GPS measurement of postseismic deformation ... - Wiley Online Library

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. B11, 2520, doi:10.1029/2003JB002396, 2003

GPS measurement of postseismic deformation following the 1999 Chi-Chi, Taiwan, earthquake Shui-Beih Yu, Ya-Ju Hsu,1 Long-Chen Kuo, and Horng-Yue Chen Institute of Earth Sciences, Academia Sinica, Nankang, Taipei, Taiwan

Chih-Chung Liu Land Survey Bureau, Ministry of Interior, Taichung, Taiwan Received 9 January 2003; revised 29 July 2003; accepted 14 August 2003; published 12 November 2003.

[1] In the 15-month period (September 1999 to December 2000) after the 1999 Chi-

Chi, Taiwan, earthquake (Mw 7.6) about 80 campaign-surveyed GPS stations in central Taiwan were occupied up to seven times. Furthermore, seven continuous GPS stations were put into operation in the epicentral region mostly within three weeks after the main shock. GPS data from these near-field stations and some pre-existing far-field stations were utilized to study the postseismic deformation following the Chi-Chi earthquake. The postseismic displacements of the GPS stations were approximated by a combination of an exponential transient decay function with a relaxation time of 86 days and a postseismic linear rate change. Stations on the hanging wall displayed west-northwesterly horizontal displacements of up to 252 mm and uplift as large as 229 mm. Postseismic displacements are much larger to the south near the epicenter of Chi-Chi main shock than in the north where the maximum coseismic displacement occurred. Stations on the footwall show only small displacements. The postseismic GPS data were inverted to infer the deeper fault geometry and afterslip distribution based on a four-segment fault model with its shallow part constrained by coseismic fault geometry deduced from GPS and seismological studies. The optimal model requires the lower fault segment to be a horizontal plane at a depth of 10.4 km, consistent with a model based on GPS data taken during the first three months after the main shock and other geological or geophysical studies. The maximum afterslip (459 mm) occurs in the hypocentral region and in the northern part of the decollement. Slip of about 430 mm also occurs in the southern part of the decollement. The afterslip moment inferred from 15 months of GPS data is 4.7 1019 N m, approximately 2.3 times the seismic moment released by aftershocks. This implies that INDEX TERMS: 1242 Geodesy a major part of the postseismic deformation is aseismic. and Gravity: Seismic deformations (7205); 1243 Geodesy and Gravity: Space geodetic surveys; 3210 Mathematical Geophysics: Modeling; KEYWORDS: postseismic deformation, afterslip, GPS, Chi-Chi earthquake, Taiwan Citation: Yu, S.-B., Y.-J. Hsu, L.-C. Kuo, H.-Y. Chen, and C.-C. Liu, GPS measurement of postseismic deformation following the 1999 Chi-Chi, Taiwan, earthquake, J. Geophys. Res., 108(B11), 2520, doi:10.1029/2003JB002396, 2003.

1. Introduction [ 2 ] The devastating Chi-Chi earthquake (M w 7.6) occurred at 1:47 a.m. local time on 21 September (or at UTC 17:47 20 September) 1999. It was the largest inland earthquake in Taiwan in the twentieth century. The Central Weather Bureau (CWB) of Taiwan located the epicenter at 120.82E, 23.85N, near the town of Chi-Chi, with a focal 1 Now at Institute of Geophysics, National Central University, Chungli, Taiwan.

Copyright 2003 by the American Geophysical Union. 0148-0227/03/2003JB002396$09.00

ETG

depth of 8 km [Shin, 2000]. The focal mechanism indicated a N20E strike and a 30SE dip with an average rake of 85 [Chang et al., 2000]. This event produced a remarkable 100-km-long surface rupture (Figure 1), mainly along the previously mapped north-south-trending Chelungpu thrust fault (CLPF) [Meng, 1963; Chang, 1971]. At the northern end of the rupture near Fengyuan, the fault trace turned east with complex branching consisting of reverse faults, anticlines, and monoclines [Central Geological Survey, 1999; Lee et al., 2002]. Vertical surface offsets averaged about 2 m along the southern half of the CLPF and about 4 m along the northern half [Chen et al., 2001]. At the northwestern corner of the rupture, near Shihkang, the vertical offset was more than 8 m.

3-1

ETG

3-2

YU ET AL.: POSTSEISMIC DEFORMATION OF CHI-CHI EARTHQUAKE

Figure 1. Coseismic displacements associated with the 1999 Chi-Chi, Taiwan, earthquake [Yu et al., 2001]. Horizontal displacements are shown in vectors, and vertical displacements by vertical bars (uplift by a hollow bar and subsidence by a solid bar). The star denotes the epicenter of the main shock. CLPF indicates the Chelungpu fault.

