A Plan for Urban Seismic Hazard Mapping in the St. Louis Area
A seismic hazard map depicts the expected level of ground shaking caused by earthquakes and how that level varies spatially. Urban seismic hazard maps differ from the USGS national seismic hazard maps in that they are higher resolution and they account for the effects of the shallow rocks, sediments, and topography on earthquake ground shaking (i.e. site effects). These maps typically show the motion at a single frequency or parameter and for a particular probability of being exceeded during a particular time span. Such maps are probabilistic, and explicitly consider the likelihood and potential contribution of all possible earthquakes, a range of possible seismic wave propagation characteristics and site effects, and account for observational uncertainties. Deterministic or scenario maps, showing the ground motion for a single set of input parameters (e.g. a specific earthquake at a particular location, etc.), can be derived from a subset of the information used to make probabilistic maps. In addition to ground motion, another type of seismic hazard maps may show the spatially variable likelihood of ground failure (such as landslide or soil liquefaction), by combining the potential of materials to lose strength or fail when shaken and the level of shaking expected.
Generating useful high-resolution seismic hazard maps is both a technical and non-technical process. The maps and derivative products should be useful in meeting the needs of a broad constituency and thus will be widely effectively advertised, disseminated and updated. These primarily non-technical aspects of mapping must be accomplished in close coordination with the technical work. Herein, we present a plan for all aspects of generating useful high-resolution seismic hazard maps. We divide the plan into three parts; the first concerns organizational issues, the second, product generation and dissemination, and the third, scientific work. Note that this is an evolving plan that will be updated as the project progresses.
I. Defining users and their needs.
Useful products satisfy user needs. Thus, a key first step is to determine who the likely users are and what each needs (products, area covered, etc.).
The area to be mapped includes all 29 quads of the St. Louis metropolitan area. Several reasons justify choosing such a large area. If the maps were going to be used, an entire political entity needs to be mapped. Much of the most time-consuming work, the geologic mapping, has already been done; e.g., in Illinois 3-D geologic mapping is complete and some has been done in Missouri. It is also likely that what's learned in mapping this area will be applied to other regions of Missouri and Illinois, so that we should attempt to sample as many different environments as possible.
II. Identifying resources, organizing, scheduling.
The project is coordinated by a coordinating committee:
Project resources may come from a variety of sources. Among these are:
III. Forming an Advisory Board.
Successful mapping requires communications between users and those involved in map production. An Advisory Board will help provide this communication, and review project plans, products, etc. The coordinating committee will select final list of invitees to the Advisory Board, drawing on this list of possible members:
Product Generation, Dissemination
In addition to ground motion maps, liquefaction maps and possibly landslide maps also are needed. All products will be consistent with the USGS national seismic hazard maps, and to the extent possible, with other USGS urban seismic hazard maps. Final decision on the ground failure maps should be based on some background calculations that show whether the potential for ground failure really is significant enough to warrant the effort required to do a full ground failure hazard assessment. All supportive data, maps, and derivative products will reside in a GIS and databases and be accessible via standard Internet Web browser interfaces.
The Missouri Dept. of Natural Resources and Illinois Geological Survey both have databases for geological and geotechnical information under development. Neither are publicly accessible yet. The USGS National Mapping Center in Rolla, MO may be able to provide support in developing and implementing the facilities required for databasing, digital mapping, etc., perhaps as part of the National Map effort. Similarly, the East West Gateway Council might provide support in archiving and serving digital information.
II. Education and Outreach
The mapping process provides an ongoing opportunity to educate the public about earthquake hazards. Education and outreach activities should commence immediately, with possible activities to include: 1) a series on earthquake hazard in the St. Louis Post Dispatch, 2) a panel discussion on earthquake hazards for a local television program aimed at St. Louis businesses during the 2004 Earthquake Awareness Week, 3)a Newspapers in Education program, 4) talks to local professional organizations, and 5) working with school organizations.
The basic input parameters to high-resolution seismic hazard map calculations include information about 1) earthquake sources, 2) ground motion attenuation, and 3) near surface materials. For the first two of these it is critical to use the same input as used in the USGS national seismic hazard maps. The difference between urban hazard maps and the national hazard maps is in the third input, which this project will provide. The near surface geological materials, and the 3-dimensional variation in their thicknesses and physical properties, either amplifies or de-amplifies the level of earthquake ground shaking and lengthens or shortens its duration. They also affect the potential for ground failures to occur. A goal of seismic hazard mapping is to forecast these effects. The Memphis, Seattle, and other urban hazard mapping projects have developed the methodology for mapping, or forecasting ground motions and liquefaction susceptibility. However, as noted below, we anticipate new discovery and development.
In this context 'near surface' refers to different depths, depending on the type of map (ground motion, liquefaction or landslide susceptibility). The key controlling characteristics of the rocks and sediments also depend on the map type, but some are common to all. We discuss these common inputs below, and then those specific to each type of map in separate sections.
II. Surficial and 3-D geologic maps
The general state-of-the-art is to assume that relevant characteristics of near surface materials (or 'soil' in engineering terms) vary with their geologic classification or 'lithology'. If such an assumption is true, it is useful because surficial geologic maps and 3-dimensional pictures of the lithology can be made at higher resolution than one can map most other material properties. Thus, 3-D geologic maps can be used as proxies, or interpolation tools, for mapping the physical properties. The appropriateness and accuracy of using such proxies determines the accuracy of the maps, and undoubtedly varies regionally and locally. Thus, in addition to the geologic mapping itself, a major effort must be dedicated to establishing the degree to which the lithology correlates with the relevant material properties.
