Documentation of soil conditions at liquefaction and non-liquefaction sites from 1999 Chi–Chi (Taiwan) earthquake

The 1999 Chi–Chi, Taiwan, earthquake provides case histories of ground failure and non-ground failure that are valuable to the ongoing development of liquefaction susceptibility, triggering and surface manifestation models because the data occupy sparsely populated parameter spaces (i.e. high cyclic stress ratio and high fines content with low to moderate soil plasticity). In this paper, we document results from several large site investigation programs conducted in Nantou, Wufeng and Yuanlin, Taiwan. The seismic performance of the investigated sites include non-ground failure building and free-field sites, building sites with partial foundation bearing failures, free-field lateral spread sites, and free-field level ground sites with sediment boils. Field and laboratory investigation protocols for the sites are described, including cone penetration testing (some with pore pressure and shear wave velocity measurements) and rotary wash borings with standard penetration testing (including energy measurements). Implications of the SPT energy measurements with respect to established guidelines for the estimation of SPT energy ratio (including short rod corrections) are presented. Finally, data for three example sites are shown that illustrate potential applications of the data set, and which also demonstrate a condition where existing liquefaction analysis procedures fail to predict the observed field performance.     

Role of fluids in surface deformation caused by the 1999 Chi‐Chi earthquake in Taiwan

The 1999 Chi‐Chi earthquake significantly altered the landscape of central Taiwan. Surface deformation produced by the earthquake along the trace of the Chelungpu thrust can be classified into two styles: (1) uplift without significant surface rupture, and (2) uplift accompanied by surface rupture. Here we examine areas that exhibited the first style of deformation (e.g. Wufeng). Seismic stress at the time of the main shock may have been relieved by high pore‐fluid pressure in a 300‐m‐thick sand and gravel aquifer. Along the thrust fault, frictional heating of these sediments resulted in thermal expansion and an increase in pore‐fluid pressure. High pore‐fluid pressure damped seismic‐wave energy and enhanced intergranular slips of unconsolidated sandy and gravel sediments, which were possibly assisted by sulphuric acid corrosion, leading to a high sulphate content in the groundwater (c. 70 mg L^?1). These changes permitted surface folding and terrace‐style uplifting to occur without significant rupture. In contrast, other areas in which the second style of deformation is dominant (e.g. Fengyuen‐Shihkang) have thin (0–10 m) sand and gravel deposits and lower concentrations of sulphate (c. 30 mg L^?1) in groundwater. In these areas, sediments were heated but not sufficiently to produce significant thermal expansion and increase in pore‐fluid pressure; accumulation of stress in these locations led to rupture at the ground surface, with the formation of steep fault scarps. The areas exhibiting the first deformation style are characterized by the presence of high pore‐fluid pressure, frictional heat conduction, and possibly chemical corrosion related to sulphuric acid attack and formation of sulphate, in contrast to those involving significant uplift and surface rupture. The areal distribution of these two surface deformation styles suggests that the aforementioned fluid‐related subsurface processes may have altered the characteristics of sediments and caused diverse responses to the quake. Copyright © 2002 John Wiley & Sons, Ltd.     

Nonlinear identification of the soil response at Dahan downhole array site during the 1999 Chi–Chi earthquake

Records of the 1999 Chi–Chi earthquake, provided by the Dahan downhole array were analyzed, and nonlinear hysteretic stress–strain relations in the soil layers were estimated by the method previously used for studying the response of soils during the 1995 Kobe and the 2000 Tottori earthquakes. The obtained models of the soil behavior were applied for evaluating changes of the shear moduli in the soil layers and for the nonlinear identification of the soil behavior at Dahan site during the Chi–Chi earthquake. We found that reduction of the shear moduli in the soil layers did not exceed 5%, and the soil response was virtually linear. The content of the nonlinear components in the soil response was about 5% of the intensity of the response, and it was mostly due to the odd-order nonlinearities. A similarity in the stress–strain relations describing the behavior of soils during the 1995 Kobe earthquake, the 2000 Tottori earthquake, and the 1999 Chi–Chi earthquake was found, which indicates the possibility to describe the behavior of similar types of soils at different sites by similar stress–strain relations and predict soil behavior in future earthquakes.     

