1

Next-Level ShakeZoning for earthquake hazard definition in the Intermountain West

JOHN N. LOUIE, WILLIAM SAVRAN, BRADY FLINCHUM, GABRIEL PLANK, GRAHAM KENT, KENNETH D. SMITH, Nevada Seismological Laboratory, Nevada System of Higher Education, MS 0174, University of Nevada, Reno, NV 89557 (louie@seismo.unr.edu) SATISH K. PULLAMMANAPPALLIL, AASHA PANCHA, Optim Seismic Data Solutions, 200 S. Virginia St. Suite 560, Reno, NV 89501 (satish@optimsoftware.com) WERNER K. HELLMER, Clark County Department of Development Services, 4701 W. Russell Rd., Las Vegas, NV 89118 (wkh@co.clark.nv.us)

ABSTRACT

Our multi-institutional collaboration is developing “Next-Level ShakeZoning” procedures

tailored for defining earthquake hazards in the Intermountain West. The current Federally sponsored tools- the USGS hazard maps and ShakeMap, and FEMA HAZUS- were developed as statistical summaries to match earthquake data from California, Japan, and Taiwan. The 2008 Wells and Mogul events in Nevada showed in particular that the generalized statistical approach taken by ShakeMap cannot match actual data on shaking from earthquakes in the Intermountain West, even to first order. Next-Level ShakeZoning relies on physics and geology to define earthquake shaking hazards, rather than statistics. It follows theoretical and computational developments made over the past 20 years, to capitalize on detailed and specific local data sets and more accurately model the propagation and amplification of earthquake waves through the multiple geologic basins of the Intermountain West. Excellent new data sets are now available for Las Vegas Valley. Clark County, Nevada has completed the nation's very first effort to map earthquake hazard class systematically through an entire urban area. Using the new Parcel Map in computing shaking in the Valley for scenario earthquakes is crucial for obtaining realistic predictions of ground motions. In an educational element of the project, a dozen undergraduate students are computing 50 separate earthquake scenarios affecting Las Vegas Valley, using the Next-Level ShakeZoning process. Despite affecting only the upper 30 m, the Vs30 geotechnical shear-velocity from the Parcel Map shows clear effects on even the longer-wavelength 0.1-Hz to 0.5-Hz shaking predictions. The fully 3-d Next-Level ShakeZoning scenarios show many areas of shaking amplification and de-amplification that USGS ShakeMap scenarios cannot predict.

INTRODUCTION

Our multi-institutional collaboration is developing “Next-Level ShakeZoning” procedures tailored for defining earthquake hazards in the Intermountain West. The current Federally sponsored tools- the USGS hazard maps and ShakeMap, and FEMA HAZUS- were developed as statistical summaries to match earthquake data from California, Japan, and Taiwan. The 2008 Wells and Mogul events in Nevada showed in particular that the generalized statistical approach

2

taken by ShakeMap (Worden and others, 2010) cannot match actual data on shaking from earthquakes in the Intermountain West, even to first order (Louie, 2008). Next-Level ShakeZoning relies on physics and geology to define earthquake shaking hazards, rather than statistics. It follows theoretical and computational developments made over the past 20 years, to capitalize on detailed and specific local data sets and correctly model the propagation and amplification of earthquake waves through the multiple geologic basins of the Intermountain West.

The Building Code demands that peoples’ lives be protected in the event of earthquake shaking. A significant challenge for engineers and urban planners is how to meet the demand for life safety while not making the cost of building impossibly expensive. Two municipalities in southern Nevada, together with the Nevada System of Higher Education, have addressed this challenge with a comprehensive Earthquake Parcel Mapping program. A systematic campaign of ten thousand microzonation measurements across the Las Vegas urban area was the main component of the Parcel Mapping program. With the measurements completed, the partners can now assess the effects that the Parcel Map measurements will have on earthquake ground-shaking predictions, using newly realistic and accurate ShakeZoning procedures. Motivation and objectives

Construction within the last thirty years has experienced a synergistic evolution of seismic code provisions as structures designed and built to contemporary codes have been “put to the test” by experiencing significant ground motions. One of the most significant recent changes to seismic provisions in the building code involves the addition of the “Seismic Site Class” factor that can modify the intensity of ground motion that a building or structure is required to resist. Current building codes include the 2006 and 2009 IBC, which are largely based upon the design recommendations outlined in ASCE 7-05 and the 2003 NEHRP Provisions. These codes all require that the seismic site class factor must be determined based upon the results of onsite tests and evaluation, or otherwise default to the typically low site class value of “D”. In most instances utilizing the default site class value of D produces a conservative design (i.e. over designed). The seismic site class (and several other parameters) are utilized to determine appropriate choices for a structure’s Seismic Force-Resisting System and will also impact the requirements for restraint/anchorage of nonstructural elements: mechanical; plumbing; electrical; and architectural items. Compliance with all of these requirements directly impacts construction costs; in a general sense more seismic force equates to higher costs.

