The Site Response and Strong Ground Motion Estimation of the 2006/05/27 Yogjakarta, Indonesia Earthquake Tao-Ming Chang National Center for Earthquake Engineering Research, Taipei, Taiwan Jer-Ming Lin Institute of Geophysics, National Central University, Chung-Li, Taiwan Data Iranata Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia ABSTRACT: On May 27, 2006, a magnitude 6.3 earthquake struck Yogjakarta region of the Java Island, Indonesia where no significant earthquake happen for the past. From Harvard University and USGS, the announced earthquake parameters are highly consistent. But the widely spread damage areas suggest very high ground shaking which conflict with the teleseismic results. To find a possible answer for this, we measure micro-tremor for 44 sites around Yogjakarta region. Using the spectrum H/V ratio method, the survey results can match the basic geology of Yogjakarta and suggest those highly damaged area were suffer from basin effect due to soft loosely consolidate soil which is volcanic material erupted from Merapi volcano. We also use these site response results to construct the synthetic seismograms to estimate the strong ground motion for this earthquake Keywords: site response, micro-tremor, strong ground motion estimation 1 INTRODUCTION On 05:53:58, May 27, 2006, a magnitude Mw6.3 earthquake struck the Yogjakarta, the largest city of central Java, Indonesia. There is no seismometer installed in this region by governmental agencies and research institutes due to its low historical seismicities. Therefore, the available source parameters were announced by the Harvard University and USGS using teleseismic signals recorded by global seismic networks. Comparing these two results (Table 1), the differences are very small and can be considered as highly reliable except the epicenter locations which are normal for studying the teleseismic event without any local data. Despite the consistency of source parameters announced by Harvard and HSGS, there is an intensity map (Figure 1) in USGS s website showing a very large region (roughly 40x60 km) in the Yogjakarta regime had ground shaking intensity as high as VIII (MMI scale). As mentioned previously, there is no seismometer installed in this region, therefore the data used to plot the intensity map is reported from end-users via internet. Although this kind data may be influenced by human s subjective judgments but also reveals the statistical importance while the samples is lager enough. This intensity map can explain the distribution of the casualties and collapsed residential buildings very well, but it is very difficult for seismologists to image why a magnitude 6.3 earthquake will produce the similar intensity that a magnitude 7.7 earthquake can produce! Therefore a discrepancy exists between the teleseismic source parameters and the local intensity map. There are two possible reasons for this discrepancy. First of all, the sediment may be poorly consolidated because Yogjakarta city is located at the foot of Meripi volcano, therefore the soft soil will produce the site amplification basin effect. Second of all, the construction quality of local residential buildings may not good enough to stand for the ground shaking. And these two possible reasons will cause the intensity over-estimated, thus the internet reported intensity may be over-estimated. This is reasonable because the MMI scale was made for the building in USA. To clarify this discrepancy, we carry two seismometers to Yogjakarta city to measure the microtremor signals of different locations. Using the spectrum ratio method, the analyzed microtremor data can shows the site response of different locations. During five days, 44 measurements were taken. The map of these measurements and the dominant frequency is shown as Figure 2 and 3. 2 MICRO-TREMOR AND SITE RESPONSE Microtremor, also called as earth noise, is the summation of many seismic signals which caused by various sources including the nature and man-made vibration energy. The vibration sources include, for
