Geodetically observed surface displacements of the 1999 Chi-Chi, Taiwan,
earthquake
Ming Yang1, Ruey-Juin Rau2, Jyh-Yih Yu1, and Ting-To Yu2
1Department of Surveying Engineering, National Cheng Kung University, Tainan, Taiwan 2Satellite Geoinformatics Research Center, National Cheng Kung University, Tainan, Taiwan
(Received February 18, 2000; Revised May 17, 2000; Accepted May 18, 2000)
The 21 September 1999 Chi-Chi, Taiwan, earthquake of magnitude MW=7.6(ML =7.3)severely deformed
the Earth’s crust in the central Taiwan region. The earthquake created an 85-km-long surface rupture along the Chelungpu fault. The epicenter was located at 23.85◦N, 120.81◦E, near the southern end of the rupture zone. Three-dimensional displacements of 285 geodetic control stations were determined in this study from Global Positioning System (GPS) observations collected before and after the earthquake. The detailed surface displacement field shows that individual stations are vertically uplifted by up to 4 m and displaced horizontally by up to 9 m, with the largest displacement occurring near the northern end of the ruptured thrust fault. The azimuth of the surface displacement field is approximately parallel to the direction of tectonic convergence of the Eurasian and Philippine Sea plates. The maximum three-dimensional displacement of 9.9 m is among the largest fault movements ever measured for modern earthquakes.
1.
Introduction
Taiwan lies at the junction of the Eurasian plate and the Philippine Sea plate, which converge near latitude 24◦N at ∼7 cm/yr in a N50W direction (DeMetset al., 1990; Senoet al., 1993; Yuet al., 1997; Yuet al., 1999). As a result, this region is one of the world’s most unstable areas with frequent major earthquakes. The 01:47 local time (17:47 GMT the previous day), 21 September 1999 Chi-Chi earthquake was the largest earthquake(MW =7.6, ML = 7.3)that struck
the island in this century. This massive earthquake not only produced significant horizontal and vertical surface displace-ments over a very large area of 10000 km2(100 km×100 km),
but also caused approximately 10000 collapsed structures in several cities along the ruptured Chelungpu fault, and over 2300 fatalities and 8700 injuries.
In this study, we used Global Positioning System (GPS) geodetic data to quantify the three-dimensional surface dis-placement pattern associated with the Chi-Chi earthquake. Pre-earthquake data were collected between 1995 and 1998 from global and regional permanent GPS tracking networks, as well as at Taiwan’s first- and second-order geodetic con-trol stations. Post-earthquake data were collected within one month of the earthquake, including field surveys at geodetic control points located in the central Taiwan area and contin-uous observations at stations of the regional GPS tracking network. The resultant high precision co-seismic displace-ment field of the earthquake can be further exploited to pro-vide valuable information for seismologists and earthquake engineers to study the complex pattern of surface faulting in Taiwan.
Copy right cThe Society of Geomagnetism and Earth, Planetary and Space Sciences (SGEPSS); The Seismological Society of Japan; The Volcanological Society of Japan; The Geodetic Society of Japan; The Japanese Society for Planetary Sciences.
2.
GPS Data Processing
Since 1990, researchers from Taiwan started to use GPS interferometry to characterize regional tectonic motion and estimate relative velocities near and within the island (Yuet al., 1997; Yuet al., 1999). It is clear in Fig. 1 that significant velocity contrasts are present in the central, south and east regions of the island. In 1995, a decision was made by the Ministry of the Interior, R.O.C. to establish a new geodetic datum, the Taiwan geodetic datum 1997 (TWD97), to replace the old reference frame created in the late 1970s by means of triangulation. The GPS-based TWD97 was directly linked with the International Terrestrial Reference Frame (ITRF) by connecting a regional GPS permanent tracking network with globally distributed International GPS Service (IGS) stations. The regional tracking network was further intensified with 726 first- and second-order geodetic control stations that cov-ered the entire region of Taiwan with an average spacing of 10–15 km, including the island of Taiwan and several islands across Taiwan Strait, adjacent to Mainland China.
