S51A-0210
Quantifying and Qualifying USGS ShakeMap Uncertainty
We describe newly-developed algorithms for quantifying and qualifying uncertainties associated with USGS ShakeMap ground motions. We calculate an uncertainty measure as a function of location on the map dependent on the underlying ground motion prediction equation. Uncertainty is high for larger magnitude earthquakes when finiteness is not yet constrained (and hence the rupture distance is also uncertain); uncertainty at each grid cell is decreased by the proximity of seismic stations. This grid-based "uncertainty map" is essential for probabilistically evaluating the losses that derive from hazard modeling based on ShakeMaps. However, many users may prefer a qualitative grade for the entire ShakeMap. This grading scale will allow them to quickly gauge the appropriate level of confidence when using rapidly produced ShakeMaps as part of the post-earthquake critical decision- making process. To this end, we describe a new uncertainty letter grading ("A" through "F") based on the uncertainty map. The letter grades are calculated from the mean uncertainty over the map area at grid cells that have intensities of VI or greater. This constraint focuses the calculation of uncertainty only on portions of the map with potentially damaging ground motions (inclusion of lower intensities would also make the average uncertainty dependent on map scaling.) A middle range ("C") grade corresponds to a ShakeMap for a moderate magnitude earthquake modeled with a point source. Lower grades "D" and "F" are assigned for larger events (M>6) where finite source dimensions are not constrained. The addition of ground motion observations (or observed macroseismic intensities) to constrain ground motions has the opposite effect, reducing uncertainties over data-constrained portions of the map. Higher grades ("A" and "B") correspond to ShakeMaps with constrained fault dimension and numerous stations, depending on the density of station/data coverage. Thus the letter grade changes with subsequent ShakeMap revisions as more data are added or when finite-faulting dimensions are established. http://earthquake.usgs.gov/eqcenter/shakemap/
S51A-0211
USGS ShakeMap Developments, Implementation, and Derivative Tools
We discuss ongoing development and enhancements of ShakeMap, a system for automatically generating maps of ground shaking and intensity in the minutes following an earthquake. The rapid availability of these maps is of particular value to emergency response organizations, utilities, insurance companies, government decision- makers, the media, and the general public. ShakeMap Version 3.2 was released in March, 2007, on a download site which allows ShakeMap developers to track operators' updates and provide follow-up information; V3.2 has now been downloaded in 15 countries. The V3.2 release supports LINUX in addition to other UNIX operating systems and adds enhancements to XML, KML, metadata, and other products. We have also added an uncertainty measure, quantified as a function of spatial location. Uncertainty is essential for evaluating the range of possible losses. Though not released in V3.2, we will describe a new quantitative uncertainty letter grading for each ShakeMap produced, allowing users to gauge the appropriate level of confidence when using rapidly produced ShakeMaps as part of their post-earthquake critical decision-making process. Since the V3.2 release, several new ground motion predictions equations have also been added to the prediction equation modules. ShakeMap is implemented in several new regions as reported in this Session. Within the U.S., robust systems serve California, Nevada, Utah, Washington and Oregon, Hawaii, and Anchorage. Additional systems are in development and efforts to provide backup capabilities for all Advanced National Seismic System (ANSS) regions at the National Earthquake Information Center are underway. Outside the U.S., this Session has descriptions of ShakeMap systems in Italy, Switzerland, Romania, and Turkey, among other countries. We also describe our predictive global ShakeMap system for the rapid evaluation of significant earthquakes globally for the Prompt Assessment of Global Earthquakes for Response (PAGER) system. These global ShakeMaps are constrained by rapidly gathered intensity data via the Internet and by finite fault and aftershock analyses for portraying fault rupture dimensions. As part of the PAGER loss calibration process we have produced an Atlas of ShakeMaps for significant earthquakes around the globe since 1973 (Allen and others, this Session); these Atlas events have additional constraints provided by archival strong motion, faulting dimensions, and macroseismic intensity data. We also describe derivative tools for further utilizing ShakeMap including ShakeCast, a fully automated system for delivering specific ShakeMap products to critical users and triggering established post-earthquake response protocols. We have released ShakeCast Version 2.0 (Lin and others, this Session), which allows RSS feeds for automatically receiving ShakeMap files, auto-launching of post-download processing scripts, and delivering notifications based on users' likely facility damage states derived from ShakeMap shaking parameters. As part of our efforts to produce estimated ShakeMaps globally, we have developed a procedure for deriving Vs30 estimates from correlations with topographic slope, and we have now implemented a global Vs30 Server, allowing users to generate Vs30 maps for custom user-selected regions around the globe (Allen and Wald, this Session). Finally, as a further derivative product of the ShakeMap Atlas project, we will present a shaking hazard Map for the past 30 years based on approximately 3,900 earthquake ShakeMaps of historic earthquakes. http://earthquake.usgs.gov/shakemap
S51A-0212
ShakeCast: Automating and Improving the Use of ShakeMap for Post-Earthquake Decision- Making and Response
ShakeCast is a freely available, post-earthquake situational awareness application that automatically retrieves earthquake shaking data from ShakeMap, compares intensity measures against users" facilities, sends notifications of potential damage to responsible parties, and generates facility damage maps and other Web-based products for emergency managers and responders. ShakeMap, a tool used to portray the extent of potentially damaging shaking following an earthquake, provides overall information regarding the affected areas. When a potentially damaging earthquake occurs, utility and other lifeline managers, emergency responders, and other critical users have an urgent need for information about the impact on their particular facilities so they can make appropriate decisions and take quick actions to ensure safety and restore system functionality. To this end, ShakeCast estimates the potential damage to a user's widely distributed facilities by comparing the complex shaking distribution with the potentially highly variable damageability of their inventory to provide a simple, hierarchical list and maps showing structures or facilities most likely impacted. All ShakeMap and ShakeCast files and products are non-propriety to simplify interfacing with existing users" response tools and to encourage user-made enhancement to the software. ShakeCast uses standard RSS and HTTP requests to communicate with the USGS Web servers that host ShakeMaps, which are widely-distributed and heavily mirrored. The RSS approach allows ShakeCast users to initiate and receive selected ShakeMap products and information on software updates. To assess facility damage estimates, ShakeCast users can combine measured or estimated ground motion parameters with damage relationships that can be pre-computed, use one of these ground motion parameters as input, and produce a multi-state discrete output of damage likelihood. Presently three common approaches are being used to provide users with an indication of damage: HAZUS-based, intensity-based, and customized damage functions. Intensity-based thresholds are for locations with poorly established damage relationships; custom damage levels are for advanced ShakeCast users such as Caltrans which produces its own set of damage functions that correspond to the specific details of each California bridge or overpass in its jurisdiction. For users whose portfolio of structures is comprised of common, standard designs, ShakeCast offers a simplified structural damage-state estimation capability adapted from the HAZUS-MH earthquake module (NIBS and FEMA, 2003). Currently the simplified fragility settings consist of 128 combinations of HAZUS model building types, construction materials, building heights, and building-code eras.