Review and new interpretation for the propagation characteristics associated with the 1999 Chi‐Chi earthquake faulting event

The Chi‐Chi earthquake (M_W = 7.6) took place in central western Taiwan in 1999. The earthquake caused reactivation of the Chelungpu Fault and resulted in 100‐km‐long surface ruptures. The fault strikes mostly north–south to NNE–SSW; however, the northern tip of the southern segment of the surface ruptures rotates clockwise to define an east–west trend, then jumps to a shorter NNW‐trending rupture. The largest vertical displacement is recorded in the Shihkang area of the Shihkang–Shangchi Fault Zone, where vertical slips are up to 8–10 m. The Shihkang–Shangchi Fault Zone displays a complex fault pattern as a linkage damage zone between two fault segments with the greatest concentration of faults and fractures. Our new interpretation, based on recent published geometric, kinematic, and geophysical studies on the Chi‐Chi earthquake fault, suggests that the Shihkang–Shangchi Fault Zone is not a simple termination zone, but may be an ‘overstep zone’ or a ‘transfer zone’. Slip analysis along the surface ruptures indicates that they are composed of three fault segments and the amount of slip partly depends on the intersection angle between slip direction and fault strike. Our numerical modeling for the area indicates that Coulomb stress changes are mainly concentrated on tips and bends of the surface ruptures. Slip patterns indicate that the fault propagates toward the northeast. Therefore, this study suggests high potential for future earthquake activity along the unruptured Shangchi segment. Hence, future geohazard studies should focus on the Shangchi segment to evaluate potential earthquakes, determine recurrence intervals, and reduce future earthquake hazards.     

Non-linear site response simulations in Chang-Hwa region during the 1999 Chi–Chi earthquake, Taiwan

The destructive 1999 Chi–Chi earthquake (M_w 7.5) was the largest inland earthquake in Taiwan in the 20th century. Several observations witness the non-linear seismic soil response in sediments during the earthquake. In fact, large settlements as well as evidence of liquefaction attested by sand boils and unusual wet ground surface were observed at some sites. In this paper, we present a seismic response simulation performed with CyberQuake software on a site located within the Chang-Hwa Coastal Industrial Park during the 1999 Chi–Chi earthquake in Taiwan. A non-linear multi-kinematic dynamic constitutive model is implemented in the software. Computed NS, EW and UP ground accelerations obtained with this model under undrained and two-phase assumptions, are in good agreement with the corresponding accelerations recorded at seismic station TCU117, either for peak location, amplitudes or frequency content. In these simulations, liquefaction occurs between depths 1.3 and 11.3 m, which correspond to the observed range attested by in place penetration tests and other liquefaction analyses. Moreover, the computed shear wave velocity profile is very close to post-earthquake shear wave velocity profile derived from correlations with CPT and SPT data. Finally, it is shown that in non-linear computations, even though a 1D geometry is considered, it is necessary to take into account the three components of the input motion.     

Inversion of near-field waveform data for earthquake source rupture process ( II ): Inversion of Chi Chi earthquake source, 20 September, 1999, Taiwan

The new inversion algorithm developed based on the recent progress in the nonlinear programming study by us is used to invert the earthquake source process of Chi Chi earthquake M _w7.6, 20 Semptember, 1999, Taiwan. A curve fault model is constructed in our inversion to make the fault model close to the real rupturing fault to reduce the influence from the discrepancy between the constructed fault model and the real rupturing fault. The results show that (1) the rupture process of the Chi Chi earthquake source lasted about 32 seconds and the main faulting occurred between 6th to 21st second after the start of the ruptures and the high slip area were mainly located at the northern segment of the fault. (2) The slip was dominated by thrust faulting. The average rake angle was 64.5°, which was very consistent with those inverted by USGS, Harvard and CWB (Central Weather Bureau of Taiwan). The amount of the moment inverted in this paper was 7.76×10²⁰ NM, which was a slightly bigger than those inverted by USGS and Harvard. (3) A clear nucleation step existed in the source faulting process and it lasted about 6 seconds. The moment release rate accelerated obviously at the end of the nucleation step. The faulting started from the southern segment and mainly occurred at the northern segment after 10 seconds. At the end of this paper, we analyzed the reliability of the inversion result via comparing with the GPS observations and discussed its scientific signification.     

