SMIP 2008 Seminar

The Working Group on California Earthquake Probabilities (WGCEP) has developed the Uniform California Earthquake Rupture Forecast version 2 (UCERF 2). This model was developed by the USGS, CGS, and SCEC, with significant support from the California Earthquake Authority. The time-independent model was developed jointly with the USGS National Seismic Hazard Mapping Program (NSHMP). As with past WGCEP and NSHMP efforts, the model depends on accurate fault locations and slip rates. This study updated fault locations based on the SCEC Community Fault Model and slip rates based on recent studies. The overall fault slip across the region was constrained by GPS deformation rates and long term plate rates. The resulting model is consistent with moment rates from GPS, long term plate tectonic models, and historic seismicity. A careful analysis of historical seismicity rates revealed an over-prediction of the rate of M>6.5 earthquakes in previous models, which we reduced to a 30% over prediction in UCERF 2 (within observed 95% confidence bounds).

Our study differs from previous WGCEP efforts by: 1) reporting earthquake probability for the entire state of California; 2) using uniform methodology across all regions; 3) compiling and using updated, uniform, and publicly accessible statewide data; 4) developing new methods to make models more rigorously adherent to observational data; 5) implementing in a modular (object-oriented), extensible framework, so that alternative logic-tree branches can easily be investigated and future updates can be quickly accommodated as new data and methods emerge. Advice and comment from the scientific and engineering communities was sought regularly through open meetings and workshops. Time- dependent probabilities were applied to fault-based sources using an empirical model (where long-term earthquake rates were adjusted according to any recent changes in observed seismicity) and using an elastic-rebound-motivated renewal model on major faults where the date of previous event was known.

Although UCERF2 is more complete and more consistent with geologic and seismic data than previous models, there are aspects of the model that we believe need improvement in the future. These include: assumptions regarding fault segmentation, the lack of fault-to-fault ruptures, and the lack of earthquake triggering and temporal clustering effects.

The UCERF2 served as the input fault model for calculation of the California portion of the National Seismic Hazard Map. The 2008 hazard maps are significantly different from the 2002 maps in many parts of the United States. The new maps for the Western United States indicate about 10-percent reductions for 0.2-s spectral acceleration and peak horizontal ground acceleration and up to 30-percent reductions in 1.0-s spectral acceleration at similar hazard levels. Most of the changes in the new maps can be attributed to the introduction of new attenuation relations for crustal and subduction earthquakes; however, changes to the fault and seismicity parameters also can be significant. In California, the ground motion calculated for the same hazard level increased up to 10 percent near major faults and up to over 15% in the Santa Barbara area and above the Cascadia Subduction Zone for 0.2-s spectral acceleration and peak horizontal ground acceleration. For 1.0-s spectral acceleration ground motions calculated for the same hazard level are lower everywhere except above the Cascadia Subduction Zone, but decrease less in the Santa Barbara area and the Mojave Desert, where the fault model was modified to reflect increased slip rates or new fault models.

Near-Fault Instrumentation

In the early 1980s a strong motion array was deployed near Parkfield in central California because of the expectation of an earthquake in the area. Important data had been obtained during the 1966 Parkfield, and given the prediction of a repeat earthquake in the 1980s, special arrays were deployed by both CGS and USGS. The CSMIP array included 45 strong motion stations, a significant investment at the time. The Parkfield earthquake eventually did occur on September 28, 2004 (magnitude 6.0 Mw). The resource investment was well warranted as very important data was recovered including some of the strongest accelerations ever recorded from a moderate earthquake. Beyond that, because of the design and scale of the array, one of the most dense set of near-fault recordings was obtained. These recordings showed near-fault motion that was highly variable, spatially, with accelerations well over 1 g within 1 or 2 km of accelerations near 0.1g – a factor of 10. The earthquake showed that near fault motion could be quite variable, and that more observations would be important to understand this result and its generality.