S51A-0213
ShakeMap® implementation in Italy
Italy is a seismically active country which has been site of several large and extremely damaging earthquakes causing hundreds to tens of thousands of casualties since historical times. In recent years, the "Dipartimento per la Protezione Civile" (DPC; Italian Civil Protection – an office dependent directly from the prime minister) has supported the project S4 driven specifically toward fast assessment of ground motion shaking in Italy in order to organize the emergency and direct the rescue teams. The Istituto Nazionale di Geofisica e Vulcanologia (INGV) has implemented the software package USGS-ShakeMap® to obtain maps of the peak ground motion parameters (PGM), and of the instrumentally derived intensities. The calculation of shakemaps at INGV relies mainly on broadband and strong motion data acquired by the Italian seismic network. Shakemap PGM data feeding relies on two concurrent seismic acquisition systems and maps are be determined as quickly as 5 minutes from origin time automatically or within 30 minutes using manually revised locations for earthquakes occurring on the national territory. In Italy, attenuation has been found to vary between different regions. For the smaller events (up to M 5.5) we have implemented a six-areas regionalized model (three separate sets of equations) following the relations of Malagnini and co-workers (Malagnini, et al., 2000; Malagnini, et al., 2002; Morasca, et al., 2006), the same used by the National Seismic Hazard Working Group (2004) for the compilation of the national hazard map; for larger earthquakes, the strong-motion-based equations by Ambraseys et al. (1996) are used. For the site corrections, we have implemented a classification based on the 1:100,000 geology map of Italy compiled and published by the "Servizio Geologico Nazionale". In this case, the geologic units have been gathered into five different classes A, B, C, D and E according to the EuroCode8 provisions, EC8, after Draft 6 of January 2003 on the base of the ground acceleration response. This site classification is compared to that that can be obtained from the analysis of the topographic relief (Allen and Wald, 2007). Examples of shakemaps for both recent M4 size earthquakes and for large, instrumentally recorded 20th century earthquakes that have occurred in Italy will be shown.
S51A-0214
Real-Time Shake Maps In The Southern Alps-Dinarides Junction Area
Shake maps are generated for the Southern Alps-Dinarides Junction area within 5 minutes from the earthquake occurrence. They are based on the TriNet "ShakeMap" software (Wald et al., 1999) interfaced with the Antelope real-time system (ARTS) installed at DST in the framework of an EU Interreg project. The system retrieves the real time waveforms and the results are posted automatically on the Department web pages (www.dst.units.it/RAF06). For the calibration of the Shake Maps regional model we use the NEHRP classification and two ground-motion relations for different magnitude ranges. The regional ground-motion relations (Costa et al., 2006), computed using strong-motion data recorded from the Friuli Venezia Giulia and the Italian National Accelerometric integrated network, are used for moderate (ML<6.3) events, whereas the Sabetta and Pugliese (1996) relation valid for the Italian territory is used for ML>6.3 events. The system has been successfully tested for medium to moderate events (ML<4.5) occurred in the area since January 2007. Ground-shaking synthetic scenarios are generated for three important past seismic events in the studied area: Cansiglio 1936, Friuli 1976 and Bovec 1998. The used synthetic seismograms are computed applying the reflectivity method (Kennett, 1981) at a cutoff frequency of 10 Hz. The satisfactory validation of these scenarios has been done by computing the misfit values between the observed and estimated accelerations, and the misfit values between the macroseismic observations (INGV-DBMI04 database) and the instrumental intensities derived from the application of different relationships (Wald et al., 1999; Faccioli and Cauzzi, 2006; Kastli and Fah, 2006). http://www.dst.units.it/RAF06
S51A-0215
Implementation of the Near Real Time ShakeMap System in Romania
The USGS ShakeMap program is a tool for the near real-time generation of ground-shaking maps following data from seismic station sites equipped with accelerometers and velocity sensors. These maps provide vital information within minutes after an earthquake to emergency response agencies. ShakeMap data at the Romania National Data Center (RO_NDC) is based on input data from the Antelope system used for real-time seismic data acquisition and automatic detection. Shakemaps are produced on the basis of observed ground motion values (peak velocities, peak accelerations) and Intensity for the Vrancea region based on epicentral distance. Currently, the ShakeMap system is functional and is generating real-time shakemaps in test mode using data from more than 20 broadband and 20 strong motion stations. In the future about 35 strong motion stations will be added for real time acquisition in the Antelope system and in ShakeMap. The site corrections are based on the average shear-wave velocity values for the uppermost 30 meters (Vs30) estimated from the surface topography using the global Vs30 server (Wald et al., 2004 and Wald and Allen, 2007). However, research to utilize geologic information for the Vs30 calculations is in progress. Attenuation relations based on local data are included with the ShakeMap installation for interpolation of ground motions between the recording stations. Shakemaps generated based only on these relations for historical earthquakes leads to satisfactory estimates of the macroseismic intensity distribution due to these events. Inclusion of recorded data in such estimates, however, will lead to more reliable shakemaps.