Having a regional microzonation map that correctly identifies the seismic site class provides a valuable asset to a developing community as it assures that structures can be safely designed to meet building code requirements without experiencing unnecessary additional costs due to the inherent conservatism associated with selecting the default parameter value. Undeveloped areas will take full benefit of this information to plan new construction; and developed areas will also benefit when repairs, alterations or additions are performed. The community as a whole will benefit from having an improved knowledge of how local subsurface geology is likely to effect ground motion.

Background

Measurement of shear-wave velocity (Vs) in the shallow subsurface is essential for the estimation of seismic hazard, the development of seismic-hazard maps, and the calibration of recorded ground motion data. The site class measurements are represented by the average shear

3

wave velocity value for the top 100 feet or 30 meters (Vs100-foot or Vs30-meter) as per IBC 2006 Section 1613.5.5. It is an integral to seismic design of structures per the International Building Code and International Residential Code, (IBC) and (IRC) respectively.

Because methods for direct measurement of Vs30 have only recently become cost-effective (Louie, 2001), earthquake-hazard mapping has to date relied upon extrapolation of statistical averages of a few measurements along geological units. Wills and others (2000) built a microzonation map of all of California from only 500 spot measurements of Vs30. Their database shows that adjacent, nearby measurements can vary widely. Thelen and others (2006) and Scott and others (2006) made systematic examinations of the spatial variance of Vs30 from transects of closely spaced measurements stretching across the Los Angeles and Las Vegas urban areas. The velocities would not correlate with any existing soil or geologic mapping, with the standard deviations of the average Vs30 in any unit much larger than the differences in the average between units. These results motivated Clark County and the City of Henderson to embark on a program of comprehensive microzonation measurement.

Earthquake-hazard mapping cannot stop with Vs measurement. The earthquake source, wave-propagation path, and site effects must be combined into a prediction of ground-shaking intensities at a building site, for each of the likely earthquake scenarios that could produce damaging shaking. The USGS ShakeMap tool (Wald and others, 1999) is one of the current, comprehensive means of making scenario computations. ShakeMap is based only on the statistics of relatively sparse recordings of the motions of historical earthquakes. Once shaking levels are predicted from individual scenarios, the scenarios can be combined into a probabilistic seismic hazard map (Frankel and others, 1996).

Our new ShakeZoning procedure for seismic hazard mapping applies the physics of wave propagation through a geologically complex earth. Our Parcel Mapping results in Las Vegas Valley represent a revolution in geotechnical characterization, allowing us to predict wave propagation and shaking amplification across terrain that has been measured to an unprecedented degree of detail. This paper shows the some effects of the new Parcel Map on earthquake-hazard assessment in southern Nevada, by using the map as a ShakeZoning input.

A predecessor to the ShakeZoning concept, the Community Modeling Environment (CME), was developed at the Southern California Earthquake Center (SCEC) under U.S. National Science Foundation Information Technology Research sponsorship. SCEC’s CME combines, in part, geologically based 3-d velocity and fault databases, developed as consensus models in the regional geophysical community (e.g., Magistrale and others, 2000), with a seismic-modeling computational engine (e.g., Olsen, 2000).

The purpose of ShakeZoning (known earlier as MA-CME) was initially to provide a community velocity model and seismic modeling environment for Nevada urban areas. Las Vegas is subject to earthquake hazards both from below the local basin, and from earthquake faults up to 200 km away. ShakeZoning computations feed geological and geotechnical results such as the Parcel Map to the E3D code (Larsen and others, 2001) from Lawrence Livermore National Laboratories for elastic wave-propagation computation. E3D was most recently vetted for ground-shaking prediction at the March 24-25, 2004 Next Generation Attenuation (NGA) Workshop. Such vetting helps to physically and geologically validate ShakeZoning results. Validation will be completed by comparing ShakeZoning seismogram output against earthquake recordings.

