Site Classification of Taiwan Free-field Strong-Motion Stations
by Chyi-Tyi Lee, Chin-Tung Cheng, Chi-Wen Liao and Yi-Ben Tsai
Based on the available geologic and geomorphologic data, 708 Taiwan Strong-motion Instrumentation Program (TSMIP) free-field strong-motion station sites are classified using a scheme compatible with the 1997 UBC provisions. Results show that an extensive usage of geologic maps and geomorphologic data for site classification is satisfactory, and is a quick and effective method for categorizing large numbers of strong-motion station sites. The response spectral shapes (RSS) method and the horizontal-to-vertical spectral ratio (HVSR) method both provide good supplements for checking purposes. Field checks are definitely necessary, especially for sites located near geologic boundaries or geomorphologic boundaries. Final comments for each station site have been made after these checks. Some problematic sites do exist and are mentioned in the text. Data from these sites must only be used with care. To increase the accessibity of our results, we have developed a web-based query system at http://gis.geo.ncu.edu.tw/query/site/.
The local site conditions are an important factor in the recorded waveform of earthquake ground motions. Different site conditions can induce amplifications of different period ranges in the response spectra (Seed et al., 1976; Mohraz, 1976). A building may be more severely damaged if the natural periods of the structure happen to be close to the amplified periods of the ground. Therefore, the local site conditions become important in ground motion analysis and in earthquake resistant designs. An earthquake response spectrum compatible with local site condition, anchoring to an appropriate peak ground acceleration (PGA), is a common input for structural dynamic analysis. There are presently 708 free-field strong-motion stations operating in Taiwan. This network has recorded huge numbers of waveforms. These data are invaluable for strong-motion studies and site effect analyses, as well as for the study of a practical site classification system.
The objective of site classification is to classify a group of strong-motion station sites into several classes so that the conditions within the same site class are similar and design engineers may understand the general site condition by the class that it belongs to. The geologic condition of interest at a site is commonly restricted to the upper-most layers. Most site effect studies of earthquake ground motion are based on the soil properties in the upper 30 meters (Anderson et al., 1996). Because quantitative subsurface soil properties are not commonly available for every site, the use of surface geology becomes important in understanding the subsurface geologic conditions. Empirical relationships between the surface geology and the subsurface shear-wave velocity have been developed and are used in ground motion amplification predictions (Joyner and Fumal, 1985; Boore et al., 1993; Borcherdt, 1994; BSSC, 1998; Park and Elrick, 1998; Castro, 1997). Some researchers had combined the use of surface geology and shear-wave velocity for site classification (e.g. Borchertdt 1994a, 1994b). The classification of site conditions and an estimation of the site amplification ratio obtained by using surface geology and geomorphologic units have also been performed in Tokyo and Kanagawa, Japan (Yamazaki et al., 2000).
Earthquake response spectra alone are also useful for site classification, when geologic data are not sufficient. Seed et al. (1976) used the normalized shape characteristics of strong-motion response spectra at a 5% damping ratio to explore site conditions. However, strong-motion data from different sources and paths may produce different spectral shapes (Yamazaki and Ansary, 1997). To avoid this, Yamazaki and Ansary (1997) used the horizontal-to-vertical ratios from the velocity response spectra of JMA strong-motion data for site classification. They concluded that the horizontal-to-vertical spectral ratios are more independent of the earthquake magnitude, path, and depth at a specific site. Atkinson (1993) and Theodulidis et al. (1996) found similar results from near-field strong-motion records. Zare et al. (1999) also succeeded in site classification using the horizontal-to-vertical spectral ratios of the velocity response spectra and the Fourier spectra from an Iranian strong-motion network.