S51A-0216
ShakeMaps at the Swiss Seismological Service: Current Status, Innovations, and Outlook
The ShakeMap-related efforts at the Swiss Seismological Service (SED) are motivated by the desire for improved emergency response and information dissemination following a large earthquake. The SED implementation approach has evolved over the years from developing in-house "ShakeMap"-type software to that of customizing the USGS ShakeMap codes to Swiss conditions. As part of efforts to calibrate the Swiss ShakeMap system, we have undertaken the following: 1) We use recordings from small-to-moderate Swiss earthquakes in combination with relevant strong motion data from the PEER strong motion database of crustal earthquakes to develop attenuation relationships for PGA, PGV, and response spectra at 0.3, 1.0, and 3.0 second periods valid in the magnitude range 2.5 &<& M &<& 8 using the Cua and Heaton (2007) approach. Aside from their applications in the ShakeMap system, these equations can be used to quantify the similarities (and differences) of average ground motion characteristics between southern California and Switzerland, and provide additional insight into the question of whether attenuation relationships derived from a particular region are transportable to other regions of the same tectonic setting. 2) The USGS ShakeMap system uses a uniform grid of estimated Vs30 (average shear wave velocity in the upper 30 meters) to characterize site amplification. However, Kästli and Fäh (1st European Conference on Earthquake Engineering and Seismology, 2006) have suggested that using geotechnical characteristics such as Vs30 are not as effective in predicting site amplification in Switzerland, due to its heterogeneous geology and small-scale structures, than in other regions with more homogenous sedimentary basins. We explore the use of intensity-based amplification factors derived from an extensive Swiss database of macroseismic intensity observations in characterizing site amplification. We use the Kästli and Fäh (2006) ground motion to intensity relationship to convert intensity amplification factors into "equivalent Vs30" values, and using these as inputs to our ShakeMap system. We evaluate the effectiveness of these approaches by comparing the ShakeMap-estimated instrumental intensities from a suite of well-recorded M&>&3 over the past 10 years with the macroseismic intensity reports from these events.
S51A-0217
ShakeMap Implementation in Alaska
The ShakeMap (SM) system was developed by the USGS for generating and distributing real-time ground- shaking maps in the aftermath of significant earthquakes. SMs provide vital information within minutes after an earthquake to emergency response agencies, the media and the general public. It is also a tool to produce earthquake planning scenarios and to estimate losses from hypothetical strong earthquakes. SM production in Alaska is based on observed ground motion data (maximum peak ground accelerations and velocities of two horizontal components) and complemented by calculated values using empirical attenuation relationships. These data are collected from more than 80 broadband and 25 strong motion stations throughout the state. The real-time seismic operations in Alaska, including the SM system, are maintained at the Alaska Earthquake Information Center (AEIC) of the Geophysical Institute in Fairbanks. The earthquake parameters and waveform measurements are obtained within the Antelope seismic monitoring system. Currently, SMs are produced for events with magnitudes greater that M3.5 with at least 10 associated arrival picks. Moreover, the calculated intensity of the eligible events should be greater than 2.5 at the epicenter. With these settings, about 20 to 30 SMs are triggered in Alaska per month. The maps are generated and posted on the AEIC website 2-3 minutes after the event. The processing time mostly depends on the number of waveforms utilized in the calculation. Several SM updates may be issued for the same event as more reliable data become available. A manual run may be executed afterwards for significant events in order to utilize any additional information, such as extended source geometry or data from external sources.
S51A-0218
ShakeMap at the Pacific Northwest Seismic Network
We summarize efforts to tailor ShakeMap to the Pacific Northwest Seismic Network (PNSN), and to increase the resolution in the major urban areas. Our initial implementation of ShakeMap employed parameters based on data from mostly larger earthquakes outside the Pacific Northwest. The PNSN automatically generates a 45- arcsec ShakeMap for any earthquake of Md ≥ 3.0 in the Puget Sound region and for Md ≥ 4.0 earthquakes in Washington and Oregon. ShakeMap uses 3-component, real-time data from 91 strong motion, 34 broadband, and 70 Earthscope Transportable Array stations. We also automatically incorporate data from dial- up stations of the National Strong Motion Program. High-resolution (7.2 arcsec) ShakeMaps for the Seattle area have just come on-line thanks to the availability of a new, more detailed geologic map. We use data from the PNSN and other sources to derive new region-specific ground motion attenuation relations and site corrections. Preliminary results suggest that default attenuation relations included with the ShakeMap package over-predict Pacific Northwest ground motions, especially at larger distances and for deep (≥ 20km) earthquakes. We will make similar comparisons with the Next Generation Attenuation relations, to be used in future ShakeMap releases, and modify the relations if warranted. To improve site corrections we compared site amplification measurements and local magnitude residuals (available for most of the PNSN stations) to measured values of Vs30 at co-located and nearby sites. These correlate well and thus provide useful proxies for Vs30 for use in ShakeMap. We also found correlations between Vs30 estimates and the age and rock type of mapped geologic units, providing a geologically constrained means of interpolating between sites with more direct Vs30 estimates. Finally, our ShakeMaps will improve because we continue to add stations to the PNSN, providing additional direct measures of ground motion.