The CLPF is one of three major east-dipping thrust faults in the western margin of the fold and thrust belt. Two other N-S-trending Quaternary faults in central Taiwan are the Shuangtung fault and Changhua fault to the east and west of the CLPF (Figure 1). [3] GPS measurements had been taken at many geodetic stations in central Taiwan by several institutions some 0.2– 2.7 years before the 1999 Chi-Chi earthquake. Most of these stations were resurveyed two days to three months after the main shock. These GPS data were utilized to estimate the coseismic displacements [Yu et al., 2001]. NW-NNW directed horizontal movements of 1.1– 9.1 m were observed on the hanging wall of the CLPF (Figure 1). There was a northwardly increasing trend in the magnitude and a clockwise rotation of about 50 in the directions of horizontal displacement vectors. In contrast, much smaller SE-ESE directed movements of 0.1 – 1.5 m with slight subsidence of 0.02 –0.26 m were detected on the footwall. The GPS observed horizontal and vertical offsets along the surface rupture were 2.4 –10.1 m and 1.2– 4.4 m, respectively [Yu et al., 2001].

[4] Seven continuous monitoring GPS stations were set up in the epicentral region mostly within three weeks after the Chi-Chi main shock and about 80 campaignsurveyed stations were resurveyed up to 7 times from September 1999 to December 2000. These GPS data provide an unprecedented opportunity to study the postseismic deformation following the 1999 Chi-Chi earthquake. Hsu et al. [2002] estimated the postseismic displacements during the first three months after the Chi-Chi earthquake. They inverted the available GPS data to find the afterslip distribution and deeper fault structure constrained by the shallow fault geometry based on the surface rupture and models fit to the coseismic deformation. Their afterslip distribution for the first three months shows a maximum slip of 250 mm in the hypocentral region at a depth of 7 – 12 km. Their study confirmed the existence of a nearly horizontal decollement at a depth of 8 – 12 km and showed that there was significant slip, about half of maximum, on this decollement. It also suggested that deep afterslip extended well south of the main shock rupture.


ETG

3-3

Table 1. Summary of Postseismic GPS Campaigns in Central Taiwan From September 1999 to December 2000a Campaign

Survey Period

Number of Station

Session, hours

Instrument

Survey Agency

1 2 3 4

23 Sep. – 6 Oct. 1999 18 Oct. – 4 Nov. 1999 13 – 21 Dec. 1999 22 – 24 Feb. 2000

60 285 90 45

6 – 12 3.5 14 3

Trimble 4000 SSE/SSI Trimble 4000 SSE/SSI Trimble 4000 SSE/SSI Trimble 4000 SSE

IESAS/CGS MOI LSB/IESAS LSB

5

27 – 31 Mar. 2000

42

6–8

LSB

6

25 – 28 Apr., 1 – 5 May 2000 10 – 20 Jul. 2000 4 – 15 Dec. 2000

82 55 88

8 8 7 – 13

Trimble 4000 SSE Leica SR9500 Trimble 4000 SSE Novatel PDCSTD Trimble 4000 SSE Trimble 4000 SSE/SSI

7 8

LSB LSB LSB/IESAS

a CGS, Central Geological Survey, Ministry of Economic Affairs; IESAS, Institute of Earth Sciences, Academia Sinica; LSB, Land Survey Bureau, Ministry of Interior; MOI, Ministry of Interior.

[5] This paper presents the postseismic temporal variations of baseline components for continuous GPS stations and the postseismic displacements estimated from campaign-surveyed GPS data collected in the first 15 months after the Chi-Chi main shock. The data employed in this paper were inverted to find the fault geometry and afterslip distribution following the similar strategy as Hsu et al. [2002]. The results were then compared with the coseismic slip as well as the afterslip for the first three months after the main shock.

2. GPS Data Acquisition and Processing [6] Within a 15-day time period after the Chi-Chi earthquake, about 60 pre-existing campaign-surveyed GPS stations in and around the epicenteral region (Figure 1) were resurveyed through a joint effort by the Institute of Earth Sciences, Academia Sinica (IESAS) and Central Geological Survey, Ministry of Economic Affairs (CGS) [Yu et al., 2001]. In mid-October 1999, the Satellite Survey Division, Ministry of Interior (MOI) conducted a more extensive postearthquake GPS survey including 285 first and second-order geodetic control stations in the Taiwan area [Yang et al., 2000]. [7] Subsequently in mid-December 1999, a GPS campaign was conducted in central Taiwan in a joint effort by the Land Survey Bureau (LSB), MOI and the IESAS. During this survey, 40 MOI stations having better sky visibility and most of the CGS/IESAS stations were each occupied once or twice. Then five more repeated surveys of part or all of these stations were conducted in the following year (2000) by the LSB and IESAS. Table 1 gives a summary of pertinent information for these postseismic GPS campaigns. [8] In the GPS campaigns conducted by the IESAS and LSB, each station was measured for one or two sessions with a dual-frequency geodetic receiver. A session was composed of 6- to 14-hour observations. Dual-frequency receivers were also used in the MOI post-earthquake GPS survey and each station was occupied for one or two 3.5-hour sessions. [9] Furthermore, in addition to two pre-existing permanent GPS continuous stations (SUNM and SANI), seven temporary continuously recording stations (CPUL, KZON, I007, S016, 5936, WUFN, and PINT) were installed in the epicentral region within 3 weeks after the Chi-Chi main shock. Except for CPUL, these stations and a new perma-