4

METHODS

Parcel Mapping The Nevada System of Higher Education (NSHE) was contracted by Clark County

Department of Development Services (CCDDS) and the City of Henderson (CoH) to create a soil classification map based on site-specific shear-wave velocities (“microzonation map”) of the urbanized part of the County (Figure 1). NSHE subcontracted Optim SDS to perform this work under their supervision. The areas characterized included: the urbanized and urbanizing areas of unincorporated Clark County, the urbanized area of the City of Henderson, and portions of the City of Las Vegas within Las Vegas Valley; the proposed Coyote Springs development on the northern border of Clark Co.; the Interstate Highway 15 southern corridor including Jean and Primm; Moapa Valley, Logandale, and Overton; and Laughlin (all shaded on Figure 1). An extension of the project to additional urbanized areas of the City of Las Vegas, and to the City of North Las Vegas, is under development.

Figure 1. Areas within Clark County mapped for earthquake parcel classification, shaded. The areas mapped were defined by CCDDS (shaded red) and CoH (shaded green).

Optim SDS performed all necessary tasks for completion of the work, including field data

acquisition, data processing and reports, working with University personnel to make the products of the work available in a format to be mutually agreed upon during the term of this contract. Optim SDS measured shear velocity as a function of depth at each of the locations using the refraction-microtremor (ReMi) surface array technique. Louie (2001) developed the refraction microtremor technique (commercially available as SeisOpt®ReMiTM, © Optim 2001-2010) as a rapid and cost-effective method of measuring Vs30-meter to meet the IBC code (BSSC, 1997), and to derive site conditions. This method has been peer reviewed and blind tested against both

5

borehole measurements and MASW (multi-channel analysis of surface waves) results (Louie, 2001; Stephenson and others, 2005; Thelen and others, 2006). Refraction microtremor is a volume-averaging surface-wave measurement, averaging velocities where geology is laterally variable, thus providing a more appropriate measure of site effects on earthquake wave propagation than single-point data obtained from downhole logs.

The NEHRP/IBC Site Class is calculated from the measurement results using the equations provided in the 2006 International Building Code book. Section 1613.5.5 of IBC 2006 defines the Vs100-foot vertically averaged shear-wave velocity. Table 1613.5.5 of IBC 2006 defines the NEHRP/IBC site class from the Vs100-foot measurement. Note that even though Vs100-foot values may give you an average velocity that is Site Class A or B, additional criteria defined in the same section of the IBC 2006 code book must be applied by a geotechnical engineer.

In theory, the accuracy and the resolution of any model is a direct function of the spatial density of the database. In determining the appropriate data density to be used, Optim SDS utilized the Southern Nevada Building Code Amendments as a basis for a data density of one velocity sample per 36 acres. To aid in the capturing of evenly dispersed data, based on 1/36 acres, Optim SDS, CCDDS, and the CoH internally developed a gridded subsection scheme based on the existing map-book and section-projections grid of Nevada. Site locations typically were reached via street-approved vehicles, off-highway vehicles (OHV), and hiking; then ReMi arrays were placed alongside roadways, trails, in open fields, or across open desert. However, due to limitations imposed by natural terrain, inaccessible areas, large structures, and infrastructure obstructions, the data density across the entire project area varied, rendering the 1/36 acre data density more of a guideline than an absolute standard.

In addition, in several areas, due to proximity to key infrastructure, identified geology of interest, and velocity anomalies discovered in the data set as the project progressed, data density was increased to better define the interpolated modeled surface in those areas. For instance, seismic array locations were intentionally placed with strategic orientation in mind, crossing known faults and fissure zones when possible. In order to maximize the potential to actually cross any subsurface fault or fissure related features, arrays were placed perpendicular to the assumed projected strike when possible.

All seismic array locations were intended to maintain an average test coverage or data density of one array per 36 acres with no less than 1000 feet measured midpoint to midpoint between each array location. The location of the seismic array is further refined from conditions encountered at the proposed location, such as: traffic and other safety and accessibility concerns; private property concerns; ability of the location to fit an array length of 604 feet (up to 2 channels may be omitted from the array under certain circumstances); location of driveways; relative slope and curvature of the array; the availability of parallel noise sources; and avoiding very high energy perpendicular noise sources and artificial underground voids.