S51A-0219
Development of Shakemap Methodologies
For almost-real time estimation of the ground shaking after a major earthquake in the Euro-Mediterranean region the JRA-3 component of the EU Project entitled "Network of research Infrastructures for European Seismology, NERIES" foresees: 1. Finding of the most likely location of the source of the earthquake using regional seismotectonic data base, supported, if and when possible, by the estimation of fault rupture parameters from rapid inversion of data from on-line regional broadband stations. 2. Estimation of the spatial distribution of selected ground motion parameters at engineering bedrock through region specific ground motion attenuation relationships and/or actual physical simulation of ground motion. 3. Estimation of the spatial distribution of site-specific ground selected motion parameters using regional geology (or urban geotechnical information) data-base using appropriate amplification models. 4. Correlation/verifyication/enrichment of the estimated ground shaking information with the available on-line strong motion data. These shakemaps maps will enable the first estimates of damage and casulaties, and consequently provide vital information within minutes after an earthquake to European emergency response agencies. There are regional dependencies and multiple sources of uncertainty in such estimations stemming from empirical ground motion predictions, non-existent or sparse ground motion measurements, considerations involving fault finiteness and directivity, data interpolations and site modifications. The methodology used by USGS is not directly transportable, due to a region specific characteristics of the empirical attenuation relationships and methods of modification of ground motion for the near-surface geology. Together with researchers from Imperial College, NORSAR and ETH-Zurich, we are developing and testing algorithms for the preparation of shakemaps and the quantification of associated uncertainties. In this connection we have estimated the distribution of EMS'98 intensities and assesed the associated uncertainties in the Marmara Region associated with the 1999 Kocaeli Earthquake 1. Through available strong motion data and using various intensity correlations with PGV, PGA, Fourier Amplitude Spectrum and Response Spectrum 2. Through use of local and regional one- and two-dimensional intensity attenuation relationships based on regression of regressions of intensity with the size of the earthquake, distance and site geology 3. Through synthetic simulation of strong ground motion and using various intensity correlations with PGV, PGA, Fourier Amplitude Spectrum and Response Spectrum, and 4. Through routine application of the USGS Shakemap software. These procedures are critically compared with the observed intensity distributions.
S51A-0220
A New Method For Rapid Computation Of Earthquake Ground Shaking Maps
Strong ground-shaking mapping soon after a moderate-to-large earthquake is crucial to recognize the areas which suffered for largest damages and losses. These maps play a fundamental role for directing first-aid emergency rescues, for loss estimation and the planning of emergency actions. The most critical issue in strong ground-shaking map computation, are the assumed correlation between one or more selected ground motion parameters (e.g., Pga, Pgv, Sa(T)) and damage and reliability of ground motion predictions in areas not covered by the seismic network. We propose a new technique for rapid computation of ground-shaking maps after moderate-to-large earthquakes based on an optimal data gridding. The gridding technique uses a triangulation where recording stations are the vertices of the triangles and ground motion measurements are used to correct predicted data at the barycentres of each triangle. This scheme allows to account for bidimensional characteristics of the ground-shaking with respect to the predictions obtained by using empirical ground motion models which depend only on magnitude and distance. As a consequence, the computed map accounts for earthquake specific features, such as rupture extension, radiation pattern and directivity to which larger part of damages can be correlated. The method is tested off-line on a set of worldwide M > 6.5 earthquakes with different fault characteristics and located in different tectonic environments. The resulting maps, which refer to rock site conditions, are compared with those obtained by using the software ShakeMap®. The comparison is devoted to outline the main differences in the interpolation and gridding schemes adopted by the two techniques and the capability to display some features of the selected earthquake by looking at the ground motion parameters distribution. However, due to the main goal of the ground-shaking maps, more remarkable is the comparison between predicted and observed instrumental intensities field when available.
S51A-0221
Near Real Time Earthquake Loss Assessment Methodology for Europe
Main objective of this study is to develop a methodology for real time estimation of losses after a major earthquake in the Euro-Mediterranean region. This study complements the work on ShakeMap methodologies that is being conducted within the framework of the JRA-3 component of the EU Project "Network of research Infrastructures for European Seismology, NERIES". The multi-level methodology being developed together with researchers from Imperial College, NORSAR and ETH-Zurich is capable of incorporating regional variabilities and sources of uncertainty stemming from ground motion predictions, fault finiteness, site modifications, and inventory of elements subjected to earthquake hazard and the associated vulnerability relationships. Within the scope of this paper, results obtained from a pilot application of this methodology for the 1999 Kocaeli earthquake are presented and comparisons with the observed losses are made.