Currently, the two areas with the highest likelihood of a moderate or larger earthquake in California, according to the Working Group on California Earthquake Probabilities (WGCEP), are the Hayward fault in the San Francisco Bay area and the Coachella segment of the San Andreas fault in southern California. Considering the Hayward area, given the experience of the Parkfield ground motion, it was clear that if the ground motion were to be as variable as in the Parkfield event, the station density was far too limited to capture the variability. The CSMIP Strong Motion Instrumentation Advisory Committee (SMIAC) recommended that the Hayward area instrumentation be substantially increased. This represents a departure from the urban ground motion instrumentation focus of recent years, aimed at improving ShakeMap for response, to a focus of improving the learning from an event.

Significant progress has been made in the past two years on the Hayward area, partly because the USGS has also been working to expand the instrumentation. With the completion of stations planned by CSMIP, and those being considered by the USGS, station density should become comparable to Parkfield. Special purpose arrays are also important. For example, the USGS deployed a special subarray of 14 stations with spacing of a km or less, called UPSAR. Important questions, such as tracking the rupture process, can be addressed with such arrays. Although not directly equivalent, some special USGS arrays in the San Jose area will provide important close-spacing ground motion data.

In contrast with the Hayward East Bay area, the Coachella segment of the Southern San Andreas has relatively few stations. The expected event is larger, near a magnitude 7 according to the Working Group. The number of records from the close-in region of M>7 earthquakes is very limited. For input ground motion in designing for large earthquakes, artificially generated strong motion records are sometimes needed. To address this data paucity, a relatively large number of instruments are needed within the near fault zone of large events. Since the focus of these instruments, especially those located in lightly populated areas, is not for immediate usage in ShakeMap, but to obtain data to guide future design assumptions, some of the features necessary for ShakeMap-caliber instruments can be relaxed and simpler, more economical instruments can be deployed. For example, these instruments do not need communication capability and certain other features that drive up cost and power usage. Over the next two years, a significant number, perhaps over 100, of these simple low-cost instruments are planned to be deployed in the near-fault zone of the expected Coachella earthquake. They will of course be complemented by conventional seismic instruments of the USGS and CGS arrays.

Relative Displacement Accuracy

The accuracy of displacement computed from strong-motion records is important in accessing structural response. The inter-story drift, or relative horizontal displacement of adjacent floors, is a major factor in the seismic response of a building. High inter-story drifts are likely associated with incipient damage in the structure.

For early strong motion accelerographs, displacements could only be obtained after analog film records were laboriously digitized and processed. Because of the high noise intrinsic to this procedure, the displacements obtained from doubly integrating the digitized acceleration generally had high noise. Thus, relative displacements, obtained after differencing records from nearby sensors, often had high noise, especially at long period.

Error in computed displacements increases with period. At short period (e.g., 1 second or less), the error amplitude is small (a fraction of a cm). At periods of 5 to 10 seconds, the error in displacement can be significantly greater (several cm). In comparison with early instruments, modern 18-bit digital accelerographs have very low noise. Because of this, serious consideration can be given to utilizing the inter-story drift obtained through differencing accelerations obtained at nearby floors. Some tests were recently conducted as payload instrumentation on a NEES 3-story test structure at the UCSD shake table. Accelerometers were attached to the structure as well as on a nearby stationary tower, and a relative displacement sensor recorded the motion between them. The tests indicate that except for permanent displacements, the inter-story displacements obtained from a nearby pair of modern, low-noise accelerometers is quite accurate even at periods of several seconds.

This paper describes an investigation of the ground motions recorded at the Turkey Flat test site, and of the predictions of those motions in the blind prediction symposium that took place in 2006. The subject investigation is currently in its early stages, so the current paper focuses on the site, measured subsurface conditions, and some of the characteristics of the recorded motions. A brief summary of the results of the ground motion predictions is also provided.