Nominally, each site was measured by a linear ReMi array with 24 channels of vertical geophones spaced at 8 meters. Ambient noise microtremor was recorded using standard vertical 4.5 Hz geophones recorded at a rate of two milliseconds for intervals of 30 seconds each. In all, a total of 12 to 20, 30-second recordings were collected for each seismic array. In addition, hammer hits using an 8-pound or 10-pound sledge and strike plate were collected at approximately 15 feet and 30 feet off both ends during normal passive data collection, to increase high frequency energy. This was especially useful in low energy environments, typically rural settings away from traffic and other ambient noise sources.

6

The noise records were processed using the SeisOpt® ReMi™ software (© Optim, Inc., 2001-2010) that uses the refraction microtremor method (Louie, 2001). Processing steps were: 1) create a velocity spectrum (p-f image) from the noise data by wavefield transformation; 2) Rayleigh-wave dispersion-curve picking along a ``lowest-velocity envelope'' bounding the energy appearing in the p-f image; and 3) interactive shear-wave velocity-profile modeling. The modeling can be subjective and must be done with constraints: that a minimum number of layers be used to accommodate the observed Rayleigh-wave dispersion; and tests on the necessity and sensitivity of the data to both layer thickness and layer velocity. The resultant model is therefore the simplest to explain the data, following the Occam's Razor principle.

Quality assurance of results begins by plotting of the modeled velocity profiles within each section, to assure consistency. If one model differs from surrounding measurements, or is anomalous given known topographic changes, remodeling of the dispersion curve picks is the first step. An alternate model that is more consistent may be able to be derived. For some data sets, re-analysis of the original data may then occur. This may involve adjustment of picks along indistinct dispersion curves or reconsideration of the curve itself. Consistency of models with adjacent sections is also verified. If there are anomalously high-velocity layers within the upper 100 feet, the reliability of low-frequency dispersion curve picks is examined. In some cases, the accuracy of the dispersion curves at low frequencies is unclear. Again, picks may be adjusted and the dispersion curve at low frequencies is reconsidered. If the site is classified as Class B, the high frequency data are more carefully scrutinized to ensure there is velocity information to determine layers in the upper 20 feet.

“Blind” tests were periodically conducted to test the repeatability of the data and analysis results. Seismic data were collected using comparable acquisition equipment by engineer James O’Donnell, of Las Vegas, independently at the same location of seismic arrays obtained by Optim SDS. Test locations were spatially dispersed in the map area where data collection was currently occurring. Data acquisition for the two datasets occurred simultaneously. Excerpts of the same ambient noise were therefore recorded by both datasets. Analysis results of the blind test data were compared with those obtained by Optim SDS. Earthquake ground-shaking prediction

The ShakeZoning tools assemble the available geological and geophysical data sets into a numerical grid for wave-propagation computation, as described by Louie (2008). Louie tutored a group of undergraduate geosciences students in the setup of the simulations. The ShakeZoning results presented here were set up, run on a small Linux cluster, and visualized by the students. We initially develop an example grid to compute the shaking induced in Las Vegas Valley (LVV) by the June 1992 magnitude 5.5 Little Skull Mountain (LSM) earthquake. Figure 2 shows the location of the tilted grid.

The grid includes two datasets at various scales on Neogene basin thicknesses that are roughly stitched together: 1) Saltus and Jachens’s (1995) USGS basin gravity inversions for the Basin and Range, including both sedimentary and Tertiary volcanic basins at 2-km resolution; and 2) the Langenheim and others (1998) Las Vegas basin model from gravity, refraction, and a few deep wells, at 0.4-km resolution. Figure 3 (top center) shows the composite basin-floor model, which sets the LVV basin into its context of the many neighboring basins.

7

Figure 2. Map of southern Nevada centered on Las Vegas showing the ShakeZoning grid for computing a scenario of the 1992 magnitude 5.5 Little Skull Mountain (LSM) earthquake.

We also incorporate geotechnical data sets, despite their having spacings varying from 0.1 to

3 km or more, including isolated point measurements. The ShakeZoning tools interpolate all the disparate data sets onto a regular grid, following instructions for how one data set may take precedence over another where they overlap. For the LSM model, three datasets are combined in this way: 1) the Saltus and Jachens (1995) regional geologic map controls the default shallow geotechnical Vs30-meter value, 250 m/s for basin sites and 760 m/s for rock sites; 2) the Las Vegas refraction microtremor transect, sites measured by B. Luke at UNLV, stratigraphy correlated to 1145 wells by W. Taylor of UNLV and G. Wagoner of LLNL (all from Scott and others, 2006); and 3) the raw 10,721 Parcel Map measurement results discussed below. Figure 3 (top right) shows a geotechnical model developed solely from actual measurements, with no Vs30 prediction.