nent station (S167) established in November 1999 were kept in operation until the end of 2000. The GPS data from these 9 near-field continuous stations and those from 34 permanent far-field continuous stations as well as 80 campaignsurveyed stations in the Taiwan area were utilized to study the postseismic deformation following the Chi-Chi earthquake. [10] The campaign-surveyed and continuous GPS data were processed with the Bernese v.4.2 software [Rothacher and Mervart, 1996] using standard procedures. We used the double-differenced ionosphere-free carrier phase observations (L3) as the basic observables. The International GPS Service (IGS) final combined orbits were employed and fixed during post-processing. The residual tropospheric zenith delay, which is the difference between the actual zenith delay and that calculated from a standard atmosphere model [Saastamoinen, 1973], was estimated every 2 hours per station simultaneously with the station coordinates by least squares adjustment. The campaign-surveyed data were processed session by session, while the continuous data were processed daily. Daily solutions from these two types of data were further combined to form the integrated daily solutions with SINEX (Software Independent Exchange) format for using in this study. The International Terrestrial Reference Frame 1997 (ITRF97) was adopted by taking the ITRF97 coordinates and velocities of an IGS core site, TSKB (Tsukuba, Japan) as a reference point. Data from several IGS sites in the AsiaPacific region including TSKB were used in our routine GPS data processing.

3. Postseismic Displacement 3.1. Time Series of Continuous GPS Stations [11] Continuous GPS data provide important information for the temporal evolution of postseismic deformation. The daily Cartesian XYZ coordinates for each station were extracted from SINEX files to form time series of geodetic positions. The XYZ time series were then rotated to a topocentric north, east, and up (NEU) coordinate system. The NEU position data were cleaned and modeled independently in each of the three coordinate directions, since correlations between components are very small [Zhang, 1996]. To find the postseismic deformation, the secular crustal motion of the station during the 15-month period was subtracted out using the 1993 – 1999 interseismic velocity field estimated by

ETG

3-4


Figure 2. Stack of power spectral densities in the east, north, and up directions for data from all Taiwan continuous GPS stations used in this study. Best-fit straight lines are estimated in log-log space. S. B. Yu et al. (Interseismic crustal deformation in the Taiwan plate boundary zone (1993 – 1999), unpublished manuscript, 2003, hereinafter referred to as Yu et al., unpublished manuscript, 2003). [12] The observed postseismic displacement y(t) of continuous GPS station in each direction after the Chi-Chi earthquake can be modeled as m X yðti Þ ¼ a þ b ti Teq þ Cj H ti Tcj j¼1

þ d exp ti Teq =t þ ui

ð1Þ

where ti for i = 1, . . .N are the daily solution epochs in units of years, and H is the Heaviside step function [Nikolaidis, 2002]. Since only 15-month data were used in this study, we did not attempt to model the annual or semi-annual periodic motions. The first term, a, is the station position and the second term accounts for the post-Chi-Chi linear rate change, b, as the secular crustal motion has been subtracted out. Teq is the occurrence epoch of Chi-Chi earthquake. The third term corrects for any number (m) of offsets with magnitude Cj at epoch Tcj. The offsets might be due to coseismic movements of large aftershocks or antenna/receiver changes. An exponential decay function with magnitude d and relaxation time t is employed to model the transient postseismic motion following the Chi-Chi earthquake. The relaxation time, 86 days, from best-fitting of 7 months postseismic GPS data by Yu et al. [2001] was adopted in Equation (1) for all stations. The last term is the measurement errors vi

initially assumed to be statistically uncorrelated from one another (white noise). [13] GPS measurement noise can be approximated as a power law process, or one with time domain behavior that has a power spectrum of the form k Pðf Þ ¼ P0 f =f 0

ð2Þ

where F is the temporal frequency, P0 and f0 are normalizing constant, and k is the spectral index [Mandelbrot and Van Ness, 1968], which is the slope of the spectra in log-log space. [14] Some recent studies showed that continuous GPS data are best described as a combination of white (k = 0) and flicker noise (k = 1) or as a fractional white noise (1 < k < 1) [e.g., Zhang et al., 1997; Mao et al., 1999; Nikolaidis, 2002]. Figure 2 shows the stacked power spectra of postfit residuals in the east, north and up directions for all continuous GPS stations used in this study. The slopes of the spectra (k) are 0.37 ± 0.04, 0.30 ± 0.04, and 0.27 ± 0.05, respectively for the E, N, U directions. These spectral indices may be biased low because of the high-frequency components and a noise peak at 1/(14 days). If we only used data with frequency less than 1/(15 days), or about 24.3 cycle/year, the estimated slopes (k) become 0.62 ± 0.01(E), 0.43 ± 0.01(N), and 0.46 ± 0.01(U), respectively. Thus the previous conclusion regarding the noise characteristics of GPS measurements is confirmed here again. [15] To obtain the final parameter estimates, several iterations using the weighted least squares algorithm were

YU ET AL.: POSTSEISMIC DEFORMATION OF CHI-CHI EARTHQUAKE Table 2. Noise Amplitude of the Model With White Noise Plus Flicker Noisea East

North

Up

Station

aw

b1

aw

b1

aw

b1

5936 BANC CHEN CHIA CHNL CHYN CK01 FCWS FIVE FLNM GAIS HENC HOKN HSIN HUAL I007 ILAN KDNM KULN KZON LIUC NCTU PINT PKGM S011 S012 S016 S058 S101 S102 S103 S104 S105 S106 S167 S23R SANI SUAO SUN1 TMLM TSHO WUFN YUSN