Soil-structure interaction can affect the response of buildings with subterranean levels by modifying the characteristics of input motions relative to those in the free-field and through the added system compliance associated with relative foundation/free-field translation and rocking. While procedures are available to account for these effects, they are seldom utilized in engineering practice. Our objective is to examine the importance of these effects on the seismic response of a 54 story building with four subterranean levels. We first generate a “most accurate” (MA) model that accounts for kinematic interaction effects on input motions, depth-variable ground motions along basement walls, compliant structural foundation elements, and soil flexibility and damping associated with translational and rocking foundation deformation modes. With reasonable tuning of superstructure damping, the MA model accurately reproduces the observed response to the 1994 Northridge earthquake. We then remove selected components of the MA model one-by-one to test their impact on building response. Factors found to generally have a modest effect on building response above ground level include compliance of structural foundation elements, kinematic interaction effects (on translation or rocking), and depth-variable ground motions applied to the ends of horizontal soil springs/dashpots. Properly accounting for foundation/soil deformations does not significantly affect vibration periods for this tall building (which is expected), but does impact significantly the distribution of inter-story drifts over the height of the structure. Two approximations commonly used in practice are shown to provide poor results: (1) fixing the structure at ground line with input consisting of free-field translation and (2) modeling subterranean soil layers using a series of horizontal springs which are fixed at their far ends and subjected to free- field ground accelerations.

This presentation describes some significant milestones in the development of strong motion monitoring of earthquakes as judged by the author. Strong motion earthquake monitoring was motivated by the Great Tokyo earthquake of 1923 and was strongly influenced by Prof. Romeo Martel in the US and Prof. Kyoji Suyehiro in Japan. Also greatly influential in the development of a strong motion instrument was John R. Freeman who became interested in earthquakes at age 70. The first strong motion instrument was constructed by the US Coast and Geodetic Survey in 1932 and the first significant strong motion record was obtained during the Long Beach earthquake of 1933. This presentation traces the development of strong motion instruments and the analysis of strong motion data from the era of the Wood-Anderson Seismograph to more recent digital recorders.

Early strong motion instruments were analog and data was recorded on photographic film. There were many challenges in getting the recorded data into a form that was useful to engineers and others studying strong earthquake motions. Initially, photographic records were examined visually to determine notable features of the motion including peak acceleration, duration of shaking, and the nature of the envelope of the time history of motion. But in 1934 Prof. Hugo Benioff of Caltech introduced the Response Spectrum of an earthquake. This concept was later refined for engineering use by his colleague Prof. George Housner. The Response Spectrum provided earthquake engineers with an easily applied tool that could be used to estimate the response of a structure to earthquake excitation. Computation of Response Spectra from early film records was no easy task and relied heavily on the use of analog computers. However, in spite of these difficulties, sufficient data were analyzed so that the first “Design” Response Spectrum was published by Prof. Housner in 1959. Later, Prof. Newmark and Prof. Hall of the University of Illinois produced a further refined Design Response Spectrum that was widely distributed in a 1982 monograph by EERI.

Due to a landmark program instituted by the City of Los Angeles which mandated the installation of strong motion instruments, a treasure trove of approximately 400 strong motion records was obtained during the San Fernando earthquake of 1971. It was also significant that this earthquake occurred at the time when analog computation was giving way to digital computation in many fields of engineering. Capitalizing on the convergence of these two events, the NSF funded a project at Caltech to digitize and distribute the time history and Response Spectra data for all of the San Fernando records as well as other key historical records. The process of digitization revealed certain base-line problems with the data and band-pass filtering algorithms were developed to eliminate drift in the integrated acceleration data. New digital programs were also developed to compute Response Spectra.