In Taiwan, Kuo (1992, 1993, and 1994) used the 1991 UBC provisions (ICBO, 1991) to classify 156 free-field strong-motion station sites into four categories, namely, S1, S2, S3, and S4 (from hard to soft, respectively). His classification scheme is based on the surface geology and on borehole data. In 1998, an earthquake loss estimation system - HAZUS (FEMA, 1997) was introduced to Taiwan by the National Science Council. Since that time, we have participated in this project, known as “Haz-Taiwan”, being responsible for the preparation of the PESH (potential earth science hazard) data for the system. We consider both the method proposed by Borchertdt (1994a, 1994b) and the one by the 1997 NEHRP (BSSC, 1998) as a basis for a classification scheme, using available geologic data, for a preliminary study of strong-motion site classification in Taiwan (Cheng et al., 1998; Liao et al., 2000). Our preliminary site classification has been tested in the system. After the 1999 Chi-Chi earthquake, a large amount of strong-motion data has been recorded. Meanwhile, more geologic/geomorphologic data have become available, so we have conducted a more refined site classification, by using surface geology from large-scale geologic maps, geomorphologic data and borehole data. The results are presented here and are also compared with the response spectral shapes and the horizontal-to-vertical response spectra ratios from a large number of well-recorded strong-motion data. With some necessary field checks, a preferred site classification has been made for each free-field strong-motion station.
In order to understand the source effects, wave propagation effects, and site effects of Taiwan earthquakes, the Seismology Center of the Central Weather Bureau in Taiwan operates about 700 modern digital accelerographs at free-field sites (Shin et al., 2000). The instruments installed at each station include triaxial accelerometers, a digital recorder and a timing system (Liu et al., 1999). The selection criteria for strong-motion sites are set as follows: (1) in the nine metropolitan areas in Taiwan; (2) near fault zones; (3) at different geologic sites - rocks, soil and soft soils; and (4) near important structures, industrial sites and nuclear power plants. Within these criteria, the free-field stations are spaced approximately 5 km apart, except in the Central Mountain Range, where stations are only spotted. The distribution of strong-motion stations in the TSMIP network is shown in Figure 1.
The classification scheme used in this study basically follows the 1997 UBC (ICBO, 1997), 1997 NEHRP (BSSC, 1998) classification scheme. Since subsurface geologic data is relatively deficient compared to surface geologic and geomorphologic data, we have also adopted Borcherdt’s geologic criteria (Borcherdt, 1994a, 1994b) and have intensively used large-scale geologic maps of Taiwan in our working procedure (Table 1). The method is similar to that used in the Haz-Taiwan program (Cheng et al., 1998; Liao et al., 2000). The site classes B, C, D, and E, comparable to SB, SC, SD, and SE in the 1997 UBC provisions, are used in the present classification system. Class A and class F are not used at present. Details of this simplified working scheme will be described in the following section.
Since the use of geologic maps may place some limitation on the accuracy and may introduce some errors into the classification scheme, we also performed classifications by using the response spectral shape (RSS) (Seed et al., 1976) and the horizontal-to-vertical spectral ratios (HVSR) methods (Yamazaki and Ansary, 1997; Zare et al., 1999) for purpose of comparison and checking. We use the four site conditions (Type 1 through Type 4 in Figure 2) from the mean acceleration spectral shapes utilized by Seed et al. (1976) for the categorization of soil types. We also use HVSR to categorize the site conditions. According to the Japanese code (Japan Road Association, 1980), the dominant period (TG) for rock and hard soil sites is less than 0.2 seconds, for hard soil TG=0.2-0.4s, for medium soil TG =0.4-0.6s, and for soft soil TG≧0.6s. We determine the soil types according to average HVSR peaks, which fall within the above-mentioned period bands. Figure 3 illustrates the HVSR for the site classification of K-NET (http://www.k-net.bosai.go.jp/overview.html) strong-motion stations (J.X. Zhao, unpublished data).
The results are compared among different classifications. The classifications of Kuo (1992, 1993, and 1994) are also used for comparison. Then, an evaluation of the site class is made for each strong-motion station. If there is no consistency between the results from different classification methods, then these questionable sites should be checked in the field before any final site classification can made.
Data used in this study include: geologic maps, a digital elevation model (DEM), borehole data, shear-wave velocities, and ground motion records. In Taiwan, the available regional scale geologic maps include 1/50,000 scale and 1/250,000 scale geologic maps from the Central Geological Survey (CGS), and 1/100,000 scale geologic maps from the Chinese Petroleum Corporation (CPC). For the use of geomorphologic analyses, a set of 40-meter-resolution digital elevation models (DEM) produced by the Taiwan Agriculture and Forestry Aerial Survey Institute is available. Borehole data are mostly available in the urban areas, especially in the Taipei basin. The small amount of such data outside urban areas has been provided by some of the new highway projects. Shear-wave velocity data can be found in published reports. Abundant strong-motion records can be obtained from the TSMIP network. A total of 1676 3-component strong-motion records from 11 large earthquakes are used in this study.