example, biology activities, traffic, wind, ocean tides and etc. Short period microtremor is consisting of Rayleigh waves induced by local traffic vibrations from many directions. The major advantage of microtremor survey is the less cost, fast, convenient, and easy to analyze. The microtremor data is very easy to obtain than the traditional strong ground motion data which require installing acceleration seismometers for many years. Microtremor survey is very efficient, especially in the urban area where the seismic reflection survey and drilling wells are not easy to be done. Using microtremor data to study the underground geological structures, Professor Kanai (1962) of Tokyo University did a lot pioneer works. One of his conclusions is that the amplitudes of certain period waves exist in the alluvium or weathered sediment sites are larger than those at rock sites. This is because the multiple reflections and resonance of seismic signals in the sediment layers. After that, Kanai and Tanaka (1962) discuss the relation between predominant period and geological layer structures using the distribution curves of seismic wave periods. Katz (1976) did similar research using power spectrums of microtremor. Microtremor can be applied for different purposes, such as investigation on structure vibrations, characteristics of geological structures, site selection for important facilities, sediment thickness, velocity structures of sediments, amplification effect of soft soils. Usually the most used method in studying the effect of site response or the amplification effect of soft soil layer is the two station spectrum ratio method (Borcherdt, 1970; Chávez-Gracía et al., 1990; Field et al., 1992) which is a simple and effective way to eliminate the source and propagating path effects for the regions with many earthquakes happen in the vicinity (Lermo and Chávez-García, 1993; Field and Jacob, 1995; Bonilla et al., 1997; Riepl et al., 1998). The critical point for using twostation spectrum ratio method is that a good reference site can be identified with respect to the soil site which will be studied. Usually the reference site is the site on the outcrop of bedrock which is not too far away from the soil site, therefore the spectrum ratio can really eliminate the source and path effects for the strong motion data from the same earthquake. However it is really difficult to find a good reference site, and this limit the use of this method. To overcome the difficulty in searching a good reference site, Nakamura(1989) proposed an empirical single station horizontal/vertical spectrum ratio (H/V Ratio) method which utilizing the micotremor data measured on site to study the site effect. In the beginning, the H/V ratio method is applied in studying site response for using microtremor data. After that, Lermo and Chávez- Gracía(1993) applied the same technique to strong motion data of Mexico and suggested similar site response results were obtained from S wave of strong motion data and microtremor data for four Mexican cities. Field et al.(1990) pointed out that microtremor data can be used in site response and micro-zonation studies. Lermo and Chávez-Gracía(1994) analyzed the weak motion, strong motion, and microtremor data of Mexico city, and found the microtremor data can be used in estimating the predominant frequency and amplification factor of sediment layers. Most microtremor researches confirm that the dominant frequency of soft sediments can be perfectly identified using Nakamura H/V ratio method. Nakano et al. (2000) shown a detail dominant frequency distribution map which is created using 341 microtremor measurements, strong ground motion data and local geology, and proposed for future seismic microzonation and seismic design standards. 2.1 Field Measurement and Data Processing In this study, we use a three-component seismometer VSE-311C and a recorder SAMTAC-801B made by Tokyo Sokusin corporation, Japan. The bandwidth of this system is 0.07~100Hz. During this five days microtremor survey in Yogjakarta region, 44 measurements were made with roughly 5km spacing. Each measurement lasts 18 minutes and using sampling rate 200 sps. To avoid the unwanted vibrating influenced by wind, the sensor was isolated using a plastic bucket with sponge lay between the ground and bucket. Figure 4 shows the field measurement photo. The data processing method was proposed by Nakamura(1989), who used single station horizontal to vertical spectrum ratio to analyze the microtremor data. The spectrum analyzed results can show the specific characteristic of dominant frequency of each site. The processing procedures are described as follows. (1) Select the acceleration three components microtremor data. (2) Cut data using window length 4096 points. Apply 5% cosine taper on two ends of each windows. (3) Deselect noisy windows which strongly influenced by artificial vibrations due to nearby human activities during the measuring. After this step, selected window will reduce to 30~50. (4) Transform each component of selected data to frequency domain using FFT. (5) Combine the north-south and east-west component Fourier spectrum into horizontal RMS spectrum. (6) For every window, divide horizontal Fourier spectrum by vertical Fourier spectrum and smooth five times to avoid the spectral holes.