In 1976, the Great Tangshan earthquake occurred in China killing hundreds of thousands of people. The following year, at the 6WCEE in New Delhi, India, a new international committee was formed on strong motion instrumentation. At that time, there were about 5,000 strong motion instruments deployed worldwide, 3,000 of which were in the US. In 1978 an International Workshop on Strong-Motion Instrument Arrays was held in Hawaii. The participants of that workshop concluded that understanding strong ground motion was critical to earthquake safety, that there was a scarcity of engineering data near the source of destructive earthquakes, and that countries needed to make a concerted effort to deploy instrument arrays capable of resolving the nature of the source mechanism, wave propagation, and local site effects associated with earthquakes. As a result of this workshop, a number of digital strong motion arrays were deployed worldwide including in Taiwan (SMART-1) and China.

The digital strong motion array deployed in China was in the aftershock region of the Great Tangshan earthquake. This array recorded more than 1050 near-field accelerograms from more than 400 events. On October 19, 1982, nine digital instruments recorded the ML=5.7 Lulong event with the closest instrument being only 5 km from the epicenter. After overcoming some processing challenges, this record showed an interesting new type of “pulse-like” ground motion that had not been previously reported. After some initial dispute over the validity of this record, it was gradually accepted as indication a real phenomenon. This result was further validated by the 1992 ML=7.5 Landers earthquake. An analog instrument installed by the Southern California Edison Company was located within 2 km of the fault trace of that event. The instrument was retrieved from the field and subjected to extensive testing at Caltech. The integration algorithms developed for the Lulong record were then applied to the Landers record. What was revealed was a clear indication of the pulse-like near-field ground motions. The same techniques were applied to recorded data from the 1992 ML=6.7 Erzincan earthquake and the results were very similar. By this time, there was no disputing the existence of near-field pulse-like ground motions.

The 1995 Hyogoken-Nambu (Kobe) earthquake in Japan triggered a significant expansion of strong motion networks in Japan as well as in other Asian countries. The 1999 Chi-Chi earthquake in Taiwan yielded accelerograms from more than 600 instruments. The State of California presently has over 2,000 strong motion instruments deployed on the ground and in buildings.

As strong motion networks have gone from analog instruments to digital instruments, data retrieval and processing has also changed. It is now possible to retrieve and process data in near or true real-time. This opens up many exciting opportunities for enhanced decision making using string motion data. The presentation gives an example of a currently operating real-time monitoring system in the Millikan Library Building on the Caltech campus. This system is capable of providing real-time inter-story hysteresis diagrams to assist in damage assessment. Other possible applications of decision making based on real-time string motion monitoring are also given. These applications address the needs of a broad spectrum of stakeholders throughout society including public officials, building owners, building occupants, and individual citizens.

Strong motion monitoring has reached a level of maturity where we no longer celebrate the good fortune of obtaining one additional ground motion record from a distant earthquake. Therefore, it is the position of the author that our efforts need to be refocused from the capture of isolated records to obtaining integrated real-time information that can be used for rapid decision making.

Healthy communities continuously grow by leveraging their intellectual capital to drive economic development while protecting their cultural heritage. Success, in part, depends on the support of a healthy built environment that is rooted in contemporary urban planning, sustainability and disaster resilience. In many parts of the country, the ability to rebound from major earthquakes is an important facet of community health, one that depends on the expertise of the nation’s earthquake professionals. We, as earthquake professionals, have the responsibility to deliver that expertise in an understandable fashion that can be interwoven into public policy while recognizing the community’s natural ability to rebound. No one else has the technical knowledge to bring that perspective to the policy table.

Earthquake professionals -- Emergency Response Planners, Earth Scientists, and Earthquake Engineers -- have made great strides toward understanding how to record, characterize, build for, and recover from major earthquake events. Today’s seismic hazard maps, performance based building codes, and integrated emergency response plans all demonstrate remarkable progress in just the past 30 years. Seismic hazards nationwide are understood and procedures are available to adequately predict performance. EERI’s Securing Society Against Catastrophic Earthquake Losses defined a research and outreach plan in 2003 that would arrest the growth of seismic risk nationwide to acceptable levels. In 2006, EERI, SSA, and California OES co-convened a conference commemorating the 1906 San Francisco Earthquake and published Managing Risk in Earthquake Country; an action plan for reducing losses, to acceptable levels in future earthquakes. Unfortunately, progress on implementing these plans appears to be stalled due to a lack of funding and political will caused by complacency, misunderstanding, and an absence of persistent lobbying by the earthquake experts.