The geologic maps used in this paper are the 1/50,000 scale ones from the CGS. The 1/100,000 scale maps from the CPC are used, when 1/50,000 scale maps are not available. Finally, 1/250,000 scale maps from the CGS are used, when neither the 1/50,000 nor the 1/100,000 scale maps are available. Figure 4 shows a composite map of the above-mentioned geologic maps. A total of 31 sheets of 1/50,000 scale CGS geologic maps are used in this study, which cover most of the foothills and coastal regions in northern, southern, and eastern parts of Taiwan. The CPC 1/100,000 scale geologic maps cover most of the western coastal plains and foothills, and just fill in the important parts in central western Taiwan (See Figure 4).
In order to establish the site classes as shown in Table 1, the soil and rock types indicated in the geologic maps are examined in detail. Due to the wet climate, the weathering rate in Taiwan is extremely high; therefore site class A (hard rock) is not applicable here. Site class B (rock) and site class C (soft rock or very dense soil) can be distinguished by their geologic age and rock type. Site class B may include igneous rocks, metamorphic rocks, limestone, and hard volcanic deposits. Sandstones, shales, conglomerates, and slates of Miocene age or older are all classified as class B. Pliocene and Pleistocene Sandstones, shale/mudstones, and conglomerates are considered as soft rocks and thus classified as class C. Volcanic breccias and pyroclastic rocks of a similar age can also be grouped in class C. Taiwan’s lateritic highlands consist of a several-meter-thick laterite layer on top with weathered bedrock underneath. Lateritic terraces commonly consist of several-meters-thick laterite with several-meters to more than ten-meters-thick sandy gravel deposits overlaying the bedrock. Both the lateritic highlands and the lateritic terraces are classified as class C.
Late Quaternary deposits, other than lateritic terraces, such as loose sand, silt, clay, and gravel deposits, are considered as engineering soils, and classified as class D (stiff soil) or class E (soft soil). Soils in class D are fluvial terraces, stiff clays and sandy gravel deposits, while Holocene alluvium flood plains or recent fills usually form soils in class E. To differentiate between stiff soil and soft soil sites, SPT-N values from borehole data of the new rapid transit system and some highway projects are used. The average SPT-N values are calculated for the top 30 meters of soil. Soils with average SPT-N values greater than 15 are defined as class D; others are defined as class E. In areas without boreholes, geomorphologic units control the class E boundary.
All the geologic maps have been digitized using careful coordinate control under the MapInfo Geographic Information System (GIS). The geologic ages, lithology and names of the strata were also entered into the database for further analysis. DEMs were transferred to a color-shaded image with appropriate sun angles to enhance the relief. The borehole data were analyzed and the average SPT-N values for the upper-most 30 meters were put into the database. The shear-wave velocity data were also examined, but the amount of them are too small to be used in the analysis. After the map layers and the attribute database are established, overlay analysis and a spatial query of the GIS can be performed and the geologic site classes can be obtained. The resultant geologic site class map is showed in Figure 5.
After this, the locations of free-field strong-motion stations have been overlaid on the geologic site class map, so that the site class can be easily seen for each strong-motion station. By using the GIS spatial query function, the strong-motion stations are easily categorized with the geologic site class map. A complete list of the geologic site classes and related information for each of the 708 strong-motion station sites can be found in the CD-ROM attached in this volume and on our web page (http://gis.geo.ncu.edu.tw/query/site/).
In this section, we compare the geologic site classification results with that of three other classification systems: (1) RSS classification, (2) HVSR classification, and (3) Kuo’s classification (Kuo, 1992, 1993, and 1994). In order to obtain good quality accelerograms for the RSS and HVSR analyses, we carefully selected 11 large shallow earthquakes that occurred at different locations in Taiwan, triggering many stations during their occurrence. The distribution of these earthquakes is shown in Figure 6. To reduce the uncertainty, the MW7.6 Chi-Chi earthquake (Ma et al., 1999), which reveals significant source effects, is not selected; no subduction zone earthquakes are used in this study either. Descriptions of these earthquakes and the number of stations triggered by each earthquake are listed on Table 2. A total of 1676 strong-motion records from 554 stations are used for the present study.