Investigating the impact of the Chi‐Chi earthquake on the occurrence of debris flows using artificial neural networks

Debris flows have caused enormous losses of property and human life in Taiwan during the last two decades. An efficient and reliable method for predicting the occurrence of debris flows is required. The major goal of this study is to explore the impact of the Chi‐Chi earthquake on the occurrence of debris flows by applying the artificial neural network (ANN) that takes both hydrological and geomorphologic influences into account. The Chen‐Yu‐Lan River watershed, which is located in central Taiwan, is chosen for evaluating the critical rainfall triggering debris flows. A total of 1151 data sets were collected for calibrating model parameters with two training strategies. Significant differences before and after the earthquake have been found: (1) The size of landslide area is proportioned to the occurrence of debris flows; (2) the amount of critical rainfall required for triggering debris flows has reduced significantly, about half of the original critical rainfall in the study case; and (3) the frequency of the occurrence of debris flows is largely increased. The overall accuracy of model prediction in testing phase has reached 96·5%; moreover, the accuracy of occurrence prediction is largely increased from 24 to 80% as the network trained with data from before the Chi‐Chi earthquake sets and with data from the lumped before and after the earthquake sets. The results demonstrated that the ANN is capable of learning the complex mechanism of debris flows and producing satisfactory predictions. Copyright © 2009 John Wiley & Sons, Ltd.     

Characteristics of strong ground motion during the 1999 Chi-Chi earthquake (Taiwan) and large aftershocks: comparison with the previously established models

In the Taiwan region, the empirical spectral models for estimating ground motion parameters were obtained recently on the basis of recordings of small to moderate (5.0≤M_L≤6.5) earthquakes. A large collection of acceleration records from the recent M_L 7.3 (M_W 7.6) Chi-Chi earthquake (20 September, 1999) makes it possible to test the applicability of the established relationships in the case of larger events. The comparison of ground motion parameters (peak accelerations and response spectra), which were calculated using the stochastic approach based on the modeled Fourier amplitude spectra, and the observed data demonstrates that the models may be successfully used for ground motion prediction for earthquakes of magnitudes up to M_L=6.8–7.0 and hypocentral depth more than 10 km. To satisfy to the peculiarities of ground motion during shallow (depth less than 10 km) and larger (M_L>7.0) events, the models were revised.     

Evaluation of models for Fourier amplitude spectra for the Taiwan region

In the Taiwan region, the empirical spectral models for estimating ground-motion parameters were obtained recently on the basis of recordings of small to moderate (5.0≤M_L≤6.5) earthquakes. A large collection of acceleration records from the M_L=7.3 Chi-Chi earthquake (21 September, 1999) makes it possible to test the applicability of the established relationships in the case of larger events. The comparison of ground-motion parameters (Fourier amplitude spectra, peak accelerations and response spectra), which were calculated using the models, and the observed data demonstrates that the models could provide an accurate prediction for the case of the Chi-Chi earthquake and the largest aftershocks. However, there are some peculiarities in the ground-motion frequency content and attenuation that, most probably, are caused by the features of the rupture process of the large shallow earthquake source.     