S51A-0222
A new Tool for Estimating Losses due to Earthquakes: QUAKELOSS2
WAPMERR and the Swiss Seismological Service are developing new software for estimating mean damage to buildings, number of injured and number of fatalities due to earthquakes worldwide. The focus for applications is real-time estimates of losses after earthquakes in countries without dense seismograph networks, and results that are easy to digest by relief agencies. Therefore, the standard version of the software addresses losses by settlement, subdivisions of settlements and important pieces of infrastructure. However, a generic design, an open source policy and well defined interfaces will allow the software to work on any gridded or discrete building stock data, to do Monte-Carlo simulations for error assessment and to plug in more elaborate source models than simple point and line sources and thus to compute realistic loss scenarios as well as probabilistic risk maps. It will provide interfaces to SHAKEMAP and PAGER, such that innovations developed for the latter programs may be used in QUAKELOSS2, and vice versa. A client server design will provide a front-end web interface where the user may directly manage servers as well as run the software in one's&pown laboratory. The input-output features and mapping will be designed to allow the user to run QUAKELOSS2 remotely with basic functions, as well as in a laboratory setting including a full-featured GIS setup for additional analysis. In many cases, the input data (earthquake parameters as well as population and building stock data) are poorly known for developing countries. Calibration of loss estimates, using past earthquakes that have caused damage and WAPMERR's experience of four years" estimating losses, will help to produce approximately correct results in countries with strong earthquake activity. A worldwide standard dataset on population and building stock will be provided as open source together with the software. The dataset will be improved successively, based on input from satellite images, international databases, local specialists and formalized feedback by the seismological user community. http://www.wapmerr.org
S51A-0223
Linking ShakeMap and Emergency Managers in the Utah Region
In 2001, the University of Utah Seismograph Stations (UUSS) locally customized and began producing automatic ShakeMaps in Utah's Wasatch Front urban corridor as part of a new real-time earthquake information system developed under the Advanced National Seismic System. In 2005, motivated by requests from Utah's Division of Homeland Security and FEMA, ShakeMap capabilities were expanded to cover the entire Utah region. Now in 2007, ShakeMap capabilities throughout the region will again be enhanced by increased station coverage. The increased station coverage comes both from permanent stations funded by a state initiative and from the temporary deployment of EarthScope USArray stations. The state initiative will add ~22 strong-motion instruments and ~10 broadband instruments to the UUSS network. The majority of these stations will be located in southwestern Utah--one of the fastest growing regions in the U.S. EarthScope will evenly distribute 70 broadband stations in the region during 2007 that will be removed after 18 to 24 months. In addition to the enhanced station coverage for producing ShakeMaps in the Utah region, the transfer of information to the emergency response community is also being enhanced. First, tools are being developed that will link ShakeMap data with HAZUS loss-estimation software in near-real-time for rapid impact assessment. Second, ShakeMap scenarios are being used in conjunction with HAZUS loss-estimation software to produce customized maps for planning and preparedness exercises and also for developing templates that can be used following a significant regional earthquake. With the improvements to ShakeMap and the improved dialogue with the emergency managers, a suite of maps and information products were developed based on scenario earthquakes for training and exercise purposes. These products will be available in a timely fashion following a significant earthquake in the Utah region.
S51A-0224
ShakeMap-Based Earthquake Emergency Response for Lifelines
Pacific Gas and Electric (PG&E), Bay Area Rapid Transit (BART), Bonneville Power Administration (BPA), and a number of other lifeline operators and utilities are using Geographic Information System (GIS)-based products, including ShakeMap, to enhance their earthquake emergency response capabilities. PG&E uses ShakeMap in conjunction with improved digital hazard maps for the San Francisco Bay area as a decision support tool to prioritize response activities. Used as a screening tool, this information helps to rapidly identify potential gas and electric transmission problem areas, prior to the receipt of damage reports from the field. Earthquake Risk Values, which combine ground shaking estimates with surface fault rupture, ground failure, and pipeline performance factors, are used to identify potentially vulnerable transmission pipeline segments and gas handling facilities for inspection following an earthquake. Scenario ShakeMaps are used for emergency training exercises and, in the event a large earthquake causes the Internet to be temporarily out of service, can be used for initial damage assessments until actual ShakeMaps become available. BART integrates the ground motions developed in near real time using ShakeMap with a BART vulnerability model in a System Earthquake Risk Assessment (SERA) to establish the likely post earthquake damage to the BART system. After most M3 or larger earthquakes, the SERA model for BART is run, using the ShakeMap- developed ground motions, to assess the likely and near-lower bound performance for more than 15,000 individual structures and components in the BART system. If any of these structures or components are predicted to have greater than a 10% chance of material nonlinear performance that might invoke a life safety situation, BART sends out people to visually inspect the facilities so targeted by the analysis. This software has been used to analyze the BART system for 15 earthquakes (M3.5 to M4.4) using ShakeMaps since 2003; only the July 20, 2007 earthquake along the Hayward fault triggered the need for visual inspections. BPA uses SERA to rapidly evaluate the potential for damage at its high voltage substations throughout the Pacific Northwest. The BPA model includes many thousands of pieces of equipment at BPA's 115 kV, 230 kV and 500 kV substations. As BPA's service area is so large, the ShakeMap analysis helps identify where the likely damage might have occurred within BPA's system, and helps establish the likely extent of substation damage and thus the magnitude of the needed emergency response. This software has been used to analyze the BPA system for 8 earthquakes (M2.7 to M4.0) using ShakeMaps since 2004; none has yet triggered the need to perform post- earthquake visual inspections.