Planners and policy makers are deeply concerned with all aspects of their communities, including its seismic safety. Their reluctance to implement the latest plans for achieving seismic safety is rooted in a misunderstanding of the hazard they face and the risk it poses to their built environment. Probabilistic lingo and public debate about how big the “big one” drives them to resort to their own experience and intuition. “It’s never happened here before” is a common justification for setting aside policy changes that will improve safety and resilience. The usual misconception of how much damage the built environment will experience is based on the belief that the building official and their latest building codes assure protection in damage proof buildings. There is a fundamental lack of transparency related to what is expected to happen and it is partially blocking the policy changes that are needed.

The solution: craft the message in broad based, usable terms that name the hazard, defines performance, and establishes a set of performance goals that represent the resiliency needed to drive a community’s natural ability to rebound from a major seismic event. With the assistance of the local earthquake professional community, urban planners, policy makers, and local City officials, the San Francisco Planning and Urban Research Association (SPUR) has established three study groups to sort out the issues. We are in the process of determining options and developing policy recommendations to assure that San Francisco and the Bay Area will not fall to the dilemmas that are preventing the restoration of much of the Gulf Coast after Hurricane Katrina. SPUR, in its usual role as an advocate for thoughtful Urban Planning, choose to take a different tack than has been used in the past. We are using transparent goals and measures with an intuitive vocabulary for both performance and hazard, and the recommendations describe a state of resiliency that is needed to support response and recovery.

SPUR is in the process of defining performance goals, for the built environment in terms of three time frames. The first relates to the initial response and lasts seven days. The second extends to 30 days and focuses the restoration of workforce housing and meeting ongoing social needs. The third is a three-year period of long-term reconstruction. During the initial period, essential facilities such as hospitals, police stations, and emergency response facilities are needed, along with housing that can support shelter-in- place, and the infrastructure systems needed to support reconstruction. The focus of the next period is on restoring the living environment for the workforce that will reconstruct the city, by reestablishing their utilities, schools, and neighborhood businesses. The third phase expedites the achievement of a “new normal”.

SPUR is in the process of defining the hazard in terms that are consistent with current San Francisco programs and policies. Three earthquakes are named and defined for use in the recommendations. The “routine” earthquake is a 70% in 50-year event and used to define the service levels of tall buildings. The “expected” earthquake is a 10% in 50-year event and is used as the basis for the policies related to performance. The third is the “extreme” earthquake that is a 2% in 50-year event, the basis of the 2006 International Building Code.

SPUR is defining five performance measures for buildings and three for lifeline systems in an effort to establish an intuitive understanding of the expected post-event performance. Each declares whether people will be safe inside, whether the building will be able to be repaired and whether usable during repairs. Lifeline systems are further defined in terms of the time intervals to restore 90%, 95%, and full service. These transparent categories are used in conjunction with the expected earthquake level to describe the standards needed for new buildings and lifelines and the rehabilitation programs needed for existing buildings and lifelines so that the performance goals are achievable, the cultural assets protected, and the economy able to rebound. Because the definitions apply to individual types of uses of buildings and allow various time frames for restoration, the needed programs should prove to be achievable and cost effective.

Many of us strive to contribute to the greater good while doing our everyday jobs. It is a passion for me and has lead to my personal devotion to seismic risk reduction advocacy nationwide. As earthquake professionals, we are very lucky to be able to contribute an expertise that can save lives as well as communities. In 30 plus years, I’ve learned that I can be effective when working with other structural engineers on buildings codes, pace setting when working with the larger family of earthquake professionals, and actually able to change public policy when providing my technical expertise to the broader community of policy makers while helping them craft the policies needed to instill change. It takes patience and a broad understanding of all the issues being faced. It’s not unlike my trade, fitting a structure into a building. Here it is fitting seismic into my community and the results are worth the effort and frustration. I challenge each of you to do the same. Volunteer and work toward making your community healthier.

