Characteristics of Soil Liquefaction using HN of Microtremors in YuanLin area , Taiwan

The characteristics of soil liquefaction were investigated using the HIV ratios of microtremors in the Yuan-Lin area. Liquefaction at Luen-Ya-Li, Yuan-Lin in central Taiwan was clearly observed with serious sand boils bringing about massive damages during the Chi-Chi earthquakeBased on the HN ratios calculated from microtremor measurements at 42 points distributed in this area, the predominant frequencies appeared between 0.8-0.9 Hz for the liquefied area, with higher relative amplification factors compared to other areas. In the study, the ground vulnerability index (K) g values (Nakamura 1996) in the liquefied areas were higher than those in the neighboring areas without liquefaction. This study shows supporting evidence for the first time that the HN ratios of microtremors can be a good alternative indicator for an area's potential for liquefaction. (


INTRODUCTION
Devastations caused by earthquakes directly reflect the local geological condition.Al though the best approach for understanding these ground conditions is through direct observation, such studies are obviously restricted to areas with relatively high seismicity.Due to constraints such as high rates of seismicity and the availability of an adequate reference site, a wide range of different methods has been applied for site response studies.However, the use of microtremors is invaluable, as it requires no other geological information while estimating the effect of surface geology on seismic motion.
Recently, the HIV technique (Nakamura 1989) is becoming more popular with its data collection facilities and application allowance in areas with low or even zero seismicity.Previ ous studies had suggested that the vertical component of ambient noise not only maintains the characteristics of source to surface ground of sediments, but also is significantly influenced by Rayleigh waves on the sediments.Therefore, they can be used to remove both the source and the Rayleigh wave effects from the horizontal components.This technique is effective in iden tifying the fundamental resonant frequency of a sedimentary layer while providing amplifica tion factors that are more realistic than those obtained from sediment to rock site ratios.Many researchers (Ohmachi et al. 1991;Lermo et al. 1992;Field andJacob 1993,1995) have, in fact, shown how such H/V ratios of noise can be used to identify the fundamental resonant fre quency and amplification factor of sediments.Ohmachi et al. ( 1991) and Lermo and Ch<ivez-Garcfa ( 1994) applied the HIV ratio method to analyze microtremor measurements.Lermo and Chavez-Garcia (1993) used it to assess the empirical transfer function of the intense S-wave, part of an earthquake record, obtained from three cities in Mexico.Their results clearly indicated that the HIV ratio could provide a robust estimate of the frequency and amplitude of the first resonant mode, albeit not of the higher modes.In the meantime, Field and Jacob (1993) and Field et al. (1995) considered the re sponse of sedimentary layers to ambient seismic noise and claimed that the HIV ratio method was an effective and reliable tool to identify the fundamental resonant frequencies of a layered sedimentary basin.Further evidence was given by Suzuki et al. (1995) who used both microtremor and strong-motion data in Hokkaido, Japan, and ascertained that the peak fre quency determined by the B/V ratio seemed to correspond with the predominant frequency estimated from the thickness of an alluvial layer.Based on numerical calculations, many other researchers (Lermo andChavez-Garcia 1993, 1994;Lachet and Bard 1994;Dravinski et al. 1996) have shown that the HIV ratio method is obviously able to predict fundamental resonant frequency well.
The Chi-Chi earthquake (Mw::: 7.6) in central Taiwan occurred at 1:47 a.m. on September 21, 1999, was triggered by the reactivation of the pre-existing Che-Lung-Pu fault, and gener ated a rupture more than 105 km in length with a maximum offset of l l m (vertical) and lO m (horizontal).It caused the highest casualties and damages in Taiwan history.The central is land experienced extremely high shaking intensities with extensive soil liquefaction in many areas, especially in Yuan-Lin, Wu-Feng and Nan-Tou.Soil liquefaction caused houses to collapse, bridge piers to move and sway, levees to slide and settle, retaining structures to tilt and overturn as well as roadways and farm fields to crack, spread laterally and subside.Su et al. (2000) recently performed a field reconnaissance and preliminary assessment of liquefac tion in the Yuan-Lin area, while Ueng et al. (2000) reported the preliminary results of their field investigation, subsurface explorations, in-situ and laboratory tests along with their analy ses of soil liquefaction in the same area during the Chi-Chi earthquake.Su and Wang (2000) have done another site investigation and subsequently estimated the liquefaction potential in Yuan-Lin using the SPT (Standard Penetration Test) and CPT (Cone Penetration Test) methods.
However, with only one strong-motion station (TCUl 10) deployed in the Yuan-Lin area, in order to investigate the site characteristics, it is necessary to conduct dense microtremor measurements covering 42 observation points distributed at Luen-Ya-Li and its vicinity.Based on the strong-motion data at TCUl 10 and the microtremor data at the 42 sites, we analyzed the site response using the HIV ratio method, and calculated K values to estimate the potential for g soil liquefaction in the Yuan-Lin area.