We set up the example computation in the ShakeZoning tools to yield 0.1-Hz waves on a 236 NE-SW by 501 NW-SE by 41-node deep grid with a dh=dx=dy=dz grid spacing of 0.5 km. The grid includes the LSM source zone near its NW end (star in Figure 3 (lower left). A model of intrinsic seismic attenuation “Q” was developed following Olsen and others (2003). For this historical earthquake the earthquake source parameters are provided by Smith and others (2001). Source parameters for scenario earthquakes are derived from the USGS Qfaults database (USGS and CGS, 2010).

8

Figure 3. (Top left) Road map on the southeast-tilted LSM ShakeZoning computation grid. (Top center) Shaded-relief map on the grid showing the basement-surface component of the ShakeZoning model; warmer, lighter colors at greater basin thicknesses. (Top right) Geotechnical shear-velocity map developed from the Clark County and Henderson projects. Lower velocities are warmer, lighter colors, with Vs30-meter velocities under 0.2 km/sec lightest yellow. (Lower left) Map snapshot of wave propagation modeled from an LSM scenario, at 37.5 sec after earthquake initiation as Rayleigh waves enter Las Vegas Valley (LVV). (Lower right) Peak ground velocity (PGV) map resulting from this 0.1-Hz scenario, with greater shaking as warmer, lighter colors. For this scenario the lightest yellow area had greater than 1.2 cm/sec PGV.

9

Figure 3 (lower left) shows a snapshot of the wave propagation 37.5 seconds after the origin time of the earthquake. White modes in the basins show intensive horizontal shaking from waves trapped inside the low-velocity basin sediments. After 200 seconds of computed wave propagation on this grid, occupying a quad-core Intel CPU for a few hours, ShakeZoning produces a map showing the peak horizontal ground velocity (PGV) a the surface of the grid. (No topography is included.) Figure 3 (lower right) shows the PGV map, with the LVV and surrounding basins getting much greater shaking than bedrock areas at similar distances from the earthquake. Rodgers and others (2006) observed this high degree of basin amplification in recordings throughout Las Vegas Valley of ground motions from earthquakes and NTS explosions. They also observed long durations of shaking within and near the basins. These features of the recorded ground-motion data from the region provide a start to validating our Next-Level ShakeZoning modeling. RESULTS Parcel Mapping

Refraction microtremor data from a total of 10,721 sites were acquired, processed and submitted to CCDDS and the CoH. This represents more than 100% of the total number of seismic lines scheduled to be collected as part of this project. Figure 4 shows the microzonation map generated using ArcGIS and the Vs100-foot values determined from the distribution of seismic arrays. The method of kriging was used to produce this map.

Parallel “blind” tests were conducted at 93 randomly selected sites by J. O’Donnell. He analyzed and modeled the blind-test data independently of Optim SDS and NSHE. The blind-test Vs100-foot velocities were on average only 0.26% percent higher (5 ft/sec out of an average Vs100-foot of about 2000 ft/sec) than the velocities obtained at the same sites by Optim SDS. This tiny average difference shows that there was no significant systematic bias in the data analysis or modeling procedures. The root-mean-squared (RMS) difference between O’Donnell’s blind analyses and Optim SDS’s is 4.92% (98 ft/sec RMS difference). Only six of the 93 blind-test sites showed a difference with a magnitude greater than 10%, and the greatest difference was only 13.55%. These results show the consistent high quality of the Parcel Mapping measurements.

Caliche, a calcareous soil cement, is distributed across much of the Las Vegas region. The high-velocity caliche deposition within subsurface layers can result in high velocity layers within the velocity-depth profile, with lower velocity material below: velocity “reversals.” An assumption is made that the soil is not varying greatly over the short distances considered, and that the caliche is developing in the same sedimentary layer within the section.

It is to be noted that, based purely on the Vs100-foot value, Site Class B values would be shown in the microzonation map (Figure 4). IBC 2006 Section 1613.5.5 states that: “The rock categories, Site Classes A and B, shall not be used if there is more than 10 feet (3048 mm) of soil between the rock surface and the bottom of the spread footing or mat foundation.” So, when the values suggest Class B, site-specific consideration should be made (depth of foundation, competency of rock etc.) before deciding whether it is a Site Class B. At most of the higher-velocity sites, our shear-velocity profiles show more than 10 feet of low-velocity soil at the surface. Accordingly, we are suggesting a Site Class we call “C+” to describe these areas, denoted with red, darkest shading in Figure 4.