S51A-0225
LiqueMap: A Real-Time Postearthquake map of Liquefaction Probability
LiqueMap is a proposed map for real-time distribution over the internet after a large earthquake. It predicts the spatial distribution of the probability of liquefaction in the area subjected to strong ground motion. It relies on peak ground accelerations (PGA) produced by ShakeMap and the methodology for probabilistic liquefaction hazard mapping that was developed by Holzer and others (USGS OFR 02-296, 2006). This methodology relies on field- based plots of cumulative frequency distributions of the liquefaction potential index (LPI) of surficial geologic units. LPI is a scalar parameter that integrates the liquefaction potential of the whole soil column as a function of PGA and earthquake magnitude. LiqueMap is produced with ArcGIS\copyright Model Builder. Three requirements must be met to produce a LiqueMap. First, digitized maps of surficial geology must be accessible with sufficient resolution to distinguish between geologic units with different liquefaction susceptibility. Second, adequate subsurface information or knowledge must be on hand for each surficial geologic unit in the ShakeMap area to produce plots of cumulative frequency of LPI for the range of earthquake magnitudes and PGA that are expected. These plots of cumulative frequency are used to establish liquefaction probability distribution functions for each surficial geologic unit. If LPI data for local geologic units are not available, generic probability distribution functions based on the type of geologic unit can be substituted. The third requirement is the spatial distribution of PGA. PGA values from ShakeMap are gridded and are used as input to the liquefaction probability distribution function at each node to the compute the spatial pattern of liquefaction probability. Because position of the water table is a critical parameter for evaluating liquefaction potential, liquefaction probability distribution functions are computed for either idealized or actual conditions depending on the hydrologic information that is available. LiqueMap should be useful to utility and transportation agencies for identifying areas of potential damage to lifelines from liquefaction-induced permanent ground deformation. When overlain on maps of vulnerable pipelines and other lifelines, LiqueMap can help these entities set priorities for post-earthquake inspections.
S51A-0226
Progress Toward More Detailed Site-Conditions Maps for California
We have developed a map of geologic units that can be distinguished by their shear-wave velocity. In developing this map we build upon earlier work to determine the shear-wave velocity characteristics of geologic units in California (Wills and Silva, 1998) and grouping geologic units with similar Vs into NEHRP categories (Wills and others, 2000). We have further refined those categories and prepared a map showing geologic units with distinct Vs characteristics (Wills and Clahan, 2006). We are testing potential refinements of the existing statewide map based on: 1. Improving the precision of locations the geologically-defined units and 2. Improving the definition of shear-wave velocity classes. For a test area in southern California, we have compiled the available high resolution geologic mapping. We have simplified the hundreds of individual geologic units shown on those maps to simplified units based on shear- wave velocity defined units of Wills and Clahan (2006). The result is a map showing geologic units with defined Vs30 values based on the most detailed available geologic mapping. Most of the simplified geologic units have well-defined ranges of shear-wave velocity, with the very important exception of younger alluvial deposits. These deposits may be thin – and profiles to 30 m include other, higher velocity materials, and the grain size and density may vary substantially. As a result there is a wide range of measured Vs30 values in younger alluvium. To sub-divide the young alluvium Wills and Clahan (2006) attempted to use very simple geographic criteria. We developed classes called thin alluvium, deep alluvium, fine alluvium and coarse alluvium based on geographic criteria. These categories appear to have differing shear-wave velocity characteristics, but the criteria for distinguishing them was not explicitly defined, nor were options for defining these classes explored. In order to make the mapping more consistent, and more applicable to other areas, we are exploring the use of two simplified rules for sub-dividing young alluvium. Preliminary results suggest that distance from bedrock and slope may both correlate with shear-wave velocity in young alluvium. Further work will define the relations between these factors and shear-wave velocity, and result in methods for producing improved maps of Vs30 for estimating seismic amplification in southern California and elsewhere.
S51A-0227
Development of Vs30 map in Taiwan using multivariate geostatistical method
Average shear-wave velocity of the upper 30 meters soil profile (Vs30) is a key to indicate the site response characteristics and amplification of ground-motion. The purpose of this study was mainly on mapping the distribution of Vs30 in Taiwan with available data and multivariate geostatistical techniques. 320 drill and PS logging data at existing strong-motion stations were used to establish the spatial relationship between shear- wave velocity and SPT-N. And then, this model was used to estimate the shear-wave velocity of each drilling section from GEO2005 drill database of the Central Geological Survey. Finally, a geostatistical method called ¡§Kriging with varying local means¡¨ was applied to generate a Vs30 distribution map of Taiwan. A site classification result of Taiwan's strong-motion stations is also updated based on this new Vs30 map.
S51A-0228
Vs30 from site response and local magnitude corrections
We used site response amplifications and local magnitude (ML) station corrections to derive Vs30 estimates (S- wave velocity in the upper 30 m) in the U. S. Pacific Northwest. To obtain Vs30 from site response amplifications, we used the average of the amplifications at 1, 2, 3 and 5 Hz to estimate an impedance ratio, which we assumed represents the ratio between bedrock (Vs=760 m/s) and the shallow, 30-m thick layer (Vs30). We used the velocity and density relations in Brocher (2005, BSSA) to convert this impedance ratio into a velocity and density for each layer. Comparing these Vs30 estimates with nearby (<300 m distance) velocity measurements generally shows agreement within about 30%. To estimate Vs30 from ML station corrections, we show that the station correction is proportional to the logarithm of the site amplification. The magnitude corrections are relative to the average of the network, so they first needed to be adjusted so that a bedrock reference site (Vs=760 m/s) has a correction of 0. We used the adjusted magnitude corrections to compute the site amplification that would be required to cause the ML correction, and from this the impedance ratio and Vs30 as before. The Vs30 estimates made from the magnitude corrections again show agreement to within 30% of nearby measured Vs30 values. The resulting Vs30 values correlate with the age of surface geologic units, showing a progressive increase in Vs30 from ~280 m/s for Quaternary alluvial deposits and artificial fill, to ~1300 m/s for Oligocene volcanic rocks. The Pleistocene glacial tills that cover much of the Puget Lowland region show an average velocity of 512 m/s.