According to the RSS shown in Figure 2, we classified the TSMIP strong-motion station sites into four types, namely Types 1, 2, 3, and 4 (Seed et al., 1976). Type 1 (rock site) and Type 4 (soft soil) can be identified easily, because the amplification period ranges are obvious. However, we commonly need some more care to differentiate Type 2 (stiff soil) and Type 3 (deep non-cohesive soil), because their amplification period ranges are rather close. In addition, strong-motion data obtained from different sources or paths may result in different spectral shapes (Yamazaki and Ansary, 1997; Silva and Green, 1989), which may increase the difficulty of the comparison. However, with careful selection of earthquake sources and by using of a modal spectral shape from many spectra, they can still be differentiated.
The results show that most Type 1 sites are correlated with site class B (rock site); most Type 2 (stiff soil) sites may be correlated with site class C (soft rock or very dense soil); most Type 3 sites may be correlated with site class D (stiff soil); and most Type 4 sites may be correlated with site class E (soft soil). Table 3 gives the detailed correlation results.
According to the HVSR method shown in Figure 3, we categorized the TSMIP strong-motion station sites into four types, namely Types I, II, III, and IV. Type I (rock and hard soil) emphasized at about 0.15 seconds shows a nearly uniform shape on the HVSR. Type IV (soft soil) can easily be recognized, because of its high peaks at long periods (1.0 second). Peak value of Type II (hard soil) is rather close to Type I, and that of Type III (medium soil) is rather close to Type IV, they must be examined with care at each site. The HVSRs are commonly more stable among different earthquakes than that of the RSSs. This may be because this method is more independent of the earthquake source and wave propagation path (Aki, 1988; Atkinson, 1993; Theodulidis et al., 1996).
Most of the type I sites correlate well with the site class B. Type II sites may be correlated with site class C, but some type I sites and type III sites also correlate with site class C. Type III sites may correspond to site class D, but some type II sites and type IV sites also correspond to site class D. Type IV (soft soil) sites correlate with site class E, but some type III sites also correlate to this class (See Table 3).
Kuo (1992, 1993, and 1994) used the 1991 UBC criteria (ICBO, 1991) for classification of site condition. He classified 156 free-field strong-motion stations into four categories, namely S1, S2, S3, and S4, based on the surface geology and borehole data. Results reveal that S2 may be correlated with both site classes C and D. S3 may correlate with both site class D and E, and S4 sites correlate well with site class E (See Table 3). This correlation is similar to the result of Dobry et al. (2000), however, category S1 is not discussed, because there are only 11 samples.
The results from the aforementioned four different classification systems and other related data have been compiled to form a database in MapInfo GIS, so that statistical analyses can be easily performed. Items in the database include: the station names, the coordinates and elevations of stations, the relief of the area surrounding a station, the geologic classification, the geologic map scale, the RSS classification, the HVSR classification, Kuo’s results, and some final comments. According to the statistics and Table 3, we find some fairly good correlations among the different classification systems. Site class B may be equivalent to type 1 of Seed et al. (1976), or type I in Figure 3, or S1 of 1991 UBC (ICBO, 1991). Site class C may be roughly equivalent to type 2, or type II, or S2. Site class D may correspond to type 3, or type III, or S3, and site class E may correspond to type 4, or type IV, or S4. These correlations can be seen in Table 3. It is noted that problematic sites are not included in the statistics in Table 3.
Generally speaking, the results after geologic site classification are supported by the results from RSS and from HVSR and are consistent with Kuo. However, a few sites are apparently not so consistent for the different classification types, and a reexamination of the original geologic and geomorphologic data is needed. For example, for stations at the edge of a terrace, the edge of a basin, or near the border of a plain or hills, the geologic classification may not be accurate enough, because of the accuracy of the map or because of complicated subsurface geology. At these types of locations, RSS and HVSR may be also insensitive in differentiating type 2/II and type 3/III. We must then carefully examine the geomorphologic characteristics and the local geology, so that a better classification can be made.