10

The measured Parcel Map shows a clearly definable C+ to C boundary on the west side of Las Vegas Valley (red to green or darkest to lightest shading in Figure 4). The boundary may be associated with the bases of the alluvial fans emanating from the Spring Mountains, where the sediments become finer-grained (Scott and others, 2006). The C to D boundary (green to blue or lightest to medium shading) is on the other hand much more complex, and difficult to associate with any known geological features.

Figure 4. Map showing the IBC seismic zoning results of the Earthquake Parcel Mapping projects sponsored by Clark County and the City of Henderson. Black lines are municipal and state boundaries. The IBC “D” zone is blue (medium shading); the IBC “C” zone is green (lightest shading); and the proposed IBC “C+” zone is red (darkest shading).

11

Earthquake ground-shaking prediction There are nine potential earthquake faults in the USGS Qfaults database (2010) that the

USGS National Seismic Hazard Maps (Frankel and others, 1996) rate as probabilistically important to earthquake shaking potential in Las Vegas Valley. The Little Skull Mountain earthquake scenario is one example. We use the LSM scenario at 0.1 Hz to demonstrate the crucial effects the Parcel Map results have in the Valley, even on such low-frequency ground shaking. Figure 5 shows this effect, which is unexpectedly strong given that the wavelengths of the 0.1-Hz waves modeled are 100 times larger than the 30-meter depth extent of the Parcel Map measurements.

At such long wavelengths the effects of the Vs30-meter variations found in the Parcel Map (Figure 5, upper left) should be minuscule. The amplifications (red, darker shading) and de-amplifications (blue) are predicted in Figure 5 (upper right) from the measured Vs30 relative to the IBC assumptions only, by the one-dimensional equations in the IBC Code (2006) and in ShakeMaps (Wald and others, 1999; Worden and others, 2010). The 1-d amplifications are up to 64% in Figure 5 (upper right). These amplifications should only appear at much higher frequencies, above 1 Hz. The 0.1-Hz waves should respond mostly to the geologic basin, Figure 5 (lower left).

However, Figure 5 (lower right) shows 0.1-Hz amplifications from Next-Level ShakeZoning, relative to an IBC default geotechnical map, of up to 36%. The 30-meter-deep Parcel Map was the only variant between the two ratioed scenarios. This is from fully 3-d, physically and geologically accurate ShakeZoning computation. Further, the unexpectedly high amplification (red, darker shading on Figure 5, lower right) is not located at either the location of the lowest Parcel Map velocities, nor at the greatest basin depth (nor steepest basin-floor slope). As well, the 64% de-amplifications (blue) on the west side of the Valley predicted by the 1-d model in Figure 5 (upper right) are not supported by the realistic 3-d computations. This observation suggests that the current IBC may be too permissive in some higher-velocity areas.

Figure 6 shows an initial test of these observations at higher frequencies of more engineering interest, up to 0.5 Hz or 2 seconds period. We configured a simulation of a M6.5 earthquake scenario on the Black Hills fault bordering Eldorado Valley, southeast of Las Vegas Valley. The Next-Level ShakeZoning simulation at this higher frequency produces peak ground velocities above 60 cm/sec throughout Eldorado Valley (yellow in Figure 6, left). Further, high ground motions escape into Las Vegas Valley through a small connecting basin filled with the basalts at Black Mountain.

At 0.5 Hz frequency, the effect of Clark County’s Parcel Map (Figure 5, upper left) should be prominent. Taking the ratio between two scenarios, the numerator with the grid having been built with the Parcel Map, and the denominator with its grid built without the Parcel Map, shows both large amplifications (red and yellow in Figure 6, right) and de-amplifications (blue and black in Figure 6, right). The inclusion of the Parcel Map was the only difference between the two scenarios.

The Clark County Parcel Map geotechnical velocities are clearly producing factors of two amplification and de-amplification over the IBC default soil properties. As well, these amplifications impact areas far beyond the coverage of the Parcel Map measurements, through the processes of wave propagation and basin trapping (Figure 6). Although the Parcel Map cannot lead to any one-to-one correspondence between velocity measurement and predicted earthquake ground motion, it is an essential component of a valid ShakeZoning simulation.