TECHNIQUE
Site effects, usually considered empirical transfer functions of the surficial layers, are commonly studied by two techniques: the standard spectral ratio and the HN ratio methods.
The standard spectral ratio, ST' is calculated by dividing the horizontal Fourier spectrum of the ground motions on an alluvium site, S Hs ' by that recorded on a nearby rock site, S H8 • The latter station is taken as the reference station.Thus s = S HS T S . (1)

HB
Following the work of Nakamura (1989), Lermo and Chavez-Garcia (1993) used a spec tral ratio Es to estimate the amplitude effect of the source where Svs and Svs are, respectively, t:.1e Fourier spectra of the vertical motions on the surface and those on the bedrock at a certain depth.Nakamura (1989)  ( 5 ) vs This suggests that the HN ratio, as defined by this transfer function, can be obtained solely from the motions on the surface, which obviously makes it easier to estimate the characteris-tics of ground motion.By employing the HIV ratio, we were able to determine the predomi nant frequency (FF) and the amplification factor (AP) of the site.Nakamura (1996) also pro posed the vulnerability index " K8 value" as a means to determine the extent of liquefaction.The K value is simply derived from strains of ground and structures (Nakamura 1996(Nakamura , 1997, . It can be defmed as In this study, we determined the site characteristics of the Yuan-Lin area using the HIV of microtremors and the K values to predict the potential for soil liquefaction at the site.Many reliable strong motion data sets recorded during the Chi-Chi earthquake allowed us to compare site responses of the strong motion event with weak motion events recorded at TCU 110 during 1993-1998 period.Figure 1 shows the epicenter of the Chi-Chi mainshock (marked with a star) and three weak-motion events (labeled 1, 2 and 3) recorded by TCUll O and TCU120.TCU120 was chosen as the reference site.In order to assess the site response, we extracted the intense S-waves with a cosine taper at a 20-sec length for the 921 mainshock and at a 10-sec length for all other ground motions.Furthermore, we applied the 0.25Hz Hanning smoothing technique for the HIV ratio calculations to avoid pseudopeaks caused by spectral holes.
To understand the characteristics of soil liquefaction in the Yuan-Lin area, we made dense microtremor measurements at 42 sites in Luen-Ya-Li and its neighboring area using the hand type seismocorder system (SPC-5 1) and three-component velocity sensors (VSE-150) with a sensitivity range of 0.1-70 Hz. Figure 2 shows the location distribution of the 42 measured points (Y01-Y42 where YOl represents microtremor station YANOOl, and so on) with the dark triangular symbols indicating the sites with obvious liquefaction.Sites YO 1-Y04 located in Luen-Ya-Li have massive sand boils.Serious damages to buildings due t o soil liquefaction were also observed at sites Y09 and YIO.The microtremor point Y24 is close to the strong motion station TCUl 10.
Velocity data were recorded at a rate of 200 samples per second on a permanent 16-bit, 16-channel data acquisition system.At each observation point, the microtremor measurements were continuously recorded for thirty minutes, and the records were divided into several 40sec segments.We calculated the Fourier spectra for these segments and then smoothed the spectra by using a 0.25Hz Hanning window.In the next stage, the spectral ratios between the horizontal and vertical components were computed.---'-------'-----'---'----'------'----' 119.5 120 120.5 121 121.5 122 122.5 123 Yuan-Lin locates at the alluvial fan of the Cho-Shui Stream (Fig. 3).According to bore hole data by the Central Geological Survey (Chiang et al. 1999), the Yuan-Lin area is mostly on an alluvial structure chiefly composed of silt, mud, clay and sand (very coarse, coarse and medium), and the depth of basement gradually increases from east to west.The S-wave veloc ity at the depth of 0-30m is from 100 m/sec to 250 m/sec and increased to 600 m/sec at the lOOm of depth.According to Ueng et al. (2000), the minimum SPT-N value may have dropped to 2 after the Chi-Chi earthquake.In general, the ground water levels are usually high with depths of 0.5m to 4.0m in this area (Su and Wang 2000).