Some questionable sites can be successfully explained by the local geology after field checks. For example: The Alishan station (ALS and CHY074) is a rock site according to the geologic map, but the local subsurface geology actually consists of thick colluvial deposits, so the site condition is very dense soil rather than rock. The station at the National Taiwan Ocean University (TAP065) is located on a garbage dumpsite, so the local conditions are very soft soil instead of rock, as indicated by the regional geologic map. The Ilan ILA020 station is located on a lake deposit upstream of an ancient natural dam, so it should be a soil site instead of a rock site, as shown on regional geologic map.
At some other sites, which are not located near geologic boundaries or geomorphologic boundaries, the RSS and HVSR observations are very strange, compared with the geologic site class. In these cases, we also need to go to the field to actually inspect the site and to observe whether any anomaly exists. During the field checks, we found that some of these sites are located on ridge tops. A typical ridge top case is the Sun-Moon lake site (TCU084), which is located on a very steep mountain summit where the bedrock is exposed nearby. It is without doubt a rock site, but the topographic effects may be significant, so that the earthquake response spectra are amplified at long periods. Sites with such ridge top conditions include: CHN5, CHY042, CHY080, CHY086, SML, TAP074, TCU084, TCU130, TTN017, WDT, WHF, and YUS.
Some of the other questionable site conditions found during the field checks include: (1) a site being located near a building, wall, tank, or scarp; (2) the concrete slab base of the station being affected by tree roots. The station at the Shinchu Science Park (TCU017) is typical of root problem. The station at the National Central University (TCU083) is typical of the effects being inside a building. The station at Minchian (TCU129), which recorded 1g during the 1999 Chi-Chi earthquake, has some complications due to the construction of the concrete platform of the station (Wen et al., 2000). The station at Chutzuhu (TAP056) shows high PGA and high frequencies for many earthquakes, but the reason for this is still not known (Wen et al., 1998). Data from these problematic sites should only be used with care.
To summarize, we have found the geologic classification method to be both quick and effective method for categorizing large numbers of strong-motion stations, and the RSS and HVSR methods both provide good supplements for checking purposes. Geomorphology also plays an important role in geologic classification, as a geomorphologic unit is commonly closely related to a geologic unit. The final site class comments for each strong-motion station are shown in Figure 7. A complete list of the coordinate, relief, site classifications, and final comments for each site, can be found on our web page (http://gis.geo.ncu.edu.tw/query/site/) for reference purpose. Statistical results (Figure 8) show that all the soft soil sites (class E) are located at low elevations (less than 50 m) or flood plains; most of the stiff soil sites (class D) are also located at lower elevations, some of them on fluvial terraces with elevations less than 150 m; most very dense soil or soft rock sites (Class C) are to be found on lateritic terraces and highlands at elevations greater than 150 m. Most rock sites (site class B) are located at hilly and mountainous highlands, expect for some on limestone, igneous and metamorphic rock sites at lower elevations.
In the present study, we carefully digitized large-scale CGS and CPC geologic maps. More than 50% of the free-field strong-motion site conditions can be determined using the CGS’s 1:50,000 scale geologic maps. And more than 62% of the free-field strong-motion site conditions can be determined using the CPC’s 1:100,000 scale geologic maps. As a result, more than 85% of the stations are covered for site classification by good geologic maps, which ensure the quality of the results.
Site class D and site class E cannot be differentiated only using geologic maps. They need to be classified by using subsurface geotechnical data and geomorphology. Because the drilling data were all collected from major highway and mass rapid transit projects, the data quality is good and consistent. The boundaries for geomorphologic units have been interpreted from good quality DEM color shaded images.
However, uncertainties may still exist, due to the limitation of the map scale and the shortage of subsurface geotechnical data, especially when a station is close to the boundary of two different geologic units. It is also difficult to estimate the thickness of subsurface soil layers by examining surface geology. Therefore, the parallel use of the RSS and HVSR methods becomes important to reduce the uncertainties.
To use the RSS and HVSR methods, we must carefully select the source and distance of earthquakes, so that very large magnitude and very distant earthquakes are not used in developing the response spectra. After the careful selection of earthquakes, the RSS method is found to be good for determining the dominant period of a site, especially for a rock site, which has a conspicuous spectral shape. However, source and path effects still cannot be separated with this method (Yamazaki and Ansary, 1997). The HVSR method may be better to isolate the influences from earthquake magnitude, depth and path, and reflects the site conditions (Yamazaki and Ansary, 1997; Theodulidis et al., 1996; Atkinson, 1993). However, after the careful selection of the earthquake sources, results in this study reveals that HVSR method is not superior to the RSS method in terms of site classification.