Inversion of near-field waveform data for earthquake source rupture process (Ⅱ): Inversion of Chi Chiearthquake source, 20 September, 1999, Taiwan

The new inversion algorithm developed based on the recent progress in the nonlinear programming study by us is used to invert the earthquake source process of Chi Chi earthquake Mw7.6, 20 Semptember,1999, Taiwan. A curve fault model is constructed in our inversion to make the fault model close to the real rupturing fault to reduce the influence from the discrepancy between the constructed fault model and the real rupturing fault. The results show that (1) the rupture process of the Chi Chi earthquake source lasted about 32 seconds and the main faulting occurred between 6th to 21st second after the start of the ruptures and the high slip area were mainly located at the northern segment of the fault. (2) The slip was dominated by thrust faulting. The average rake angle was 64.5°, which was very consistent with those inverted by USGS, Harvard and CWB (Central Weather Bureau of Taiwan). The amount of the moment inverted in this paper was 7.76×1020 NM, which was a slightly bigger than those inverted by USGS and Harvard. (3) A clear nucleation step existed in the source faulting process and it lasted about 6 seconds. The moment release rate accelerated obviously at the end of the nucleation step. The faulting started from the southern segment and mainly occurred at the northern segment after 10 seconds. At the end of this paper, we analyzed the reliability of the inversion result via comparing with the GPS observations and discussed its scientific signification.     

Ground motion characteristics of the Chi‐Chi earthquake of 21 September 1999

The purpose of this paper is to investigate the ground motion characteristics of the Chi‐Chi earthquake (21 September 1999) as well as the interpretation of structural damage due to this earthquake. Over 300 strong motion records were collected from the strong motion network of Taiwan for this earthquake. A lot of near‐field ground motion data were collected. They provide valuable information on the study of ground motion characteristics of pulse‐like near‐field ground motions as well as fault displacement. This study includes: attenuation of ground motion both in PGA and spectral amplitude, principal direction, elastic and inelastic response analysis of a SDOF system subjected to near‐field ground motion collected from this event. The distribution of spectral acceleration and spectral velocity along the Chelungpu fault is discussed. Based on the mode decomposition method the intrinsic mode function of ground acceleration of this earthquake is examined. A long‐period wave with large amplitude was observed in most of the near‐source ground acceleration. The seismic demand from the recorded near‐field ground motion is also investigated with an evaluation of seismic design criteria of Taiwan Building Code. Copyright © 2000 John Wiley & Sons, Ltd.     

Bi-normalized response spectral characteristics of the 1999 Chi-Chi earthquake

To develop uniform and seismic environment-dependent design spectrum, common acceleration response spectral characteristics need to be identified. In this paper, a bi-normalized response spectrum (BNRS) is proposed, which is defined as a spectrum of peak response acceleration normalized with respect to peak acceleration of the excitation plotted vs. the natural period of the system normalized with respect to the spectrum predominant period, Tp. Based on a statistical analysis of records from the 1999 Chi-Chi earthquake, the conventionally normalized response spectrum(NRS) and the BNRS are examined to account for the effects of soil conditions, epicentral distance, hanging wall and damping. It is found that compared to the NRS, the BNRS is much less dependent on these factors. Finally, some simple relationships between the BNRS for a specified damping ratio and that for a damping ratio of 5%, and between the spectra predominant period and epicentral distance for different soil types are provided.     

Nonlinear earthquake analysis of high arch dam–water–foundation rock systems

A nonlinear finite element model for earthquake response analysis of arch dam–water–foundation rock systems is proposed in this paper. The model includes dynamic dam–water and dam–foundation rock interactions, the opening of contraction joints, the radiation damping of semi‐unbounded foundation rock, the compressibility of impounded water, and the upstream energy propagating along the semi‐unbounded reservoir. Meanwhile, a new equivalent force scheme is suggested to achieve free‐field input in the model. The effects of the earthquake input mechanism, joint opening, water compressibility, and radiation damping on the earthquake response of the Ertan arch dam (240 m high) in China are investigated using the proposed model. The results show that these factors significantly affect the earthquake response of the Ertan arch dam. Such factors should therefore be considered in the earthquake response analysis and earthquake safety evaluation of high arch dams. Copyright © 2011 John Wiley & Sons, Ltd.     