In this study, pre-earthquake data were obtained in field surveys between April 1995 and May 1998 and from contin-uous observations at stations of the regional contincontin-uous GPS tracking network and the IGS global network. We divided these data into two groups. The first group contained daily GPS observations made at the global and regional tracking sites. The second group, however, only included measure-ments collected at the first- and second-order control sta-tions during 124 field sessions, and continuously recorded regional tracking data covering 24 hours on the day of each field session. Forty-eight IGS stations were used in the study. Taiwan’s only IGS site, TAIW, located in the northern part of the island, is the only station included in both the global and the regional networks. Figure 2 illustrates the geographical
Fig. 1. Taiwan tectonic velocityfield after Yuet al.(1997).
Table 1. GPS data reduction summary.
Models Values
Data intervals 30 sec for tracking data, 15 sec forfield survey data
Elevation angle cut-off 15◦
GPS orbit IGS precise ephemeris
Ionospheric delay Zero (canceled by forming the ionosphere-free phase)
Tropospheric delay Modified Hopfield model value with one estimated correction parameter per station every 4 hours
a prioriphase sigma ±1 cm
Data editing threshold 3-sigma value
Fig. 3. Daily solutions of baseline distance between IGS site GUAM and tracking station FLNM.
Fig. 4. Daily solutions of baseline distance between IGS site GUAM and tracking station KMNM.
distribution of the TWD97 stations. Installed in the regional tracking network were dual frequency TurboRogue receivers with Dorne & Margolin geodetic antennas. The continuous data were recorded every 30 seconds by the regional track-ing network, in accordance with the IGS standard. For the 124 field sessions, the fundamental session length was 4– 5 hours, and the dual frequency observations recorded by GPS receivers of various types were sampled every 15
sec-onds. The number of GPS receivers used in thefield
sur-veys varied greatly, with a minimum of 2 and a maximum of
Table 2. ITRF94 epoch 1997.0 velocities and velocity/coordinate standard deviations for the nine regional tracking stations in the north-south (N), east-west (E), and vertical (U) directions.
Station Velocity Velocity Sigma Coordinate (m/yr) (±m/yr) Sigma (±m)
FLNM N 0.0076 0.0002 0.0003 E 0.0098 0.0005 0.0007 U −0.0179 0.0011 0.0013
KDNM N −0.0069 0.0002 0.0003
E 0.0050 0.0005 0.0007 U −0.0170 0.0011 0.0014
KMNM N −0.0092 0.0002 0.0003
E 0.0330 0.0005 0.0008 U −0.0181 0.0011 0.0015 MZUM N −0.0096 0.0002 0.0003 E 0.0326 0.0005 0.0008
U −0.0216 0.0011 0.0019
PKGM N 0.0000 0.0002 0.0003 E 0.0270 0.0005 0.0008 U −0.0665 0.0011 0.0015 TAIW N −0.0086 0.0002 0.0002 E 0.0364 0.0005 0.0006
U −0.0374 0.0011 0.0014
TMAM N 0.0021 0.0002 0.0003 E 0.0028 0.0005 0.0008 U −0.0170 0.0011 0.0016 TNSM N −0.0063 0.0003 0.0004 E 0.0358 0.0010 0.0012
U −0.0367 0.0024 0.0028
YMSM N −0.0103 0.0005 0.0004 E 0.0352 0.0011 0.0008 U −0.0178 0.0011 0.0013
123 stations involved in a single session. As the GPS satellite geometry was different for various sessions, all stations were surveyed with multiple occupations.
Table 3. Three-dimensional displacement vectors and associated sigma values sampled at 285 TWD97 stations.