12

Figure 5. Maps of the Las Vegas Valley (LVV) section of the tilted LSM ShakeZoning scenario grid. (Upper left) Geotechnical map with ~9000 Parcel Mapping Project Vs30-meter measurements included, using the IBC default velocities for areas the Parcel Map does not cover. (Upper right) Shaking amplification expected from the ratios of measured geotechnical velocities over the IBC default velocities, under a simple 1-d model. Amplification of 120% or more is shown as red, and de-amplification to 80% or less is blue. (Lower left) LVV map showing the basin floor in shaded relief, with warmer, lighter colors for greater basin thicknesses. (Lower right) ShakeZoning PGV ratio map comparing the 0.1-Hz results of the LSM scenario, comparing the Parcel Mapping result against the IBC default result. Even at these very low frequencies, 0.1 Hz, where the shallow geotechnical velocities should have minimal effect, comparing the physically correct ShakeZoning computation against the IBC defaults (above) shows surprisingly large and complex patterns of amplification (red) and de-amplification (blue), greater than 20%.

13

Figure 6. (Left) Next-Level ShakeZoning PGV map from a M6.5 earthquake scenario on the Black Hills fault, computed at 0.5-Hz frequency. Yellow areas show above 60 cm/sec peak ground motion, filling Eldorado Valley and projecting very high ground motions into Las Vegas Valley. A gray shaded-relief view of the basin floors is superimposed, as is the main highway network as white lines and the principal fault traces as red lines. (Right) PGV ratio map showing Black Hills scenario motions computed with the Clark County Parcel Map (at left), divided by the motions computed from the same earthquake scenario but using only the IBC default geotechnical velocities, without the Parcel Map, also with gray basin-floor shaded relief. Red shows amplifications produced by including the Parcel Map, saturating at a factor of two (yellow, seen at the Apex landfill and on East Vegas Valley Blvd.). Blue shows PGV de-amplifications relative to IBC defaults produced by the Parcel Map, saturating at a factor of one-half (black, seen at Sun City). White or gray shows no relative amplification due to the Parcel Map. DISCUSSION AND CONCLUSIONS

The Parcel Mapping project was completed successfully for Clark County and the City of

Henderson. In all, 10721 individual seismic arrays/lines were assigned, deployed and spatially located in a systematic manner, maximizing the density of the database coverage within urban Clark County. This achievement has put southern Nevada at the forefront of earthquake-hazard mitigation efforts worldwide.

Based on the GIS seismic shear-velocity database developed by NSHE and Optim SDS, the project provides Clark County and the City of Henderson a single shear-wave velocity-based Parcel Map with contoured values corresponding to the site classifications of the IBC and NEHRP definitions for site class. Velocity databases from the CCDDS and the CoH were integrated into a single Vs database and the interpolated velocity map was based on the entire joint database, providing seamless interpolation across the CCDDS and CoH boundaries. This

14

Parcel Map will provide city planners, building officials, design professionals, and researchers alike the opportunity to determine the actual site class value of a particular parcel before a single site-specific investigation is performed.

We show the Parcel Map to be essential for the accurate prediction of prominent features of the earthquake ground shaking across Las Vegas Valley as observed by Rodgers and others (2006) from recorded events, such as large basin amplifications. The Parcel Map is needed even at the low wave frequencies that are of most interest to the taller buildings in the area. Scenario computations up to 0.5 Hz frequency are now practical and show even greater effects of the Parcel Map, and have greater applicability to large numbers of buildings. Further and higher frequency scenarios are under development, for the eight hazardous earthquake faults in the region, and for frequencies up to 1 Hz. The newly accurate ShakeZoning tools developed in this effort will be made available to aid engineers and planners in the region. Preliminary tools can be downloaded from http://crack.seismo.unr.edu/ma.

REFERENCES

Building Seismic Safety Council, BSSC, 1997, NEHRP Recommended Provisions for Seismic

Regulations for New Buildings and other Structures, Part1 – Provisions, Federal Emergency Management Agency, Washington D.C., and FEMA 302.

Frankel, A., Mueller, C., T. Barnhard, D. Perkins, E.V. Leyendecker, N. Dickman, S. Hansen, and M. Hopper, 1996, National seismic-hazard maps: documentation, U.S. Geol. Surv. Open-File Rept. 96-352.

Langenheim, V. E., Grow, J., Miller, J. J., Davidson, J. D., and Robison, E., 1998, Thickness of Cenozoic deposits and location and geometry of the Las Vegas Valley shear zone, Nevada, based on gravity, seismic-reflection, and aeromagnetic data: U. S. Geol. Survey Open-File Report OF 98-0576, 32 pp.