1 2 2 2 2 1 1 2 2 2 2 3 2 2 2 1 2 3 1 2 3 2 2 1 2 2 2 3 2 3 1 2 2 2 2 2 2 2 2 2 2 1 2

9 4 6 5 3 7 4 5 4 6 10 9 6 3 6 7 7 6 7 5 6 3 5 4 5 4 6 4 4 6 6 5 5 5 5 5 2 5 5 5 3 7 6

1 2 2 1 2 1 1 2 2 2 2 3 1 2 2 1 2 3 2 1 2 2 1 1 2 2 2 2 2 2 1 2 2 2 2 2 2 2 1 2 2 1 1

6 3 3 4 7 4 4 3 6 3 11 6 6 4 5 5 4 6 5 4 7 3 4 4 6 5 3 4 3 7 4 4 3 4 4 3 3 4 5 4 4 5 6

7 10 12 6 6 4 5 6 8 8 11 11 5 10 11 6 10 10 5 6 14 5 5 4 5 5 8 9 5 8 5 8 8 5 8 6 10 12 8 7 5 5 5

16 19 27 21 12 4 5 12 32 11 33 36 7 19 22 9 34 16 10 3 15 10 6 13 11 6 10 18 16 25 6 13 16 8 6 12 11 18 10 10 9 14 21

a aw (mm) and b1 (mm/yr0.25) are the magnitudes of white noise and flicker noise, respectively.

executed. The time series were firstly modeled (assuming white noise) using Equation (1) to create postfit residuals for identifying outliers. Next, the cleaned data were remodeled to create new postfit residuals that can be used for the noise analysis. Since previous studies have demonstrated that a combination of white and flicker noise is dominant in GPS position data, the maximum likelihood estimator (MLE) was utilized to estimate the amplitudes of white and flicker noise present in each time series of postfit residuals [Langbein and Johnson, 1997; Zhang et al., 1997; Williams, 2003]. Table 2 gives the estimated white noise and flicker noise amplitudes for the time series of 43 continuous GPS stations. The full covariance matrix is the sum of the temporally uncorrelated (white) and correlated (flicker) noise covariance matrices. Finally, the cleaned time series were modeled again using the full data covariance matrices, so that the parameters can be estimated with more realistic uncertainties.

ETG

3-5

[16] Figure 3 shows an example modeled time series for the east, north, and up components of station YUSN (near the highest peak in Taiwan, Yushan). The model includes one coseismic offset associated with a moderate aftershock (Mw 5.7) occurred on July 28, 2000, a linear rate change and an exponential decay following the Chi-Chi earthquake. Final postfit residuals of east, north, and up components are plotted in Figure 4. The root-mean squares (rms) values of residuals are 2.6 mm (E), 1.8 mm (N), and 7.6 mm (U) respectively. [17] Figures 5 – 7 show examples of postseismic temporal variations in the east, north, and up directions for 12 continuous stations with respect to Paisha, Penghu (S01R) (see index map in Figure 1) from September 1999 to December 2000. Paisha is situated on the relatively stable Chinese continental margin and did not show significant postseismic displacement. Thus it can be taken as a good reference for this study. The coseismic offsets caused by two major aftershocks occurred on October 22, 1999 (Mw 5.9) and June 10, 2000 (Mw 6.1) were removed in these figures. In general, the curves (solid lines) combine the linear rate change and exponential decay with a relaxation time of 86 days fit the data quite well. On the other hand, if we use a logarithmic function instead of an exponential decay model, the data could be fitted almost equally well. However, in case of lacking GPS data at early time after the earthquake, the logarithmic function may overestimate the postseismic displacements. Table 3 gives the estimated total postseismic displacements that occurred between September 1999 and December 2000 for 43 continuous GPS stations in the Taiwan area. [18] Significant westward to west-northwestward postseismic displacements of 132– 212 mm were observed at the near-field hanging wall stations (I007, 5936, S167, S016) during the 15-month period after the Chi-Chi main shock (Figure 5). I007 was uplifted 0.85 m coseismically and showed an additional postseismic uplift of about 160 mm. Conversely, S016 also registered an uplift of 1.01 m associated with the main shock, but subsided more than 79 mm during the postseismic recording period. WUFN is also situated on the hanging wall side, but only a few tens meters from the 1999 Chi-Chi surface rupture at Kuangfu Junior High School in Wufeng. It shows a southwestward postseismic displacement of 30 mm (Figure 6). PINT and KZON are located on the footwall of the Chelungpu fault and show 10– 27 mm postseismic displacements in the directions of 229– 248. PKGM, in southwestern Taiwan, shows a 18 mm postseismic displacement in the direction of 98. [19] Although the stations in eastern Taiwan (FLNM, HUAL, S058 and S105) are far from the surface rupture (95 – 110 km), they still show notable postseismic displacements of 28 – 55 mm in a west to west-northwesterly direction (Figure 7). This far-field deformation implies a deep source of postseismic slip. The postseismic displacements of other continuous stations in northern and southern Taiwan are mostly less than 20 mm (Table 3). 3.2. Campaign-Surveyed Stations [20] To find the postseismic deformation, the secular crustal motions of campaign-surveyed stations were sub-