Comparison of Strong and Weak Motion Events
Based on the high-quality strong-motion recordings, the HN ratio method was first ex amined by comparing it to the traditional spectral ratio method.The HN ratio (thick line) at station TCU 110 (a soil site) and the spectral ratio (TCUl 1 Off CU 120, thin line) are shown in Fig. 4. Station TCU120 was chosen as the reference site.These results were obtained using the S-wave windows for -the three events shown in Fig. 1.Basically, the predominant frequency estimated from both methods is close at about 0.8-0.9Hz, and their trends are similar.On account of the Chi-Chi mainshock, the PGA values at most strong-motion stations are larger than those recorded during the earlier 1993-1998 period.The site responses between the strong and weak motions at station TCUl 10 were also compared.Figure 5 compares the averaged HI V ratios between the 921 mainshock (thin lines) and the weak-motion events (thick lines) at station TCUl 10.The PGA values triggered by the 921 event are larger than 0.18g.Apparently, the predominant frequency caused by the Chi-Chi mainshock (a strong event) shifts to a lower value when compared with that of weak-motion events.This phenomenon indicates that the site probably produced nonlinear behavior.

Microtremor Measurements
In order to investigate the characteristics of soil liquefaction, as mentioned earlier, we did dense microtremor measurements at the 42 sites distributed in Luen-Ya-Li and its neighboring area of Yuan-Lin (Fig. 2).According to Huang (2002), when compared, the predominant fre quencies of the microtremor data and those of the earthquake recordings are similar at most stations although the amplification factors estimated from the microtremor data show lower values.Similar results appear at TCU stations.We compared the HIV ratios between the earth quake recording and the microtremor data at TCUl 10.In Fig. 6, the predominant frequency individually estimated from microtremors and earthquake data is respectively about 1 Hz and 0.85 Hz.
In order to understand the variations in the HIV ratios in this area, we selected 10 sites   higher amplification factors but relatively lower predominant frequencies at sites Y09 and YOl, which explains why larger K values appear at these two stations.In fact, serious liqueg faction with sand boils was observed and caused houses to collapse at these sites during the Chi-Chi earthquake.However, site Y06 near the mountain area has the highest predominant frequency and the lowest amplification factor; thus, the K value is at its lowest at these sites.
g From these observations, it is reasonable to conclude that K is clearly a value which corre g sponds to the site and can be considered as vulnerability index of that site, an indicator which might be useful in selecting weak points of ground especially in liquefied areas.
The predominant frequencies and the relative amplification factors at the 42 measured points are shown in Figs. 9 and 10, respectively.Obviously, the liquefied areas have lower predominant frequencies (0. 7-1.OHz) and higher amplification factors (above 2.5).By contrast, the higher predominant frequencies (above l.5Hz) and smaller amplification factors (near 2.0) appear at the eastern region or near the mountain area.
By employing the predominant frequencies and the amplification fa ctors, we were able to calculate the K values as shown in Fig. 11.The higher K values (above 10) are concentrated g g in the liquefied area.Based on the estimations from the HN ratio of microtremors in Fig. 11, a contour map of the Yuan-Lin area is shownin Fig. 12. Generally speaking, the results calcu lated from the HIV ratio method of microtremor data analysis and the results incorporating  with the liquefaction potential index match well with liquefaction evidence from field investigations.