Larsen, S., Wiley, R., Roberts, P., and House, L., 2001, Next-generation numerical modeling: incorporating elasticity, anisotropy and attenuation: SEG Expanded Abstracts, 20, p. 1218-1221.

Louie, J, N., 2001, Faster, Better: Shear-wave velocity to 100 meters depth from refraction microtremor arrays: Bull. Seism. Soc. Am., 91, p. 347-364.

Louie, John N., 2008, Assembling a Nevada 3-d velocity model: earthquake-wave propagation in the Basin & Range, and seismic shaking predictions for Las Vegas: SEG Expanded Abstracts, 27, p. 2166-2170.

Magistrale, H., S. Day, R. W. Clayton, and R. Graves, 2000, The SCEC Southern California Reference Three-Dimensional Seismic Velocity Model Version 2: Bull. Seism. Soc. Am., 90(6B), p. S65–S76.

Olsen, K. B., 2000, Site amplification in the Los Angeles Basin from three-dimensional modeling of ground motion. Bull. Seism. Soc. Am., 90(6B), p. S77-S94.

15

Olsen, K. B., S. M. Day, and C. R. Bradley, 2003, Estimation of Q for long-period (>2 sec) waves in the Los Angeles basin: Bull. Seism. Soc. Am., 93, p. 627–638.

Rodgers, Arthur, Hrvoje Tkalcic, David Mccallen, Shawn Larsen, and Catherine Snelson, 2006, Site response in Las Vegas Valley, Nevada from NTS explosions and earthquake data: Pure & Appl. Geophys., 163, p. 55–80.

Saltus, R. W., and Jachens, R. C., 1995, Gravity and basin-depth maps of the Basin and Range Province, Western United States: U.S. Geological Survey, Geophysical Investigations Map, Report: GP-1012, 1 sheet.

Scott, J. B., T. Rasmussen, B. Luke, W. Taylor, J. L. Wagoner, S. B. Smith, and J. N. Louie, 2006, Shallow shear velocity and seismic microzonation of the urban Las Vegas, Nevada basin: Bull. Seism. Soc. Am., 96(3), p. 1055-1067, doi: 10.1785/0120040093.

Smith, Kenneth D., Brune, James N., de Polo, Diane, Savage, Martha K., Anooshehpoor, Rasool, and Sheeham, Anne F., 2001, The 1992 Little Skull Mountain earthquake sequence, southern Nevada Test Site: Bull. Seism. Soc. Am., 91, p. 1595-1606.

Stephenson, W. J., J. N. Louie, S. Pullammanappallil, R. A. Williams, and J. K. Odum, 2005, Blind shear-wave velocity comparison of ReMi and MASW results with boreholes to 200 m in Santa Clara Valley: Implications for earthquake ground motion assessment, Bull. Seism. Soc. Am., 95, p. 2506-2516, doi: 10.1785/0120040240.

Thelen, W. A., M. Clark, C. T. Lopez, C. Loughner, H. Park, J. B. Scott, S. B. Smith, B. Greschke, and J. N. Louie 2006, A transect of 200 shallow shear-velocity profiles across the Los Angeles basin, Bull. Seism. Soc. Am., 96, p. 1055-1067.

U.S. Geological Survey and California Geologic Survey, 2010, Quaternary fault and fold database for the United States, accessed 12/20/2010, from USGS web site: http://earthquake.usgs.gov/hazards/qfaults/.

Wald, David J., Vincent. Quitoriano, Tom. H. Heaton, Hiroo. Kanamori, Craig. W. Scrivner, and C. Bruce Worden, 1999, TriNet ``ShakeMaps'': Rapid Generation of Instrumental Ground Motion and Intensity Maps for Earthquakes in Southern California: Earthquake Spectra, 15, p. 537-556.

Wills, C. J., M. D. Petersen, W. A. Bryant, M. S. Reichle, G. J. Saucedo, S. S. Tan, G. C. Taylor, and J. A. Treiman 2000, A site-conditions map for California based on geology and shear wave velocity, Bull. Seism. Soc. Am., 90, p. S187-S208.

Worden, C. B., D. J. Wald, T. I. Allen, K. Lin, D. Garcia, and G. Cua, 2010, A revised ground-motion and intensity interpolation scheme for ShakeMap, Bull. Seism. Soc. Am., 100, p. 3083–3096.