ETG

3-6


Figure 3. A modeled time series for the east, north, and up components of station YUSN. The model includes one coseismic offset (indicated by a dash-dotted line) associated with a moderate aftershock (Mw 5.7) occurred on 28 July 2000, a linear rate change and an exponential decay following the Chi-Chi earthquake. Outliers excluded are shown by open circles.

tracted out of each postseismic epoch prior to further analysis, the same as was done for the continuous stations. Velocities of the stations that were not analyzed by Yu et al. (unpublished manuscript, 2003) were determined by bilinear interpolation that fits a bilinear surface through existing data points. The value of an interpolated point is a combination of the four closest points. [21] The postseismic displacements of campaign-surveyed stations were also estimated by using the model described in Equation (1), except no attempt to correct for coseismic offsets. The number of data points for each time series of campaign-surveyed stations was not enough for noise analysis. Thus we used the formal errors and multiplied the uncertainties by scale factors empirically. The scale factors for the east, north, and up components were 10, 10, and 5 respectively. [22] Figure 8 shows examples of the time series for north, east, and up components of eight campaign-surveyed stations. In general, the data were fitted quite well with a combination of exponential decay functions and linear rate changes. Table 4 gives the total postseismic displacements during the 15-month period after the Chi-Chi earthquake, September 1999 to December 2000, for 80 campaignsurveyed stations in central Taiwan. [23] Figure 9 shows the postseismic horizontal and vertical displacements for continuous and campaign-surveyed stations in central Taiwan during the 15-month period after Chi-Chi earthquake and their 95% confidence ellipses for

horizontal components. Most of the stations on the footwall of the Chelungpu fault showed less than a 40 mm horizontal displacement in the east to southeasterly direction and a slight or insignificant subsidence. Several stations (e.g., AF11, AF13, AF14, AF16, G039, G041, and G045) near the surface rupture and first surveyed within a few days after 21 September 1999, displayed 6 – 50 mm horizontal displacements in the southwest to westerly direction (211– 264). The stations to the southwest of Changhua showed a moderate subsidence of up to 88 mm. More than 50% of the vertical postseismic displacements for campaign-surveyed stations are marginal or not significant in the 95% confidence level. [24] In contrast, the hanging wall stations showed westnorthwesterly directed horizontal displacements of up to 252 mm. Postseismic displacements are much larger on the southern part of the hanging wall near the epicenter of ChiChi main shock than in the north where the maximum coseismic displacements occurred. The stations very close to the rupture showed small or insignificant displacements. This suggests that postseismic slip took place on the deeper portion of fault plane. Note also that there were significant displacements of up to 90 mm in eastern Taiwan about 100 km east of the Chi-Chi rupture. A remarkable uplift as large as 229 mm was observed in the vicinity of the main shock epicenter. The uplifted area extended eastward to Wushe about 50 km from the surface rupture. A narrow zone of slight subsidence (up to 79 mm) was observed to the


ETG

3-7

Figure 4. Postfit residuals for the east, north, and up components of YUSN. The root-mean squares (RMS) values of residuals are 2.6 mm (E), 1.8 mm (N), and 7.6 mm (U) respectively. The dash-dotted lines show the earthquake epoch of a moderate aftershock (Mw 5.7) occurred on July 28, 2000. southeast of Fengyuan, near the northern end of the northsouth- trending Chelungpu fault.

4. Fault Geometry and Afterslip Distribution [25] The fault geometry and afterslip distribution during 15 months period after the main shock were inferred by

weighted least squares using the elastic half-space model as was done by Hsu et al. [2002] for the first three months after the main shock. However, there are some differences between these two studies in the observed data and preprocessing. In this paper, we used a combination of a linear rate change and an exponential decay function in equation (1) to estimate postseismic deformation at each station.

Figure 5. Temporal variations found in the north, east, and up components for four continuously recording stations (I007, 5936, S167, and S016) located on the hanging wall of the CLPF. The curves (solid lines) combine the linear rate change and exponential decay with a relaxation time of 86 days fit the data quite well. The dash-dotted lines indicate the earthquake epochs of two major aftershocks occurred on 22 October 1999 (Mw 5.9) and 10 June 2000 (Mw 6.1).

ETG

3-8


Figure 6. Temporal variations found in the north, east, and up components for three continuously recording stations (PINT, KZON, and PKGM) on the footwall and a near-fault station (WUFN) on the hanging wall of the CLPF. Solid lines show the linear rate change and exponential fits with a relaxation time of 86 days. The dash-dotted lines indicate the earthquake epochs of two major aftershocks occurred on 22 October 1999 (Mw 5.9) and 10 June 2000 (Mw 6.1). Further, we used a four-segment fault model and inverted both the horizontal and vertical postseismic displacement data to find the optimal fault geometry and afterslip distribution. We minimized the following functional: F ðs; b; mÞ ¼k

