Preliminary Deep Sediment Thickness Model
This is an initial run of the USGS Probabilistic Hazard Analysis code using the prototype MAE Center groud motion models for the central United States. (Ground Motion Models).
The USGS Hazard mapping page is USGS National Seismic Hazard Mapping Project
The idea is to compare hazard estimates from these models to those arising from the USGS computations for the 1996 NEHRP maps. The primary difference is that the linear effects of the deep soil deposits are accounted for. The following figure shows a prototype sediment thickness model. Since we are interested primarily in the deep sediments, we ignore anything less than 50 meters thick. This excludes most of the river valleys. This picture may be improved using the MAE Center GIS compilations.
|
Preliminary Deep Sediment Thickness Model |
The generic velocity models are given here. The 1000 meter model is a thick green line.
Prototype Deep Soil Model |
The site amplification as a function of soil column thickness currently used is given here:
Frequency Dependent Site Amplitucation |
This figure presents both the geometrical and anelastic attenuation effects. This is based on David Boore's quarter-wavelength technique. The colors represent different thickness of the soil column. The green 1000 meter column curve (appropriate for Memphis) is emphasized with the greater width. For reference, the black dashed line is the frequency dependent site amplification used in the USGS 1996 NEHRP maps. The lower blue curve is the proposed mid-America amplification at the top of competent rock (which in the St. Louis area is abouut 20-30 meters thick). It is not flat because of the slight decrease of shear-wave velocity near the surface.
In performing the computations, four ground motion models are compared.
The following pages compare the plots of the four ground motion models. The Peak Acceleration plots will focus on the high frequency component of the expected ground motion, while the 1.0 Sec Pseudo Acceleration focuses on a lower frequency.
| Map | Probability |
| Peak Horizontal Acceleration | 10% 50 year |
|
|
05% 50 year |
|
|
02% 50 year |
| 1.0 sec Pseudo Acceleration | 10% 50 year |
|
|
05% 50 year |
|
|
02% 50 year |
10% 50 year Peak Acceleration
The striking feature is the boomerang shaped high expected ground motions in the northern part of the embayment, at some distance from current microseismicity. This is quite understandable in terms of the site amplification and total attenuation in the deep soil column. At high frequencies, the wave theory amplification is essentially independent of soil thickness, but the anelastic attenuation effects, which remove high frequencies, are significant at deeper sites. Because the earthquake sources used in the simulation are roughly centered on the axis of the embayment, peak acceleartions are severely reduced at most sites when compared to the 1996 NEHRP map.
The effect of introducing a central U. S. distance attenuation relation can be seen by comparing the plots for Model X to Model XX. Peak ground motions at St. Louis are barely affected, while they are reduced in the northern part of the embayment. I believe this is due to the different predicted peak accelerations for small magnitude earthquakes observed at short distance. The prototype CUS model has longer wave propagation durations.
Finally, Models XX and XXX differ only in the source scaling relation. The Atkinson - Boore 1995 model differs in the handling of large earthquakes, and evidently in some aspects of the smaller earthquakes dominating the 10% 50 year hazard.
05% 50 year Peak Acceleration
The boomerang effect is less pronounced since larger earthquakes now contribute more to the hazard.
02% 50 year Peak Acceleration
The characteristic of these figures is that the finite faults assumed for large New Madrid earthquakes has considerably more influence at this hazard level. The same patterns persist, e.g., deep soil reduced peak motions at Memphis, Atkinson-Boore 95 has lower motions than the USGS 1996 source model.
A stimulating observation is given here. If paleoliquefaction reflects peak acceleration, and if the 2% in 50 year hazard level represents a more geographic average, then the expected motions may show more liquefaction sites in the northern embayment, than in NE Arkansas?
10% 50 year 1.0 Sec Pseudo Acceleration
The 1.0 Sec PSA will be strongly affected by the very deep soil sites. We might expect the 1996 NEHRP motions to be modified in a manner proportional to the thickness. We should see a strong influence of the embayment geometry. This is centainly seen comparing Model X to the USGS results.
Subtle differences are seen between models X and XX, but there seem to be big differences compared to the Atkinson-Boore model. This is related to motions for magnitudes greater than 6.
05% 50 year 1.0 Sec Pseudo Acceleration
The same trends continue.
02% 50 year 1.0 Sec Pseudo Acceleration
You can really see the embayment now!