Direct finite element method for nonlinear earthquake analysis of 3‐dimensional semi‐unbounded dam–water–foundation rock systems

A direct finite element method for nonlinear earthquake analysis of 2‐dimensional dam–water–foundation rock systems has recently been presented. The analysis procedure uses standard viscous‐damper absorbing boundaries to model the semi‐unbounded foundation‐rock and fluid domains and specifies the seismic input as effective earthquake forces at these boundaries. Presented in this paper is a generalization of the direct finite element method with viscous‐damper boundaries to 3‐dimensional dam–water–foundation rock systems. Step‐by‐step procedures for determining the effective earthquake forces starting from a ground motion specified at a control point on the foundation‐rock surface is developed, and several numerical examples are computed and compared with independent benchmark solutions to demonstrate the effectiveness of the analysis procedure for modeling 3‐dimensional systems.     

Fault rock analysis of the northern part of the Chelungpu Fault and its relation to earthquake faulting of the 1999 Chi-Chi earthquake, Taiwan

Abstract   The 1999 Chi-Chi earthquake in Taiwan ( M _w = 7.6) produced a surface rupture along the north–south-striking Chelungpu thrust fault with pure dip-slip (east side up) and left lateral strike-slip displacements. Near-field strong-motion data for the northern part of the fault illustrate a distinct lack of the high-frequency seismic radiation associated with a large slip (10–15 m) and a rapid slip velocity (2–4 m/s), suggesting a smooth seismic slip associated with low dynamic frictional resistance on the fault. A drillhole was constructed at shallow depths in the possible fault zones of the northern part of the Chelungpu Fault, which may have slipped during the 1999 earthquake. One of the zones consists of a 20-cm-thick, unconsolidated fault breccia with a chaotic texture lacking both discrete slip surfaces (e.g. Riedel shears) and grain crushing. Other possible fault zones are marked by the narrow (less than a few centimeters) gouge zone in which clayey material intrudes into the damaged zone outside of the gouge zone. These characteristic fault rock textures suggest that the slip mechanisms at shallow levels during the earthquake involved either granular flow of initially unconsolidated material or slip localization under elevated pore pressure along the narrow clayey gouge zone. Because both mechanisms lead to low dynamic frictional resistance on the fault, the rapid seismic slip in the deep portions of the fault (i.e. the source region of strong-motion radiation) could have been accommodated by frictionless slip on the shallow portions of the fault. The combination of strong-motion data and fault rock analysis suggests that smooth slip associated with low dynamic friction occurred on both the deep and shallow portions of the fault, resulting in a large slip between the source region and the surface in the northern region.     

Time series decomposition of groundwater level changes in wells due to the Chi‐Chi earthquake in Taiwan: a possible hydrological precursor to earthquakes

The largest and most disastrous earthquake in Taiwan (Mw: 7·3) in the 20th century, the Chi‐Chi earthquake, hit central Taiwan at 01:47 local time on September 21, 1999. The groundwater level changes were rapid at that time. Studies have found that the rapid change in groundwater levels was a co‐seismic phenomenon. This work analyzes the possibility that the abnormal change in groundwater levels may have occurred before the earthquake. Three well stations with a total of five wells are considered. They are all near the Che‐Lung‐Pu fault, which caused the Chi‐Chi earthquake. The time series decomposition method was applied to decompose the seasonal groundwater level, the trend in groundwater levels, and the period of the change in the groundwater level. Residual groundwater levels were found by subtracting the determined seasonal, trend and period data from corresponding data for the original groundwater level. The computed residual water levels in July, August and September of 1999, were transformed into a frequency spectrum by a Fourier method. Additionally, the effects of barometric pressures on the groundwater level changes were also evaluated. Analytical results show that the spectral density functions of the irregular groundwater level in the confined aquifer at the Chu‐Shan well in September behaved differently from those in July and August. We posit that a pre‐seismic hydrogeological anomaly may have existed before the Chi‐Chi earthquake, and can be considered in future studies of anomalies associated with earthquakes. Copyright © 2007 John Wiley & Sons, Ltd.     