CONCLUSIONS
The intensity of shaking in the central part of Taiwan was very high during the Chi-Chi earthquake, and extensive liquefaction occurred in many areas.Especially hard hit were the towns of Yuan-Lin, Wu-Feng and Nan-Tou.The liquefaction potential using the SPT-N and the CPT methods has been analyzed and compared in many studies.This study proposed another convenient technique for the first time to examine liquefaction potential.We investi gated the characteristics of soil liquefaction using the HIV of microtremors.The dense microtremor measurements made at the 42 points mainly distributed at Luen-Ya-Li, Yuan Lin.From the HIV ratios of microtremors, the predominant frequencies of the liquefied areas are concentrated at about 0.8-0.9Hz.The relative amplification factors are also higher than in other areas.Based on the predominant frequencies and their amplification factors, one impor tant parameter, K, could further be determined to assess the occurrence of liquefaction.In this g study, the K values in the liquefied areas are higher than those in neighboring areas without

Measured Points
Fig. 9. Distribution map of the predominant frequencies estimated by the aver aged HIV ratios of microtremors at the 42 measured points.Except for a few sites at the eastern side of the area, most of the predominant fre quencies are between 0.7 and 0.9.
assumed that the vertical com ponent of the microtremor spectrum was not amplified by low-velocity surface layers, and he estimated the effect of Rayleigh waves on the vertical components of the tremors by evaluat ing Es.He proposed that if the effect of the Rayleigh waves was the same on the vertical and horizontal components, then Es could be used to eliminate the effect of the Rayleigh waves on the transfer function.In applying this to compensate for the source effect (Es), Lermo and CMvez-Garcfa (1993) introduced a modified site effect function (Srr), namely which is equivalent to also pointed out that the ratio, S H 8: Svs• was nearly 1 which he obtained by examining microtremor measurements in a borehole.Recently, Huang and Teng (1999) exam ined the ratio using microtremors and earthquake recordings at a bedrock site in Chiawan, Taiwan.With these empirical checks, it was assumed here that a reasonable estimate of the modified site effect function could be determined from s = S HS IT S .

g 3 .
DATA AND SITESA dense strong-motion observatory network consisting of 150 stations (coded as TCUxxx where xxx represents the station number) in central Taiwan was deployed by the Central Weather Bureau (CWB) some ten years ago.Among .the 150 stations, only one (TCUllO) locates in the Yuan-Lin area.The Chi-Chi earthquake (Mw=7.6) in the early morning of 21 September 1999 caused extensive damages and liquefaction with severe sand boils at several locations in the Yuan-Lin area of Chang-Hua county.

Fig. I .
Fig. I. L ocation of stations TCUllO(open triangular symbol), reference site TCU 120 (dark triangular symbol), the epi center of the 921 mainshock (Mw=7.6, marked with a star) and three weak-motion events (labeled 1, 2 and 3) re corded by TCU 110 and TCU120.

Fig. 2 .
Fig. 2. Location distribution of 42 measured points (YO 1-Y 42) in the Luen-Ya-Li area of Yuan-Lin and its vicinity.The dark triangular symbols represent the sites with obvious liquefaction.Sites Y01-Y04 located in Luen-Ya-Li have massive sand boils.Serious damage to buildings due to soil liquefaction was also observed at sites Y09 and YlO.

Fig. 3 .
Fig. 3. Map sketching the geology of the middle reaches of the Cho-Shui Stream.The Yuan-Lin area is located at the alluvial fan of the Cho-Shui Stream (revised from Chiang et al. 1999).

Fig. 5 .
Fig. 5. Comparison of the averaged HIV ra tios between the 921 mainshock (thin lines) and the weak-motion events (thick lines) at station TCUl 10.The three panels represent the NS, EW components and their corresponding RMS results.The predominant fre quency caused by the Chi-Chi mainshock shifts to lower value when compared with that of weak motions.

Fig. 4 .Fig. 6 .
Fig. 4. Comparison of the averaged HIV ra tios (thick lines) at TCUl 10 and the averaged spectral ratios (thin lines, TCU110ffCU120).TCU120 was se lected as the reference station.The three panels represent the NS, EW components and their corresponding RMS results.The results of the two methods are similar.
Fig. 7. Comparison of the RMS compo nent of the averaged H/V ratios using microtremor data at 10 (Y06-Y24) of the 42 measured points.Except for YAN005 and YAN006, most of the predomi nant frequencies are between 0.8 and 1.0.