Both RSS and HVSR methods are also affected by local site defects and may be confused with false signals, which should be carefully checked in the field. Any deficiency in a site, for example, being adjacent to a big building or penetrated by big tree roots, can affect the spectral shape. Undoubtedly, data from a problematic site should be used with great caution. Furthermore, topographic effects (Jibson, 1987; Aki, 1988) also influence the spectral shapes, especially the ridge top effect, as has been mentioned in the previous section.
Huge amounts of strong-motion data were recorded during the 1999 Chi-Chi earthquake. Source effects can be clearly seen in the accelerograms of this earthquake, with an amplification of the long period part seen at many sites, even at some rock sites. These records are not used for evaluation of site class in this study. Some soft soil sites in the Chiayi area show a short-period response during smaller earthquakes, but a long-period response during large earthquakes. At these sites, the HVSR method is more consistent for small and large earthquakes, but the RSS method is consistent with the HVSR method only for large earthquakes. In this case, we have neglected the RSS results for the smaller earthquakes.
During the 1999 Chi-Chi earthquake, some sites did show liquefaction, e.g. the Yuanlin site (TCU110) and the Wufeng site (TCU065). We can see such liquefaction phenomenon, not only near the sites, but also in the long period waves, which appear after the PGA, on the velocity time histories of the strong-motion data. A strong-motion station located in a liquefaction area should be classified as site class F according to 1997 UBC provisions (ICBO, 1997). This has not been done for the present study and further study should be carried out in the near future on this topic.
After screening out problematic sites and records, we plot the RSS and HVSR results in Figure 9 and Figure 10, for all the records from 11 earthquakes, according to the site classes. We examine the spectral shapes for each site class and compare them with the spectral shapes proposed by Seed et al. (1976). We also examine the spectral ratio and compare them with Figure 3. The results show that the general spectral shapes for each site class are fairly consistent with those 4 site condition types of Seed et al. (1976) with similar results being found by the HVSR method. This implies that our classification results are acceptable.
Figure 10 reveals that in Taiwan site class B has a dominant period at 0.2 seconds. This is a little bit larger than the K-NET type I (0.15 second) and may indicate that the rock conditions in Taiwan are not as strong as in Japan’s rock sites. Topographic effects may be another explanation for this result. Figure 11 summarizes the RSS and HVSR curves for the four the sites classes. It is clear from this figure that site class B (rock) and site class E (soft soil) have completely different spectral shapes and spectral ratios. However, site class C (very dense soil) and site class D (stiff soil) are not so different.
All 708 TSMIP free-field strong-motion station sites in Taiwan were studied and classified to fit into 4 site classes, namely B (rock), C (soft rock or very dense soil), D (stiff soil), and E (soft soil), according to the existing geologic and geomorphologic data available. Results have been further checked by good quality response spectral shape (RSS) and horizontal-to-vertical spectral ratios (HVSR) from 1676 strong-motion records from 11 large earthquakes. After syntheses and field checks, we finally assigned a site class for each station. The results are shown in Figure 7 and are summarized in a table on our web page (http://gis.geo.ncu.edu.tw/query/site/).
Classification based on geologic criteria and the extensive usage of geologic maps and geomorphologic data is satisfactory. The results are compatible with the RSS method and the HVSR method results and also consistent with Kuo’s results. The RSS and the HVSR methods are especially helpful for evaluating a site class, when it is located at the edge of a basin or on a hillside. It appears to be an efficient way to produce a site classification map for a whole region from geologic maps and geomorphologic information. The RSS and HVSR classification schemes, however, require a lot of well-recorded ground motion or microtremor data.
The site classification results in this paper are still rudimentary. They should only be used with care. If an engineering project site is near a strong-motion station, the site class can be used directly. If there is no nearby strong-motion station, users can get a site class from Figure 7. To increase the accessibity of our results, we have developed a web-based query system at http://gis.geo.ncu.edu.tw/query/site/. This page is based on the MapInfo MapXtreme technique and provides spatial query functions, which allow users to acquire a site class by clicking on the web browser or by inputting the coordinates of a site.