The identification of modal parameters has been performed for the New Carquinez suspension bridge in California. By using multiple ambient vibration data sets recorded through a wind-motion monitoring system in the bridge, the baseline modal parameters were obtained in order to investigate dynamic behavior of the bridge in operating conditions. For the modal parameters identification, the data-driven stochastic subspace identification technique was implemented. For each data set, the modal parameters for structural modes were estimated by examining the estimation error between measured data and reconstructed one from the identified modes. Based on the results, variability of the identified modal parameters was also investigated.

The Tall Buildings Initiative (TBI) is a major multidisciplinary program coordinated by the Pacific Earthquake Engineering Research Center (PEER) in collaboration with numerous local, state and federal organizations. The goal of TBI is to address critical technical issues on seismic analysis and design of new tall buildings located in coastal California. This paper provides an overview of the TBI tasks on ground motion issues.

At 14:28 on May 12, 2008 (Beijing and Sichuan Time), a great earthquake occurred in Sichuan province, China, with a surface wave magnitude of 8.0. The earthquake epicenter was located in Wenchuan County at latitude 31.021?N and longitude 103.367?E, and a focal depth of 14 km. The associated fault rupture is mainly thrust with a strike-slip component on the Longmenshan fault belt. However, the source mechanism is quite complex. The rupture process is mainly unilateral with the main rupture spreading to the northeast about 300 km from the epicenter The post-earthquake geological studies found a 200 km long surface rupture along the central branch of the Longmenshan fault belt (i.e Yingxiu- Beichuan fault) and a rupture of 60 km along the front branch (i.e. Guanxian-Jiangyou fault). As of August 29, 2008, a total of 261 aftershocks with magnitude larger than 4.0 have occurred. Among them, 31 aftershocks were larger than 5.0, and 8 aftershocks larger than 6.0. The largest aftershock occurred on May 25 with a magnitude of 6.4.

The China Digital Strong Motion Observation Network was completed in March 2008 after nearly five years of construction and trial operation, resulting in a broad distribution of observation stations in China and intensive distribution in some local areas. During the Wenchuanearthquake of May 12, strong-motion records were obtained from about 460 permanent ground stations and three arrays for topographical effect and structural response observation. After the main shock, 59 mobile instruments were quickly deployed in the hard hit areas to record ground motions generated by strong aftershocks.

A total of about 1,400 components of ground motions from the main shock were recorded, and as of 1 August 2008, over 20,000 components from strong aftershocks have been recorded. For the mainshock records, there are more than 500 components with peak acceleration larger than 10 gal (cm/sec/sec), 200 larger than 50 gal, 115 larger than 100 gal, 42 larger than 200 gal, 16 larger than 400 gal, and 7 larger than 600 gal, and no records over 1000 gal.

The largest peak ground acceleration from the main shock is 958 gal recorded at Wolong station in Wenchuan County, Sichuan. Recorded peak accelerations and preliminary peak velocity values from some key stations along the fault are listed below:

Wolong station in Wenchuan County: Epicentral distance = 19 km; Distance to Yingxiu-Beichuan fault = 23 km PGA (gal) - 957.7 (EW), 652.9 (NS), 948.1 (UD) PGV (cm/s) - 51.5 (EW), 41.7 (NS), 30.4 (UD)

Bajiao station in Shifang City: Epicentral distance = 67 km; Distance to Yingxiu-Beichuan fault = 10 km PGA (gal) – 556.2 (EW), 581.6 (NS), 633.1 (UD) PGV (cm/s) – 62.7 (EW), 89.8 (NS), 49.6 (UD)