Surficial geologic maps showing the distribution of geologic units, particularly the unconsolidated materials, will be generated using standard mapping techniques. Geologic maps of the Illinois portion of the 29 quad area are either finished or in progress. In Missouri 1:24000 bedrock maps exist for the St. Louis area although it isn't clear that the data that went into the maps were actually collected at this scale. Support for field mapping needs to be secured before surficial mapping can be completed in Missouri. Because the surficial geologic mapping work will involve several geologists, efforts must be made to insure uniformity in mapping style, format, etc.; Rus Wheeler will help in establishing the "mapping protocol" (e.g. unit naming, etc.).
Data constraints on the stratigraphy come primarily from surface geologic mapping and from logs of various types, from which the depths to lithologic boundaries are estimated. In Illinois there’s a digital inventory of logs, although the logs themselves have not been made digital. While they have many logs, some geotechnical logs still need to be tracked down. In Missouri, water well logs interpreted by Missouri DNR have been made digital and accessible on the web. The Missouri Dept. of Natural Resources has an FY04 NEHRP project to evaluate and database existing subsurface information for Missouri, the results of which will be used to determine what additional work is required.
The basic analysis for deriving stratigraphic boundaries requires fitting discontinuous surfaces to point measurements of the boundary depths. A number of possible fitting procedures may be applied, as well as structural interpretations that are compatible with the surficial geology and basic geologic principles.
III. Ground motion maps
To estimate probabilistic ground motions that include the effects of shallow rocks and sediments we will follow the analysis procedure described in Cramer (2003). The key input information includes the shear-wave velocity (Vs), compressional-wave velocity (Vp), initial damping (Qs, Qp), and density of the sub-surface materials, how these vary in 3-dimensions, and measures of the accuracy of these characteristics. For more complex (non-linear) soil response calculations, additional soil properties will need to be known (geotechnical properties such as soil class (sand versus clay), plasticity, porosity, water saturation, and dynamic soil properties). 'Shallow' in this context refers to depths extending to the base of the unconsolidated sediments. Minimally certain characteristics of the rock just beneath the sediments also should be known (Vs, Vp, density, and damping).
Because measurements of these characteristics are relatively costly to make, we begin by testing the assumption that they correlate with lithology. The CUSEC State Geologic have an FY04 NEHRP grant to compile all existing shear-velocity profiles and create a database that will reside at the Illinois Geological Survey. Using the surficial and 3-D lithologic maps and subsurface databases described above, we will determine where additional measurements should be made to be able to assess their correlation with lithology. The possibility of leveraging the NEHRP support to obtain MODoT support to collect additional Vs measurements will be considered. (MODoT has already helped to establish a Vs database, so this would be to help populate it.)
State-of-the-art ground motion estimation considers the non-linear response of sediments to ground shaking. Non-linear effects may amplify or reduce ground motions relative to those motions estimated assuming that the output motion is simply proportional to the motion input to the base of the sediments. For sufficiently large input ground motions, the non-linear sediment response is thought to limit or cap the motion at the surface. While this implies a lower ground motion hazard, severe non-linearity ultimately may result in ground failure. Without question, predictive models of the non-linear response are highly uncertain, both in terms of the underlying theory and the input parameters that must be measured in the field. The crudest approach to include non-linear affects employ 'site-amplification factors' (e.g. those recommended in the 2000 NEHRP seismic provisions), which are standard multiplicative factors that depend on the gross sediment characteristics and input ground motion amplitude and frequency. More complete approaches account for the specific properties of the sediments and their underlying physical behaviors. In addition to requiring shear-wave and compressional-wave velocities and densities, such approaches also require specification of the 'modulus reduction' and 'damping' properties of the sediments, which describe how the sediments lose strength and dissipate energy with shaking, respectively. Although data constraining these properties do not exist for the region, funding opportunities to obtain them do. The Missouri University of Science and Technology also is building laboratory facilities to do needed in situ testing.
IV. Liquefaction susceptibility maps
Susceptibility here refers to the inverse of the 'capacity' of the sediments to maintain their strength when shaken (i.e. to not liquefy). The liquefaction potential depends both on the capacity and on the 'demand', or the shaking levels the sediments are likely to experience. Thus, liquefaction potential maps combine ground motion and liquefaction susceptibility maps. Our strategy to mapping the susceptibility makes direct use of both geologic and geotechnical information, rather than only one as is often done in more traditional approaches. As for the ground motion parameters, this work focuses on establishing the relationship between the lithology and the properties that control liquefaction capacity.
Susceptibility is best estimated from cone penetrometer test (CPT) measurements, and also may be derived from standard penetration test (SPT) measurements that generally are more abundant but are of poorer or unknown quality. For a specified ground motion amplitude and duration, standard analyses are applied with the CPT or SPT measurements to derive a 'factor of safety' profile and then a 'liquefaction potential index' (LPI). LPI is a measure of potential to liquefy (from none to major liquefaction, 0 to 15). All the LPIs estimated for each surficial geologic unit are combined to determine probability density functions. These thus provide measures of the correlation of LPI and geologic unit, or equivalently the probability of liquefaction for each unit. These correlations or probability density functions then allow the geologic maps to be transformed into probabilistic liquefaction susceptibility maps.
In addition to considering compilation of existing and collection of new CPT and SPT measurements, information about grain size and other geotechnical properties, and the ground water table are needed. In contrast to the ground motion mapping, 'shallow' in this context refers to the top few tens of meters and geologic units must be distinguished with greater resolution (e.g., different types of fill may have very different susceptibilities while ground motion estimates are completely insensitive to these). Some work on assessing liquefaction susceptibility may already have been done by William Lettis and Assoc. under a NEHRP grant.
Cramer, C.H. (2003). Site-specific seismic hazard analysis that is completely probabilistic, Bull. Seismo. Soc. Am . Vol. 93, No. 4 (August).