S51A-0229
Adapting ShakeMap to Europe: Ground-motion relations and soil response
One of the objectives in the European Community NERIES (Network of Research Infrastructures for European Seismology) project has been to develop means and facilities for shake map applications and associated damage and loss estimation. To this end the project has resolved that one will base this work on the USGS ShakeMap solutions. While operational installations in Europe are limited to centers and agencies that cover the most active seismic regions (such as Turkey and Italy), NORSAR, being located in the low-seismicity region, has installed ShakeMap only for testing and development purposes. These developments have been directed into two areas, the first of which being ground-motion (attenuation) relations where a review of existing European and more regional and local relations has been conducted. For the Mediterranean region there are new and reasonable stable spectral and PGV relations available, however, for the stable continental parts of Europe the coverage is a lot weaker. Also, intensity attenuation is less well studied, showing a much greater scatter, so more work is needed also there. The ground-motion relations depend, however, on good site corrections, and to that end we have also analyzed the use of topographic slope in the absence of direct measurements of Vs30 (Wald & Allen, BSSA, 2007). In general this approach works also for Europe, but some important exceptions have told us that a proper calibration is important also here. Finally we have also looked into using other terrain parameters to assist in the assessment of the relation between topography and soil properties. http://neries.knmi.nl/
S51A-0230
New Ground Motion Prediction Equations Spanning Weak and Strong Ground Motion Levels
We present new set of prediction equations for horizontal peak ground acceleration (PGA) and peak ground velocity (PGV) valid up to 200 km away from earthquakes over the magnitude range 2 &<& M &<& 8. We adopt a functional form used by Cua and Heaton (2007) that allows for a linear dependence of the log of ground motion amplitudes on magnitude for events with M&<&5, and a magnitude-dependent saturation term for M&>&5. We fit this functional form to observed PGA and PGV from 1) a predominantly weak motion southern California dataset, and 2) the Next Generation Attenuation (NGA) strong motion dataset in a single inversion. At the weak motion levels, the new equations are consistent with the small-amplitude equations (Quitoriano, 2003) used by the ShakeMap system for M&<&5. The resulting equations predict median PGA and PGV values consistent with the Boore and Atkinson (2006) and Campbell and Borzorgnia (2006) NGA equations, which are valid for 5&<&M&<&8, at the larger magnitudes. There are discrepancies between the median ground motions predicted by the NGA relationships and our equations at M=5 level, which is the lower bound of the applicable magnitude range for the NGA equations. These discrepancies may be indicative of a systematic bias towards higher ground motion levels at the lower magnitude bound of the NGA equations. Our findings suggest that in order to develop prediction equations that adequately characterize the median ground motion behavior at a particular magnitude (for instance, the lower magnitude bound of M=5 for the NGA equations), a sufficient quantity of data from events with magnitudes lower than the targeted lower magnitude bound must be used to constrain the regressions.
S51A-0231
Attenuation of Macroseismic Intensity - New Relations for Different Parts of Europe
When estimating strong ground motion, either using ShakeMap or other programs for seismic hazard assessment or early warning purposes, the attenuation of ground shaking is a key parameter to obtain reliable ground motion estimates. Seismic hazard studies, as well as early warning systems, are usually focused on estimating ground shaking levels in terms of peak ground acceleration (PGA), peak ground velocity (PGV) or other recorded parameters. One major drawback of such studies is the very limited strong motion dataset available in many regions even of high seismicity. Furthermore, there is hardly any direct correlation between the distribution of e.g. PGA and damage. To overcome such limitations we study strong ground motion in terms of macroseismic intensity. This makes it possible to include also historical earthquakes in an analysis by using comprehensive intensity point datasets and has the advantage of the results being directly related to the observed earthquake damages. As part of the EC project SAFER (Seismic eArly warning For EuRope), we study the attenuation of macroseismic intensities for use within early warning systems. Results will be presented for the Marmara Sea area (Turkey), the Naples area (Italy) and the Vrancea area (Romania). We consider a physically constrained attenuation model and account for the finite fault dimensions of large earthquakes in the regressions. Furthermore we derive relations between recorded ground motion parameters and macroseismic intensity for use e.g. within ShakeMap. Data from several earthquakes are joined, and for the case of Romania, anisotropy in the macroseismic field is accounted for in the derived attenuation model. Examples will be shown from implementation of the derived attenuation relation for Romania with the ShakeMap software for estimation of the intensity distributions due to previous earthquakes. Results indicate that our regression model provides a reliable estimate of macroseismic intensities for the studied regions, which can be implemented with e.g. ShakeMap for rapid estimation of the ground motion distribution after a large earthquake.