1=2

ðGðmÞs dÞ k2 þ b2 k r2 s k2

P1/2 where is the inverse square root of the data covariance matrix, G(m) is a matrix of elastic half-space Green’s functions for rectangular faults [Okada, 1985],

which depend on the fault geometry parameters m (i.e., strike, dip, length, depth, horizontal position), s is slip, d is the observed postseismic displacements and the smoothing operator r2 is the finite difference approximation of the Laplacian operator. b serves as a measure of the roughness of the slip distribution [e.g., Harris and Segall, 1987] and gives the relative weight between the model roughness and data misfit. b can be determined objectively by the statistical method of cross-validation [Matthews and Segall, 1993]. We used a four-segment fault model (K. M. Johnson and P. Segall, Imaging the

Figure 7. Temporal variations found in the north, east, and up components for four continuously recording stations (HUAL, S105, FLNM, and S058) in eastern Taiwan. Solid lines show the linear rate change and exponential fits with a relaxation time of 86 days. The dash-dotted lines indicate the earthquake epochs of two major aftershocks occurred on 22 October 1999 (Mw 5.9) and 10 June 2000 (Mw 6.1).

ETG


3-9

Table 3. Postseismic Displacements of Continuous GPS Stations in the Taiwan Area Between September 1999 and December 2000a Station

Latitude, N

Longitude, E

DE, mm

DN, mm

DU, mm

S, mm

Azis, deg

a, mm

b, mm

Azia, deg

5936 BANC CHEN CHIA CHNL CHYN CK01 FCWS FIVE FLNM GAIS HENC HOKN HSIN HUAL I007 ILAN KDNM KULN KZON LIUC NCTU PINT PKGM S011 S012 S016 S058 S101 S102 S103 S104 S105 S106 S167 S23R SANI SUAO SUN1 TMLM TSHO WUFN YUSN

24.015 24.999 23.099 23.497 23.379 23.394 22.974 24.853 25.073 23.748 23.082 22.006 23.188 24.830 23.976 23.758 24.766 21.951 23.332 23.948 22.343 24.791 23.758 23.580 23.206 23.061 24.182 23.321 25.041 22.038 23.566 22.823 22.954 23.051 23.957 22.647 24.416 24.596 23.884 22.618 22.905 24.045 23.489

121.120 121.434 121.370 120.427 120.557 120.284 120.213 121.241 121.773 121.448 120.583 120.738 120.129 121.006 121.608 120.768 121.748 120.780 120.501 120.690 120.371 120.988 120.634 120.298 120.333 120.483 120.795 121.451 121.606 121.558 120.468 121.186 121.109 120.329 120.927 120.602 120.760 121.858 120.902 121.004 120.347 120.691 120.954

173.0 1.2 32.0 18.4 28.3 1.7 6.1 0.2 5.3 48.2 18.7 21.0 6.8 0.1 41.9 210.9 0.6 26.5 19.3 25.1 22.7 1.8 7.5 18.2 4.2 19.2 131.8 53.5 4.1 10.8 1.2 20.9 22.0 16.2 133.7 15.7 14.1 7.3 135.9 11.6 7.0 28.5 94.8

44.3 16.9 16.8 3.4 18.8 1.9 3.2 17.3 12.3 25.3 15.5 9.1 6.7 11.2 8.8 19.0 18.9 12.5 2.5 9.9 4.1 5.1 6.4 2.5 2.8 0.7 12.9 14.1 19.2 19.2 13.3 12.6 17.3 1.3 72.4 8.2 2.5 23.7 93.7 14.3 3.0 10.1 49.6

38.9 ± 6.7 2.5 ± 7.3 8.2 ± 9.8 19.7 ± 7.3 16.9 ± 4.5 14.7 ± 1.7 3.6 ± 2.7 7.3 ± 4.5 7.2 ± 11.6 15.6 ± 5.1 3.2 ± 12.7 5.9 ± 12.4 8.5 ± 2.7 1.1 ± 7.0 9.5 ± 8.2 159.7 ± 4.3 5.6 ± 11.6 13.2 ± 6.4 6.5 ± 4.1 22.9 ± 2.0 1.2 ± 7.6 12.0 ± 4.0 10.8 ± 3.1 15.6 ± 4.6 11.2 ± 4.2 5.6 ± 2.8 78.5 ± 4.7 9.9 ± 7.0 10.0 ± 5.7 9.7 ± 13.0 17.4 ± 3.0 0.6 ± 5.1 3.2 ± 5.8 9.7 ± 3.2 78.3 ± 6.2 9.4 ± 4.6 14.5 ± 9.7 0.1 ± 7.2 18.6 ± 17.6 0.0 ± 4.6 13.8 ± 3.5 2.0 ± 5.3 30.3 ± 27.4

178.6 16.9 36.1 18.7 34.0 2.6 6.9 17.3 13.4 54.4 24.2 22.9 9.6 11.2 42.8 211.7 18.9 29.3 19.5 27.0 23.1 5.4 9.9 18.3 5.0 19.2 132.4 55.4 19.7 22.0 13.3 24.4 28.0 16.3 152.1 17.7 14.3 24.9 165.1 18.4 7.6 30.3 107.0

284 356 298 281 304 138 298 1 337 298 310 293 315 1 282 265 358 295 277 248 260 19 229 98 236 268 264 285 348 331 5 301 308 275 298 298 100 343 305 321 293 251 298