Station Long. (◦) Lat. (◦) de (cm) dn (cm) du (cm) σe(±cm) σn(±cm) σu(±cm)
Table 3. (continued).
Station Long. (◦) Lat. (◦) de (cm) dn (cm) du (cm) σe(±cm) σn(±cm) σu(±cm)
Table 3. (continued).
Station Long. (◦) Lat. (◦) de (cm) dn (cm) du (cm) σe(±cm) σn(±cm) σu(±cm)
Table 3. (continued).
Station Long. (◦) Lat. (◦) de (cm) dn (cm) du (cm) σe(±cm) σn(±cm) σu(±cm)
Goad and Yang, 1997; Yang and Lo, 2000). Since high-precision orbits of the GPS satellites could be obtained from
IGS data centers with an estimated coordinate RMS of 5–
10 cm (Neilan et al., 1997; Beutler et al., 1999), in this
study we usedfixed IGS precise ephemeris. To remove the
first-order ionospheric effect, the so-called ionosphere-free
phase combination was formed (Beutleret al., 1996). The
modified Hopfield model (Goad and Goodman, 1974) was
used to give the nominal tropospheric delays on GPS phase measurements. Additionally, one correction parameter per station every 4 hours was estimated to absorb unmodelled tropospheric errors. A summary of our data reduction strat-egy is given in Table 1. We primarily used the Bernese 4.0 software, developed at the University of Berne, Switzerland, as our fundamental tool for GPS data analysis (Beutleret al., 1996).
The network adjustment was carried out in two steps. In the first step only data in the first group, i.e., daily mea-surements from the global and regional tracking networks, were analyzed. The coordinates and velocities of the IGS stations were stochastically constrained to theira priori val-ues defined in the ITRF94, with the associated coordinate and velocity standard deviations defined in the same frame (Boucheret al., 1996). Given the high data density and suffi -cient elapse time, reliable coordinate and velocity estimates defined in the ITRF94 at epoch 1997.0 (approximately the mid-point of the data span) for the regional tracking
sta-tions were determined (Yanget al., 1999). The resultant
millimeter-level coordinate and velocity standard deviations for the regional tracking stations are summarized in Table 2. The coordinates of the 726first- and second-order control stations were determined in the second step, where only the GPS data in the second group were analyzed. The stochastic information regarding the regional tracking stations obtained from thefirst step, i.e., the ITRF94 velocities and coordinates and their associated standard deviations listed in Table 2, was applied in the second step. The coordinate variation rates of thefirst- and second-order stations, however, were not estimated, as the number of occupations per station was not sufficient for us to recover reliable velocity information.
The resultant 1997.0 TWD97 station coordinates, includ-ing the regional trackinclud-ing stations, were measured with accu-racies ranging from 0.1 to 3.0 cm in the east-west and north-south directions, and 0.1 to 9.7 cm in the vertical direction for most stations. The precision of the vertical component was noticeably worse than that of the horizontal components for the 726first- and second-order points, which was mainly caused by the insufficient length of observation sessions (only 4–5 hr) for effective decorrelation of the vertical component with other error sources (Beutleret al., 1989; Kuanget al., 1996; Yanget al., 1999).
The collection of post-earthquake data at the TWD97fi rst-and second-order stations located in the central Taiwan area was completed within one month of the devastating event. Each station was occupied at least twice withfield sessions of 4–5 hr in length. The data were then combined with con-tinuously recorded regional tracking data covering 24 hours on the day of each observation session. To identify seismic effects associated with the earthquake on the regional track-ing stations, we examined several long baselines connecttrack-ing
Fig. 5(b). Measured horizontal displacements associated with the Chi-Chi earthquake (nearfield).
