The 1999 Chi-Chi earthquake (=7.6; 1999/9/20, 17:47:15.9 GMT; at 23.853N, 120.816E, depth of 7.5 km; Kao and Chen, 2000) has inflicted severe damages to the region, but at the same time, generated the best strong motion dataset available (Lee et al., 1999). Previously we used the strong motion dataset to study mainshock kinematics (Chi et al.,2001) by inverting velocity waveforms from 21 stations for 416 subfaults and 10 3-s time windows. In this study, we refine our model by adding additional stations and GPS data.
We found the major asperities are within a low-elevation triangular region just east of the surface rupture on the Chelungpu fault (Figure 19.1). Also, the Shungtung fault, parallel to and 10 km east of the surface rupture, seems to be a boundary separating regions with more strike-slip component to the west and more dip-slip to the east. The triangular region can also be divided into northern/southern halves. Most of large slips occurred 15 km north of the epicenter, under the hillier region of the asperities. Maximum slip occurred at the northern end of the surface rupture where it bends to the east and becomes very complicated.
We are currently testing several possible configurations for the fault planes to the north. Using the mapped surface rupture (CGS, 1999) as a constraint, our pure GPS inversions showed that several small additional fault planes are needed to better fit the complicated GPS displacements. Scalar moments derived from GPS inversion are a bit larger than the moments derived from pure seismic data, possibly relating to contamination in the GPS data from several 6 aftershocks which occurred a few days after the mainshock.
By adding 140 GPS displacement measurements to the seismic inversion, we have successfully suppressed the asperities at depth that were previously interpreted as artifacts. For shallow asperities the overall patterns are similar to GPS inversion results and our previous seismic-only inversions. By adding small faults to the north, we were able to get even better variance reductions. F-tests show those improvements are statistically significant. However, further testing of different fault configurations are needed, as the current asperity patterns on those additional faults are not very coherent. We also tested models with a decollement, but the fits did not improve.
One of the most consistent results in our study is the strain partitioning during the rupture. The slips east of Shungtung fault are mostly dip-slip while the slips west of it have more strike-slip components. Note, however, that there was no clear surface rupture on the Shungtung fault during this event and we did not add Shungtung fault in these particular inversion calculations. Strain partitioning has been proposed to explain many convergent boundary deformations, but this study is one of the first to document evidence of co-seismic strain partitioning using strong motion data. We are also processing several of the 6 aftershocks and hope to establish a kinematic model for the deformation during the whole Chi-Chi, Taiwan earthquake sequence.
CGS, Special Report on Chi-Chi 921 Earthquake. Published by the Central Geological Survey of Taiwan, Dec. 1999, 315 p. in Chinese.
Chi, Wu-Cheng, Douglas Dreger, and Anastasia Kaverina, Finite source modeling of the 1999 Taiwan (Chi-Chi) Earthquake derived from a dense strong motion network, Bull. Seism. Soc. Am., in press, 2001.
Kao, H and W. Chen, The Chi-Chi Earthquake Sequence of September 20, 1999 in Taiwan Seismotectonics of an Active, Out-of-Sequence Thrust, Science, 288, 346-349, 2000.
Lee, W. H.K., T.C. Shin, K.W.Kuo, and K.C. Chen, CWB Free-Field Strong-Motion Data from the 921 Chi-Chi Earthquake: Volume 1. Digital Acceleration Files on CD-ROM, Pre-Publication Version (December 6, 1999), Seismology Center, Central Weather Bureau, Taipei, Taiwan. 1999.