3.4 1.5 2.1 1.8 2.5 2.3 1.6 1.8 2.2 2.5 3.9 3.1 2.1 1.5 2.1 2.5 2.4 2.3 2.4 1.9 2.5 1.3 2.2 1.4 2.2 2.0 2.3 1.7 1.5 3.6 2.3 1.8 1.8 1.9 3.2 1.9 2.4 1.8 7.4 2.1 1.5 2.4 8.2

2.3 1.2 1.2 1.4 1.2 1.4 1.6 1.2 1.6 1.4 3.6 2.2 2.0 1.2 1.8 1.9 1.5 2.3 2.0 1.5 2.4 1.3 1.6 1.4 1.9 1.7 1.3 1.6 1.2 3.4 1.6 1.5 1.1 1.6 2.7 1.2 1.8 1.5 6.9 1.7 1.2 1.8 7.6

52 48 52 52 52 52 51 52 49 51 51 47 52 49 52 52 50 50 52 52 46 52 52 52 52 52 51 51 52 50 52 51 51 52 52 51 49 51 50 51 51 52 52

a DE, DN and DU, east, north and up components of the postseismic displacements, respectively; S, postseismic horizontal displacement; Azis, azimuth of S; a and b, semimajor and semiminor axes for the error ellipse of S, Azia, azimuth of a. The uncertainties quoted are one standard deviation.

decollement, ramp and northern tear-thrust of the 1999 Chi-Chi Taiwan earthquake using coseismic GPS displacements, submitted to Tectonophysics, 2003, hereinafter referred to as Johnson and Segall, submitted manuscript, 2003) as initial fault geometry (Figure 10). The two segments on the northern end (Faults 2 and 3 in Figure 10) represent the E-W trending rupture zone of the Chi-Chi surface ruptures and their geometries were fixed as Johnson and Segall (submitted manuscript 2003). The main fault represented by upper and lower segments (Faults 1 and 4 in Figure 10), extended 110 km in length as well as 24 and 40 km in width, respectively. Since the aftershock seismicity in the shallow part of the CLPF was less than that to the east of Chi-Chi epicenter, we looked for possible deeper fault structure. The upper segment 1 was fixed to the coseismic fault geometry with a N3.3E striking plane dipping 26E (Johnson and Segall, submitted manuscript, 2003). We used a grid search and estimated the optimal slip distribution by only

varying the dip angle (q) and the hinge depth (D) of the lower segment 4 (Figure 10). In each model, the fault patches were constant and b was fixed to the cross validation result for the fault geometry with D = 10 km and q = 0. [26] To assess how well the model was constrained by the GPS data, we used the bootstrap method to estimate the confidence intervals of fault parameters. The initial fault parameters were estimated from the prior grid search and the dip angle was constrained to positive. We re-sampled the actual data set d, with its N stations, to generate 1000 synthetic data sets d1, d2, d3 . . .d1000, each with the N stations. Some stations were chosen several times, others not at all. We subjected these data sets to the same estimation procedure as performed on the actual data. Finally, we generated the distribution of fault parameters given by those solutions. The global optimum solutions of the hinge depth (D) was 10.4 km, with 95% confidence interval 6.5 13.1 km, and the dip

ETG

3 - 10


Figure 8. Temporal variations found in the north, east, and up components for (a) four campaignsurveyed stations (HTZS, AF30, G044, and G099) on the hanging wall of the CLPF, (b) four campaignsurveyed stations (AF16, AF11, A247, and G041) on the footwall of the CLPF. Solid lines show the linear rate change and exponential fits with a relaxation time of 86 days.

angle (q) was 0, with 95% confidence interval 0 0. The optimal fault geometry is consistent with the geological balanced cross sections [Suppe, 1980], model interpretations from geodetic data [Loevenbruck et al., 2001; Dominguez et al., 2003], aftershock distribution and microearthquake seismicity [Hirata et al., 2000, Chen et al., 2002, Carena et al., 2002] as well as seismic reflection results [Wang et al., 2002], all of which suggested a nearly horizontal detachment at a depth of 8– 12 km. [27] A comparison between the observed and model predicted postseismic displacements is shown in Figures 11a and 11b for the horizontal and vertical components, respectively. The surface projection of the optimal fault planes is shown by the dashed rectangles. Our model can explain about 72% of the total variance and the root mean square misfits for horizontal and vertical components are 25 and 43 mm, respectively. The majority of model displacements

are under-estimated, this may be attributed to inelastic deformation. In general, the model fits the observed horizontal displacements quite well, except for some nearfault stations (e.g., AF25, M400). However, the misfit in vertical component is large. Considering the relatively small weights due to larger standard errors of vertical components compared to the horizontal components, the poorer fit there is acceptable. [28] Figure 12 shows the inverted afterslip distribution during the 15-month period using the optimal fault geometry. Afterslip on the northern two segments are small with a magnitude of only about 50 mm, in contrast to 10 m of coseismic slip in the same area [Johnson et al., 2001]. The maximum afterslip of 459 mm occurred in the hypocentral region and in the northern part of the decollement. Moreover, afterslip of about 430 mm occurred on the southern part of the decollement. The magnitude of the afterslip