the distant IGS site GUAM with the regional tracking stations for a period of 80 days (day 230–309). Two distinct cases are discussed here. Figure 3 displays the daily solutions of
baseline length (∼2688 km) for the GUAM-FLNM (located
in the east coast of Taiwan Island) baseline. The earthquake occurred at 17:47 on day 263 (GMT), therefore the solution for that day was obtained using only data collected prior to the epoch. FLNM experienced an apparent co-seismic mo-tion of the order of 30 centimeters, the largest among all continuous tracking stations. Nevertheless, no evidence of pre- or post-seismic motions can be definitely identified in Fig. 3. Figure 4 shows the daily solutions of baseline length
(∼3000 km) for the GUAM-KMNM (located on Kinman
Is-land across Taiwan Strait) baseline. A secular decrease in the distance due to tectonic convergence can be observed from Fig. 4; however, different from the previous case, in this line we cannot positively single out any seismic signals associated with the earthquake.
Fig. 6. Measured vertical displacements associated with the Chi-Chi earthquake, where the main shock is shown as a star and station locations are indicated by triangles. Surface rupture is indicated by a thick line. Displacement of a single station is shown by a vector and corresponding 1-sigma error bar.
the earthquake epoch at 21 September 1999, annual station velocities for the TWD97 stations were interpolated from the velocityfield given in Fig. 1 (Yuet al., 1997). Since the rel-ative inter-seismic motions for the sampled stations are well known, there is little error introduced by interpolating these motions. The final surface displacement field is sampled at 285 GPS-derived vectors measured with 1-sigma errors ranging from 0.6 to 3.9 cm in the east-west direction, 0.6 to 3.4 cm in the north-south direction, and 1.0 to 11.8 cm in the vertical direction. Thefinal displacement vectors and their associated sigma values are presented in Table 3.
3.
Analysis
rup-ture extends towards the northeast for an additional 10 km in a major sub-event and splinters into complex branches (Ma
et al., 1999). Individual GPS stations within 15 km east of the fault (the hanging-wall) generally are uplifted by 0.2–
4 m and displaced northwestward by 1.5–9 m. The largest
horizontal and vertical offsets are placed at the northern end of the ruptured Chelungpu fault. The GPS stations situated further to the east of the rupture zone all the way to the east coast generally are subsided by up to 0.8 m, and displaced
northwestward by 0.1–3 m. To the west of the fault (the
footwall), in general, individual stations are subsided by up to 0.5 m and displaced southeastward by up to 1.2 m. It is also interesting to note that after the earthquake liquefaction-related ground failures have been reported in regions near the western coastline, which can be identified with the erratically large subsidence quantities up to 0.5 m in the corresponding area (see Fig. 6).
4.
Conclusion
Damaging earthquakes have previously occurred in west-ern Taiwan several times this century: in 1904 (Tou-Liu
M = 6.1), 1906 (Mei-Shan M = 7.1), 1916–17
(Nan-Tou M=6.8), 1935 (Hsin-Chu Tai-Chun M= 7.1), 1941
(Chung-Pu M=7.1), 1946 (Hsin-Hua M=6.1), and 1964
(Bai-Ho M =6.3). Although several of these earthquakes
produced vertical slips of the order of 1 m (Cheng et al., 1999), none has released as much seismic energy as the
Chi-Chi earthquake. As Taiwan’s strongest earthquake of the
century, this Mw= 7.6 earthquake has produced spectacu-lar surface ruptures among the spectacu-largest fault movements ever observed from modern earthquakes (Maet al., 1999).
The co-seismic displacements of the Chi-Chi earthquake were accurately determined in this study with geodetic data recorded by the dense Taiwanese GPS network before and after the earthquake. This paper describes the collection and analysis of the GPS data, and creates a comprehensive data
set for the displacement field. The abundant information
contained in the data set could be further inverted to deter-mine the rupture geometry and slip distribution associated with the Chi-Chi earthquake in future geophysical studies.
Acknowledgments. The authors are grateful to the Ministry of the Interior, R.O.C. forfinancially supporting this research and to the Bureau of Land Survey, R.O.C. for its contribution in the collec-tion of post-earthquake GPS data. We specially thank Dr. Kosuke
Heki and Dr. Zheng-Kang Shen for their important suggestions and constructive criticism.