Qingping station in Mianzhu City: Epicentral distance = 88 km; Distance to Yingxiu-Beichuan fault = 3 km PGA (gal) – 824.1 (EW), 802.7 (NS), 622.9 (UD) PGV (cm/s) – 133.0 (EW), 65.3 (NS), 39.6 (UD)

Zengjia station in Guanyuan City: Epicentral distance = 314 km; Distance to Yingxiu-Beichuan fault = 86 km PGA (gal) – 424.5 (EW), 410.5 (NS), 183.3 (UD) PGV (cm/s) – 44.0 (EW), 25.8 (NS), 24.6 (UD)

Preliminary analysis of these strong-motion records shows the following characteristics of ground motion from the earthquake, especially for near-fault ground motion:

(1) Large peak accelerations are recorded from stations located along the fault, and the distance to the rupturing fault, rather then the epicentral distance, clearly controls the ground motion attenuation.

(2) Peak accelerations at stations in the fault rupture propagation direction are relatively large, consistent with the fault rupture propagation or directivity effect.

(3) In the near-fault records, peak acceleration in the EW direction is in general larger than that in the NS direction.

(4) Peak accelerations at stations on hanging wall of the thrust fault are generally larger than those from stations on footwall, which shows hanging wall effect on ground motion.

(5) The area with large accelerations seems to be relatively larger near the northeast segment of the fault than the southwest segment of the fault.

(6) For some near-fault records peak acceleration is larger in the vertical direction than in one or two horizontal directions.

(7) Large velocity pulses of the ground motion appear in some near-fault records. Peak ground accelerations for both horizontal components from the main shock versus the distance to the fault were compared with other ground motion attenuation relationships. The peak horizontal ground accelerations of the Wenchuan Earthquake decrease much more slowly than those from other attenuation relationships, especially in the fault distance ranging from 100 to 300 km. Some recorded peak values are about 100 gal even at the fault distance of about 400 km. The peak ground acceleration is highly varied in the fault distance ranging from 100 to 500 km.

A team of earthquake researchers, sponsored by the Earthquake Engineering Research Institute (EERI) and the Geo-Engineering Earthquake Reconnaissance (GEER) Association,carried out a field investigation in conjunction with Chinese colleagues from August 4th to 10th to document scientific and engineering effects of the devastating earthquake (moment magnitude7.9 according to the USGS) that occurred in Wenchuan County of Sichuan Province, China, on May 12, 2008. The EERI field investigation was conducted as part of the Learning from Earthquakes Program with funding from the National Science Foundation (NSF).

The EERI-GEER field investigation team was invited by Prof. Zifa Wang, Director of the Institute of Engineering Mechanics-China Earthquake Administration (IEM-CEA), to investigate the effects of the Wenchuan earthquake. During the short period of one week in the field, the EERI-GEER team could only get a small sample of the effects of the great earthquake. Prof.Junwu Dai of the IEM-CEA accompanied the team during the field investigation.

The research team, under the leadership of Dr. Marshall Lew of MACTEC Engineering and Consulting in Los Angeles, California, included experts in structural, lifelines, and geotechnical engineering as well as disaster response and recovery. In addition to Dr.Lew, EERI team members were David Friedman and Dennis Lau of Forell/Elsesser Engineers, Inc., and Laurie Johnson, an urban planning consultant, all of San Francisco, California; Prof. Tricia Wachtendorf of the Disaster Research Center at the University of Delaware in Newark; and Prof. Jian Zhao of the University of Wisconsin at Milwaukee. The GEER team consisted of Prof. David Frost of the Georgia Institute of Technology in Savannah, Prof. J. P. Bardet of the University of Southern California in Los Angeles, and Prof. Tong Qiu of Clarkson University in Potsdam, New York.

This paper presents observations made during the EERI-GEER field investigation in Sichuan Province as well as observations made by other investigators who have visited the earthquake-affected region. Information from other sources have also been incorporated.