S51A-0232
Introducing ShakeMap to potential users in Puerto Rico using scenarios of damaging historical and probable earthquakes
The island of Puerto Rico has a long history of damaging earthquakes. Major earthquakes from off-shore sources have affected Puerto Rico in 1520, 1615, 1670, 1751, 1787, 1867, and 1918 (Mueller et al, 2003; PRSN Catalogue). Recent trenching has also yielded evidence of possible M7.0 events inland (Prentice, 2000). The high seismic hazard, large population, high tsunami potential and relatively poor construction practice can result in a potentially devastating combination. Efficient emergency response in event of a large earthquake will be crucial to minimizing the loss of life and disruption of lifeline systems in Puerto Rico. The ShakeMap system (Wald et al, 2004) developed by the USGS to rapidly display and disseminate information about the geographical distribution of ground shaking (and hence potential damage) following a large earthquake has proven to be a vital tool for post earthquake emergency response efforts, and is being adopted/emulated in various seismically active regions worldwide. Implementing a robust ShakeMap system is among the top priorities of the Puerto Rico Seismic Network. However, the ultimate effectiveness of ShakeMap in post- earthquake response depends not only on its rapid availability, but also on the effective use of the information it provides. We developed ShakeMap scenarios of a suite of damaging historical and probable earthquakes that severely impact San Juan, Ponce, and Mayagüez, the 3 largest cities in Puerto Rico. Earthquake source parameters were obtained from McCann and Mercado (1998); and Huérfano (2004). For historical earthquakes that generated tsunamis, tsunami inundation maps were generated using the TIME method (Shuto, 1991). The ShakeMap ground shaking maps were presented to local and regional governmental and emergency response agencies at the 2007 Annual conference of the Puerto Rico Emergency Management and Disaster Administration in San Juan, PR, and at numerous other emergency management talks and training sessions. Economic losses are estimated using the ShakeMap scenario ground motions (Saffar, 2007). The calibration tasks necessary in generating these scenarios (developing Vs30 maps, attenuation relationships) complement the on-going efforts of the Puerto Rico Seismic Network to generate ShakeMaps in real-time.
S51A-0233
Using ShakeMap to Map MMI Intensity for the 1868 Hayward, the 1898 Mendocino, and the 1906 San Francisco Earthquakes
The utility of the ShakeMap format for depicting ground motion and shaking intensity in recent well-recorded earthquakes has led to its use in mapping MMI intensity for large historic earthquakes. But while the mechanics of incorporating intensity sites as ground motion data in ShakeMap are trivial, the constraints of ShakeMap's native MMI scale and the hazards of ShakeMap's interpolation scheme can make implementation for historic earthquakes difficult. We illustrate these difficulties using the MMI ShakeMaps for the 1868 Hayward, the 1898 Mendocino, and the 1906 San Francisco earthquakes as case studies. To compare historic earthquakes with modern earthquakes, it is critical for ShakeMap to be fixed to a single intensity scale. Wald et al. (1997) base ShakeMap intensities on Stover and Coffman's (1993) revision of the MMI scale. Thus, intensities determined by Lawson (1908) and Toppozada et al. (1981) for these historic earthquakes must be re-evaluated before they can be incorporated in ShakeMap. To interpolate intensities in areas without stations, ShakeMap interposes "phantom" stations where the ground motions are estimated from Boore et al.'s (1997) attenuation relations. This hybrid interpolation leads to two problems. First, the ground motions are only regressed for a restricted distance from the fault (rJB ≤ 70 km, where rJB is the Joyner-Boore distance). To extend these relations to rJB ≤ 300 km, we modify these ground motion predictions using the exponential falloff exp(-0.0035 rJB). Surprisingly, this term fits the variation of intensity with distance for all three earthquakes. Second, the source strength can vary significantly along the fault for large earthquakes: interposing an "average" value often produces an artificial variation of the intensity. We damp these variations by minimizing the number of phantom stations, increasing both the threshold distance for the phantom stations and the total number of intensity sites in these areas by using all the available newspaper reports and historical narratives that describe shaking effects. We have added approximately 160 sites for the 1906 earthquake, and 30 sites for the 1868 earthquake. Locating these sites in backwoods areas, 10 to 70 km from the Mendocino coast, is particularly critical for the ShakeMaps of the 1898 and 1906 earthquakes.
S51A-0234
Estimation of Seismic Hazard Potential in Taiwan Based on Earthquakes with Mw >5 in 1900-2006
Taiwan is located in the circum-Pacific seismic belt with high seismicity. In last century many damaging earthquakes have taken place in Taiwan. Reliable assessment of seismic hazards is fundamental for effective earthquake disaster mitigation. Reliable seismic hazard assessment in turn requires accurate ground motion estimates. A catalog of more than 1840 shallow earthquakes with homogenized Mw magnitude ranging from 5.0 to 8.2 in 1900-2006 (Chen and Tsai, 2008) is used to estimate the seismic hazard potential in Taiwan in the form of ShakeMap (Ward et al., 1999). First, we use the empirical attenuation relationships obtained for Taiwan by Liu and Tsai (2005) to calculate the peak ground acceleration (PGA) and peak ground velocity (PGV) for earthquakes in above-mentioned catalog at each grid point, with exception of the Mw 7.7 Chi-Chi earthquake for which actual observations are used. Secondly, the site response factor is incorporated in the present PGA and PGV prediction models at all grid points. Such prediction models will result in more realistic peak ground motion estimates for assessment of seismic hazard potential. Finally, we combine above results with the observed PGA and PGV contour maps of Chi-Chi earthquake. As a result, the ShakeMap patterns show that rupturing of the Chelungpu fault definitively affected the spatial distribution of ground motions. High MMI intensity greater than IX located in the hanging wall of the northern part of the Chelungpu fault, is related to high PGV (about 300 cm/sec) recorded at TCU068 (Shihkang station). In addition, high MMI intensity greater than VIII in Nanao area in northeastern Taiwan, Taichung and Chianan areas in western Taiwan, Hualien and Taitung areas in eastern Taiwan are all related to large damaging earthquakes occurred in these areas. Nevertheless, it should be noted that relatively low seismic hazard potential in Changhua area with MMI intensity lower than VI may be misleading, considering occurrence of damaging historical earthquakes before 1900 and presence of Changhua fault in the area.