Attenuation relations of Arias intensity based on the Chi-Chi Taiwan earthquake data

This paper presents a derivation of attenuation relations of Arias intensity for various site conditions based on the strong-motion data recorded in the 1999 Chi-Chi Taiwan earthquake. The data are from the mainshock and three large aftershocks with stations in the footwall area and in the area away from the fault. At each station, Arias intensity is computed from two horizontal acceleration time histories. The Arias intensity data are separated into four groups according to site classes assigned to recording stations. For each site class, the attenuation relation of Arias intensity is derived using a two-step regression analysis method. The attenuation relations established in this study can be used to estimate Arias intensity from a rupture of a thrust fault for sites in the footwall area or in the area away from the fault.     

Input and system identification of the Hualien soil–structure interaction system using earthquake response data

This paper presents an input and system identification technique for a soil–structure interaction system using earthquake response data. Identification is carried out on the Hualien large‐scale seismic test structure, which was built in Taiwan for international joint research. The identified quantities are the input ground acceleration as well as the shear wave velocities of the near‐field soil regions and Young's moduli of the shell sections of the structure. The earthquake response analysis on the soil–structure interaction system is carried out using the finite element method incorporating the infinite element formulation for the unbounded layered soil medium and the substructured wave input technique. The criterion function for the parameter estimation is constructed using the frequency response amplitude ratios of the earthquake responses measured at several points of the structure, so that the information on the input motion may be excluded. The constrained steepest descent method is employed to obtain the revised parameters. The simulated earthquake responses using the identified parameters and input ground motion show excellent agreement with the measured responses. Copyright © 2003 John Wiley & Sons, Ltd.     

Collision of the Ganges–Brahmaputra Delta with the Burma Arc: Implications for earthquake hazard

We take a fresh look at the topography, structure and seismicity of the Ganges–Brahmaputra Delta (GBD)–Burma Arc collision zone in order to reevaluate the nature of the accretionary prism and its seismic potential. The GBD, the world's largest delta, has been built from sediments eroded from the Himalayan collision. These sediments prograded the continental margin of the Indian subcontinent by  400 km, forming a huge sediment pile that is now entering the Burma Arc subduction zone. Subduction of oceanic lithosphere with > 20 km sediment thickness is fueling the growth of an active accretionary prism exposed on land. The prism starts at an apex south of the GBD shelf edge at  18°N and widens northwards to form a broad triangle that may be up to 300 km wide at its northern limit. The front of the prism is blind, buried by the GBD sediments. Thus, the deformation front extends 100 km west of the surface fold belt beneath the Comilla Tract, which is uplifted by 3–4 m relative to the delta. This accretionary prism has the lowest surface slope of any active subduction zone. The gradient of the prism is only  0.1°, rising to  0.5° in the forearc region to the east. This low slope is consistent with the high level of overpressure found in the subsurface, and indicates a very weak detachment. Since its onset, the collision of the GBD and Burma Arc has expanded westward at  2 cm/yr, and propagated southwards at  5 cm/yr. Seismic hazard in the GBD is largely unknown. Intermediate-size earthquakes are associated with surface ruptures and fold growth in the external part of the prism. However, the possibility of large subduction ruptures has not been accounted for, and may be higher than generally believed. Although sediment-clogged systems are thought to not be able to sustain the stresses and strain-weakening behavior required for great earthquakes, some of the largest known earthquakes have occurred in heavily-sedimented subduction zones. A large earthquake in 1762 ruptured  250 km of the southern part of the GBD, suggesting large earthquakes are possible there. A large, but poorly documented earthquake in 1548 damaged population centers at the northern and southern ends of the onshore prism, and is the only known candidate for a rupture of the plate boundary along the subaerial part of the GBD–Burma Arc collision zone.