References
Beutler, G., I. Bauersima, B. Botton, W. Gurtner, M. Rothacher, and T. Schildknecht, Accuracy and biases in the geodetic application of the Global Positioning System,Manuscr. Geod.,14, 28–35, 1989. Beutler, G., E. Brockmann, S. Fankhauser, W. Gurtner, J. Johnson, L.
Mervart, M. Rothacher, S. Schaer, T. Springer, and R. Weber,Bernese GPS Software Version 4.0, University of Berne, Berne, Switzerland, 1996. Beutler, M. Rothacher, S. Schaer, T. Springer, J. Kouba, and R. E. Neilan, The international GPS service (IGS): an interdisciplinary service in sup-port of earth sciences,Adv. Space Res.,23, 631–653, 1999.
Boucher, C., Z. Altamimi, M. Feissel, and P. Sillard, Results and analysis of the ITRF94, IERS Technical Note 20, 206 pp., Observatoire de Paris, Paris, France, 1996.
Cheng, S.-N, Y.-T. Yeh, M.-T. Hsu, and T.-C. Shih, Photo album of ten disastrous earthquakes in Taiwan, CWB-9-1999-002-9, Central Weather Bureau and the Institute of Earth Sciences, Academia Sinica, Taiwan, R.O.C., 289 pp., 1999.
DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein, Current plate motions, Geophys. J. Int.,101, 425–478, 1990.
Goad, C. C. and L. Goodman, A modified Hopfield tropospheric refraction correction model, Fall Annual Meeting of the Amer. Geophys. U., San Francisco, CA, U.S.A., Dec. 12–17, 1974.
Goad, C. C. and M. Yang, A new approach to precision airborne GPS posi-tioning for photogrammetry,Photogrammetric Eng. & Remote Sensing,
63, 1067–1077, 1997.
Goad, C. C., D. A. G.-Brzezinska, and M. Yang, Determination of high-precision GPS orbits using triple differencing technique,J. Geod.,70, 655–662, 1996.
Kuang, D., B. E. Schutz, and M. M. Watkins, On the structure of geometric positioning in GPS measurements,J. Geod.,71, 35–43, 1996. Ma, K.-F., C.-T. Lee, and Y.-B. Tsai, The Chi-Chi, Taiwan earthquake:
large surface displacements on an island thrust fault,Eos Trans. AGU,
80, 608–611, 1999.
Neilan, R. E., J. F. Zumberge, G. Beutler, and J. Kouba, The international GPS service: a global resource for GPS applications and research, in Proc. ION GPS-97, Sept. 16–19, Kansas City, MO, U.S.A., pp. 883–889, 1997.
Seno, T., S. Stein, and A. E. Gripp, A model for the motion of the Philippine Sea Plate consistent with NUVEL-1 and geological data,J. Geophys. Res.,98(B10), 17,941–17,948, 1993.
Yang, M. and C. F. Lo, Real-time kinematic GPS positioning for centimeter level ocean surface monitoring,Proc. Natl. Sci. Counc. ROC(A),24, 79– 85, 2000.
Yang, M., C.-L. Tseng, J.-Y. Yu, and K.-H. Chen, Adjustment and accuracy analysis of Taiwanfirst-order GPS control network,J. Cadastral Surv.,
18(2), 1–27, 1999 (in Chinese).
Yu, S.-B., H.-Y. Chen, and L.-C. Kuo, Velocityfield of GPS stations in the Taiwan area,Tectonophys.,274, 41–59, 1997.
Yu, S.-B., L.-C. Kuo, R. S. Punongbayan, and E. G. Ramos, GPS observation of crustal deformation in the Taiwan-Luzon region,Geophys. Res. Lett.,
26, 923–926, 1999.