Portageville 2% 50 years	Memphis 2% 50 years

Discussion

The striking feature of the 10% 50 year plots is that our initial deep soil estimates do not fit neat any of the FEMA site categories exactly. This is because of two effects. If we use a shear-wave velocity model that has a power-law function with depth and if we assume that the surface materials are the same everywhere (which is not true toward the edges of the embayment because the assumed flat layer-cake model of deposition is not really true), then

high frequency geometric site amplification is independent of deep soil thickness
amplification > 1 moves toward longer periods as thickness increased
anelastic attenuation effects (damping) on the spectrum become more pronounced as thickness is increased

These competing effects are evidenced most pronouncedly at Memphis, where the deep soil 1 Hz level is elevated with respect to the NEHRP B-C 1 Hz values, and the deep soil PGA value is reduced from the NEHRP B-C boundary value for the 10% 50 year plot. Thus irrespective of the actual site condition near Memphis, the deep soil column perturbs the surface motions sufficiently to make the surface linear spectrum seem to mimic something between a NEHRP B or C site, and not anything worse.

For the 2%50 year comparison for Memphis, it seems as if our motions might be more like a FEMA Site D. The FEMA Site B category has a site amplification 1.0 independent of ground motion input. Our study indicates that Memphis might perform better that a Sige B at shorter periods. The site amplification at 1.0 s is very obvious.

Because of its significantly shallower thickness (100 m assumed), the Sikeston simulation exhibits significant amplification at short periods and some amplification at 1.0s for the 10% 50 year simulation. For the 2% in 50 year curves, the simulated 100 meter soil deposit motions are huge. The corresponding FEMA site spectra are already subjected to significant reduction due to presumably non-linear material response. This shows that the values from our linear simulation must be placed into an appropriate non-linear soil profile near the surface for more realistic estimates of motion.

10% 50 year	Site	Observation
	Memphis	FEMA 273 too conservative
	Portageville	FEMA 273 conservative
	Sikeston	Fema 273 wrong at 0.2 sec?
2% 50 year	Memphis	FEMA E OK, conservative
	Portageville	MAEC needs non-linear effects
	Sikeston	MAEC needs non-linear effects

Future Strategy

Veneer some aspect of FEMA non-linear on top of our Model X surface estimates
Examine real world evidence for such behavior at deep soil sites.

Last Changed fberbuary 18, 2000