Further studies on site classification should be carried out using more specific subsurface geologic and geotechnical data. The NCREE (National Center for Research on Earthquake Engineering) of Taiwan is cooperating with the CWB (Central Weather Bureau) in drilling about 50 holes at some of the TSMIP stations. We propose that they should commence drilling at some of the questionable stations as listed in the table on the website. This year, the NCREE is also performing a study of liquefaction sites in Taiwan. After the liquefaction site data become available, site class F can then be assigned to some TSMIP stations in the future. Site class C (very dense soil) and site class D (stiff soil), as revealed by Figure 11, are not so different. This also requires further study. If they are still similar, then we may consider combining them into one class to simplify the classification system.
The authors extend our deepest thanks to the Seismology Center of the Central Weather Bureau, Taiwan, for providing the strong-motion data. Out thanks are also due to the Highway Bureau, MOTC and the Department of Rapid Transit Systems, Taipei, for providing the borehole data. We also appreciate the very helpful comments of Prof. Ta-liang Teng of USC and Dr. Roger D. Borcherdt of USGS, Menlo Park. We thank Dr. J. X. Zhao of Institute of Geological & Nuclear Science Ltd. and Prof. K. L. Wen of the National Central University for their valuable suggestions and reading of the manuscript. This research was supported by the National Science Council through Grant NSC89-2625-Z008-006.
Aki, K. (1988). Local site effects on strong ground motion, Proceedings, Earthquake Engineering and Soil Dynamic II - Recent Advances in Ground Motion Evaluation, ASCE, Geotechnical Special Publication 20, 103-155.
Anderson, J. G., Y. Lee, Y. Zeng, and S. Day (1996). Control of strong motion by the upper 30 meters, Bull. Seism. Soc. Am. 86, 1749-1759.
Atkinson, G. M. (1993). Notes on ground motion parameters for Eastern North America: duration and H/V ratio, Bull. Seism. Soc. Am. 83, 587-596.
Boore, D. M., W. B. Joyner, and T. E. Fumal (1993). Estimation of response spectra and peak accelerations from Western North American earthquakes: an interim report, U.S. Geol. Surv. Open-File Rept. 93-509.
Borcherdt, R. D. (1994a). Estimates of site-dependent response spectra for design (methodology and justification), Earthquake Spectra 10, 617-653.
Borcherdt, R. D. (1994b). An integrated methodology for estimates of site-dependent response spectra, seismic coefficients for site dependent building code provisions, and predictive GIS maps of strong ground shaking, Proceedings of Seminar on New Developments in Earthquake Ground Motion Estimation and Implications for Engineering Design Practice, Applied Technology Council ATC 35-1, 10-1~10-44.
Building Seismic Safety Council (BSSC) (1998). 1997 Edition NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, FEMA 302/303, Part 1 (Provisions) and Part 2 (Commentary), developed for the Federal Emergency Management Agency, Washington, DC., 337 pp.
Castro, R. R., M. Mucciarelli, F. Pacor, and C. Petrungaro (1997). S-wave site-response estimates using horizontal-to-vertical spectral ratios, Bull. Seism. Soc. Am. 87, 256-260.
Cheng, C. T., C. W. Liao, C. T. Lee, and Y. B. Tsai (1998). A preliminary study of site classification for Taiwan Free-field Accelerograph Stations, 7th Taiwan Geophysics Symposium, Chung-Li, 115-124. (in Chinese)
Dobry, R., R. D. Borcherdt, C. B. Crouse, I. M. Idriss, W. B. Joyner, G. R. Martin, M. S. Power, E. E. Rinne, and R. B. Seed (2000). New site coefficients and site classification system used in recent building seismic code provisions, Earthquake Spectra 18, 41-67.
International Conference of Building Officials (ICBO) (1991). Uniform Building Code, Whittier, California., 1050 pp.
International Conference of Building Officials (ICBO) (1997). Uniform Building Code, Whittier, California., 492 pp.
Japan Road Association (1980). Specification for highway bridges, Part V, Earthquake Resistant Design.