(7) Average every window s H/V spectrum ratio and obtain Nakamura H/V Ratio for each site. (8) Plot the site response characteristic map (figure 3). 2.2 Microtremor dominant frequency and site response effect Base on the geological map of Yogjakarta published by Indonesia Geological Survey in 1995, the northern side of Yogjakarta city is the Merapi volcano and volcanic deposits; the western side is tertiary limestone with andesitic breccias and conglomerate hills; the eastern side is teriary limestone and tuffbreccias, volcanic breccias hills; the southern side is the tertiary limestone and marly sandstone with height less than 50 meters and quaternary volcanic undifferentiated tuff, ash, breccias and lava flows in the valley. The boundary between the southern sedimentary valley and eastern limestone hills is generally considered a blind fault; and the slope is pretty steep (>45 degree). During the microtremor survey in Yogjakarta region, we observed most sites are loosely consolidating sandy sediments. This is consistent with the H/V ratio results which indicate the deep soft soil site responses. Figure 3 shows the dominant frequency map of Yogjakarta region. The site responses of sediment valley imply the soil is very thick and at least top 30~50 meters is loosely consolidated. The small hills of limestone and volcanic breccias show characteristic of hard bedrock sites. The total number of microtremor data is limited, though, that the severe earthquake damages is somewhat related to the soft soil site response is confirmed. 2.3 Interview people live in disaster area In addition, we interviewed several local people who live in the severe damaged areas about the most impressive characteristics of main shock. Some of the testimonies are very useful in delineating the ground shaking of the earthquake. One man who lives in Jetis where the village almost completely destroyed said When the earthquake happen, I was frighten and ran out. When I got outside, I can see a thing because the dusts which due to house collapsing cover the sky and shade the sunlight. Another man who live in Bantu said My house had large cracks but did not collapse in the main shock, and fall down in the aftershock. When the earthquake happen the ground shaking last almost one minute but the violent shaking had only seven seconds. According to these testimonies, we can conclude that the ground shaking and the major rupture duration is still not deviated from the average behaviors of earthquakes all over the world. The poor construction quality of residential houses might be the major for over estimated earthquake intensities reported to USGS via internet. 2.4 The characteristics of Yogjakarta earthquake Summarize the teleseismic source parameter, microtremor survey, and earthquake victim interview; we list several conclusions of Yogjakarta earthquake as following. (1) From teleeismic location result, there is no distinct difference than other earthquakes all over the world. (2) There is no fault rupture traces on the ground surface, therefore the major rupture plane probably is deeper than 5km. (3) According to local people s testimonies, the total rupture time of this earthquake may be less than 10 seconds, the major rupture probably is between 3~5 seconds. (4) The construction quality might be the primary reason for severe damage on residential houses. The peak ground acceleration is probably less than 0.2g. (5) The microtremor survey shows that Yogjakarta city and its eastern, southern and south-eastern sides (before reach hills) have distinct soft soil site amplification effect. 3 GROUND MOTION PREDICTION Recently the computer technology has tremendous progress, some complicated and difficult tasks which can not be done ten years ago are now affordable for individual researcher by using PC-clusters. In this study, we use wave-number integration method to compute the synthetic seismograms. We will briefly discuss the source, propagation path effect, velocity mode. Since 1950s, Thomson (1950) used the matrix method to deal with the wave propagation in 1-D layered media, and Haskell (1953) promoted it into computing seismic wave propagation, the modern seismology step into its fast growing stage. Nowadays seismologists are very familiar with the synthetic seismograms computation method for 1-D layer model. Wave-number integration is one of these methods. The advantage of this method is that many type waves, such as body wave and surface waves can be calculated simultaneously, and it can deal with the elastic and inelastic attenuation properties of media. Comparing with other 3-D wave propagation techniques, wave-number integration method needs very few memory spaces and can generate high frequency seismic signals but fail to deal with the scattered waves. But when we wish to know the near-field seismic vibrations from a magnitude 6 earthquake, the source (rupture fault plane) become a plane source instead of a point source. Under such