Jibson, R. (1987). Summary of research on the effects of topographic amplification of earthquake shaking on slope stability, U.S. Geol. Surv. Open-File Rept. 87-268.
Joyner, W. B. and T. E. Fumal (1985). Predictive mapping of earthquake ground motion, in Evaluating Earthquake Hazards in the Los Angeles Region – An Earth-Science Perspective, J. E. Ziony (Editor), U.S. Geol. Surv. Profess. Pap. 1360, 203-220.
Kuo, K.W. (1992). The geological character of the CWB Strong Motion Network – Taipei area, Central Weather Bureau, 77 pp. (in Chinese)
Kuo, K.W. (1993). The geological character of the CWB Strong Motion Network – Taoyuan Hsinchu and Miaoli area, Central Weather Bureau, 76 pp. (in Chinese)
Kuo, K.W. (1994). The geological character of the CWB Strong Motion Network – Chianan area, Central Weather Bureau, 98 pp. (in Chinese)
Liao, C.W., C.M. Huang, S.W. Liao, C.T. Cheng, and C.T. Lee (2000). Constructing a database for seismic hazard potential analysis - a preliminary study, Programs and Abstracts, 2000 Annual Meeting, Geol. Soc. China, Taipei, 246-248. (in Chinese)
Liu, K. S., T. C. Shin, and Y. B. Tsai (1999). A free field strong motion network in Taiwan: TSMIP, TAO 10, 377-396.
Ma, K. F., C. T., Lee, and Y. B., Tsai (1999). The Chi-Chi, Taiwan Earthquake: Large Surface Displacements on an Inland Fault, EOS, Transations, American Geophysical Union 80, 605 & 611.
Mohraz, B. (1976). A study of earthquake response spectra for different geological conditions, Bull. Seism. Soc. Am. 66, 915-935.
Park, S. and S. Elrick (1998). Predictions of shear-wave velocities in southern California using surface geology, Bull. Seism. Soc. Am. 88, 677-685.
Seed, H. B., C. Ugas, and J. Lysmer (1976). Site-dependent spectra for earthquake-resistant design, Bull. Seism. Soc. Am. 66, 221-243.
Shin, T. C., K. W. Kuo, W. H. K. Lee, T.-L. Teng, Y. B. Tsai (2000). A preliminary report on the 1999 Chi-Chi (Taiwan) earthquake, Seism. Res. Lett. 71, 1, 23-29.
Silva, Walter j. and Robert K. Green (1989). Magnitude and distance scaling of response spectral shapes for rock sites with applications to North American tectonic environment, Earthquake Spectra. 5, 591-623.
Theodulidis, N., P.-Y. Bard, R. Archuleta, and M. Bouchon (1996). Horizontal-to-vertical spectral ratio and geological conditions: The case of Garner Valley downhole array in southern California, Bull. Seism. Soc. Am. 86, 306-319.
Wen, K. L., H. Y. Peng, K. W. Kuo, and K. S. Liu (1998). Comparison between the strong motion records at sites TAP056 and TAP066, Symposium on Taiwan Strong Motion Instrumentation Program (III), 114-122.
Wen, K. L., H. Y. Peng, Y. B. Tsai, and K. C. Chen (2000). Why 1 g was recorded at TCU129 site during the 1999 Chi-Chi, Taiwan earthquake? Bull. Seism. Soc. Am. (Submitted)
Yamazaki, F., and M. A. Ansary (1997). Horizontal-to-vertical spectrum ratio of earthquake ground motion for site characterization, Earthquake Eng. Struct. Dyn. 26, 671 & 689.
Yamazaki, F., K. Wakamatsu, J. Onishi, and K. T. Shabestari (2000). Relationship between geomorphological land classification and site amplification ratio based on JMA strong motion records, Soil Dyn. Earthquake Eng. 19, 41-53.
Zare, M., P. Y. Bard, and M. Ghafory-Ashtiany (1999). Site characterizations for the Iranian strong motion network, Soil Dyn. Earthquake Eng. 18, 101-123.
Institute of Applied Geology
National Central University at Chung-Li, Taoyuan County,
Taiwan 32054, R.O.C.
Institute of Geophysics
National Central University at Chung-Li, Taoyuan County,
Taiwan 32054, R.O.C.
(C.T.C., C.W.L., Y.B.T.)