situation, the synthetic seismograms computed using wave-number integration are all right to use because the seismograms are now sensitive to source rupture patterns instead of scattered waves. The disaster regions are very close to the epicenter of Yogjakarta earthquake, therefore the wave-number integration method with the plane source is a proper way to study the strong ground motion patterns. To generate high frequency seismic signals from a plane source, the fault plane need to be divided into many subfaults and thus the finite subfault area will have shorter source time functions which mean the high frequency seismic signals. In this study, we summarize the geological, geophysical, and seismological background information, and will do the strong ground motion numerical simulations of Yogjakarta earthquake in the following manner. First of all, we will construct the rupture plane using the source parameters published by Harvard University and USGS. Second of all, the rupture fault plane will be divided into many subfaults. The size of each subfault is 1km by 1km.. Third of all, the source rupture parameters will be set as follow. The earthquake depth will be different for different scenarios. When the earthquake began to rupture, the rupture front will propagate outward using roughly 0.85Vs and the maximum fault displacement will be decreased exponentially outward from the earthquake focus. The rupture velocity, source time function, slip length, slip direction, the initial rupture time for each subfault will be randomly changed slighly. 4 DISCUSSION AND CONCLUSION On the morning 05:53:58, 2006/05/27, an earthquake struck Yogjakarta region, Indonesia and caused huge casualties and property losses. It also raised a question for seismologists: why a magnitude 6.3 can cause such tremendous disaster? From a reconnaissance in situ, we found two possible answers for it. First of all, Yogjakarta is located in the south side of Merapi volcano. A lot of places, such as river valleys, are full of unconsolidated volcanic rupturing materials; therefore it becomes the perfect conditions for seismic amplification effect (pretty much like the basin effects). Second of all, the construction quality of residential houses is not good enough to stand the strong shaking. Now we have many microtremor data, it is possible to begin the numerical simulations to reconstruct the strong ground motion of this earthquake. In the future, the simulation results will be provides to civil engineers for developing new and cost effective way to construct residential houses. REFERENCES Bonilla, L. F., Steidl J. H., Lindley G. T., Tumarkin A. G., and Archuleta R. J. 1997. Site amplification in the San Fernando Valley, California:variability of site-effect estimation using the S-wave, coda, and H/V method. Bull. Seism. Soc. Am. 87:710-730. Borcherdt, R. D. 1970. Effects of local geology on ground motion near San Francisco Bay. Bull. Seism. Soc. Am. 60:29-61. Chávez-Gracía F. J., Pedotti G., Hatzfeld D., and P.-Y. Bard P.-Y. 1990. An experimental study of site effects near Thessaloniki(Northern Greece). Bull. Seism. Soc. Am. 86:646-654. Field, E. H., Hough S. E. and Jacob K. H. 1990. Using microtremors to assess potential earthquake site response:a case study in Flushing Meadows, New York City. Bull. Seism. Soc. Am. 80:1456-1480. Field, E. H., Jacob K. H., and Hough S. E. 1992. Earthquake site response estimation: a weak-motion case study. Bull. Seism. Soc. Am. 82:2283-2306. Field, E. H., and Jacob K. H. 1995. A comparison and test of various site-response estimation techniques, including three that are not reference-site dependent. Bull. Seism. Soc. Am. 85:1127-1143. Haskell, N. A., 1953. The Dispersion of Surface Waves in Multilayered Media, Bull. Seism. Soc. Am. 43:17-34. Kanai, K. 1962. On the spectrum of strong earthquake motions. Bull. Earthq. Res. Inst. 40:71-90. Kanai, K. and Tanaka 1962. On the predominant period of earthquake motions. Bull. Earthq. Res. Inst. 40:855-860. Katz, L. J. 1976. Microtremors analysis of local geological conditions. Bull. Seism. Soc. Am. 66(1):45-60. Lermo, J. and Chávez-García F. J. 1993. Site effect evaluation using spectral ratios with only one station. Bull. Seism. Soc. Am. 83:1574-1594. Lermo, J. and Chávez-García F. J. 1994. Are microtremors useful in site response evaluation?. Bull. Seism. Soc. Am. 84:1350-1364. Nakamura, Y., 1989. A method for dynamic characteristies estimation of subsurface using microtremor on the ground surface, QR of RTRI, 30(1): 25-33. Nakano, M., Fukuwa N. and Tobita J. 2000. Regional variation of ground motion in Nobi Plain, Japan, based on seismic records, microtremor and geological data. An international conference on Geotechnical & Geological Engineering, GeoEng 2000. Riepl, J., Bard P. Y., Hatzfeld D., Papaioannou C., and Nechtschein S. 1998. Detailed evaluation of site response estimation methods across and along the Sedimentary Valley of Volvi(EURO-SEISTEST). Bull. Seism. Soc. Am. 88: 488-502. Thomson W. T. 1950. Transmission of elastic waves through a stratified solid. J. of Applied Physics 21:89-93.
Figure 2. The location map for microtremor measurements in Yogjakarta region. Table 1. The source parameters. Institute Lon. Lat. Depth strike dip rake dyne-cm M W Harvard 110.49 8.00 33.4 51 90 14 4.0 10E25 6.3 USGS 110.286 8.007 37 241 85 10 4.2 10E25 6.3 Figure 1. The computer generated shake map for the Yogjakarta earthquake. This is under the Community Internet Intensity Map (CIIM) project. The data used is reported by endusers who are in the vicinity region of earthquake. This map showing a fairly large area (roughly 100x80 km) with strong shaking (intensity VI) which is very unlikely caused by a magnitude 6.3 earthquake. Figure 3. The Yogjakarta dominant frequency map created using Hspectrum H/V ratio method. Figure 4. The photo of field measurement.