ETG


3 - 11

Table 4. Postseismic Displacements of CampaignSurveyed Stations in Central Taiwan Between September 1999 and December 2000a Station

Latitude, N

Longitude, E

DE, mm

DN, mm

DU, mm

S, mm

Azis, deg

a, mm

b, mm

Azia, deg

6389 A247 AF01 AF04 AF05 AF09 AF11 AF13 AF14 AF15 AF16 AF17 AF18 AF21 AF22 AF24 AF25 AF26 AF30 CPUL G039 G040 G041 G042 G044 G045 G090 G091 G098 G099 G102 G103 G104 HTZS M043 M049 M075 M093 M312 M314 M315 M324 M326 M330 M360 M365 M395 M398 M400 M402 M408 M426 M428 M436 M453 M493 M501 M507 M509 M805 M808 M904 M906 M907 M909 M910 M911 M916 M918 M961 MERK S027 S030

24.154 24.021 23.770 23.873 23.900 24.039 23.896 23.948 24.017 24.096 24.038 24.158 24.217 24.218 23.962 24.037 24.125 24.223 23.885 23.929 23.822 23.716 23.765 23.751 23.834 23.838 24.313 24.381 24.293 24.173 24.301 24.262 24.244 23.976 24.131 23.980 24.281 23.695 24.173 24.105 24.301 24.222 24.256 24.227 23.868 23.986 24.053 23.914 23.680 24.104 23.779 24.049 24.329 24.063 24.154 23.775 23.953 23.726 23.816 24.173 24.112 24.300 24.258 24.069 24.011 23.957 23.894 24.333 23.954 23.982 23.798 23.486 24.283

121.278 120.400 120.556 120.527 120.578 120.506 120.677 120.690 120.636 120.640 120.661 120.624 120.518 120.564 120.739 120.718 120.744 120.643 120.760 120.627 120.680 120.648 120.680 120.752 120.744 120.655 120.553 120.580 120.901 120.887 120.780 120.710 120.780 120.975 120.851 120.438 120.841 120.464 120.825 120.742 120.759 120.735 120.567 120.799 120.936 120.620 120.578 120.704 120.754 120.904 120.843 120.918 120.613 120.705 120.475 120.720 120.901 120.857 120.892 120.673 120.678 120.598 120.514 120.625 120.433 120.406 120.417 120.704 120.499 121.126 120.302 120.884 121.020

95.8 15.2 18.4 12.7 7.6 10.3 21.0 3.0 10.0 12.2 17.6 2.3 14.5 14.9 156.3 61.9 47.9 6.6 238.2 4.1 32.2 4.3 49.0 206.5 189.5 34.9 12.7 33.0 54.0 127.0 11.4 0.5 58.6 177.1 91.3 5.8 112.4 7.0 95.6 41.3 0.8 53.4 20.9 83.5 158.8 18.0 9.6 13.9 183.9 142.6 191.8 148.6 16.2 48.2 14.8 86.1 109.5 162.0 168.6 4.0 12.8 17.9 20.3 20.8 2.6 12.5 4.5 9.4 12.2 168.2 4.7 92.2 28.6

49.8 1.4 18.3 25.5 9.8 15.3 3.8 5.0 9.0 12.1 12.3 0.2 9.2 18.3 57.6 52.3 0.9 1.5 83.4 10.9 22.6 5.8 9.5 27.6 102.6 3.7 8.1 15.6 6.6 36.2 24.2 2.5 7.9 57.7 21.8 11.6 20.5 22.0 11.9 35.1 7.9 21.4 17.1 34.6 53.3 0.7 6.7 42.4 71.8 51.0 57.1 72.6 9.4 1.5 3.9 9.6 91.5 48.9 71.5 15.4 3.9 18.1 24.2 5.5 1.0 10.3 3.9 22.8 19.0 13.2 9.7 91.5 78.4

106.5 ± 30.7 6.7 ± 16.3 9.4 ± 14.9 87.8 ± 23.8 29.3 ± 14.1 19.8 ± 21.1 10.9 ± 12.5 24.2 ± 17.8 17.2 ± 14.8 3.9 ± 16.8 8.4 ± 13.0 36.2 ± 22.0 4.5 ± 22.3 18.7 ± 17.2 61.3 ± 13.8 6.3 ± 14.2 70.2 ± 14.7 3.0 ± 15.3 104.2 ± 12.5 7.1 ± 10.9 33.3 ± 16.0 8.5 ± 30.1 5.2 ± 14.6 183.4 ± 19.3 23.2 ± 13.6 9.9 ± 16.6 1.8 ± 18.3 18.0 ± 19.5 27.4 ± 23.7 18.0 ± 16.2 23.4 ± 16.4 7.2 ± 16.3 112.9 ± 25.9 121.7 ± 15.9 26.6 ± 17.2 28.0 ± 18.7 143.9 ± 36.7 36.3 ± 17.1 38.2 ± 26.1 61.5 ± 20.3 4.3 ± 26.6 41.1 ± 18.1 6.8 ± 19.1 16.7 ± 18.2 28.7 ± 85.4 34.2 ± 20.6 8.0 ± 18.9 65.4 ± 31.2 62.5 ± 31.0 137.9 ± 30.1 228.7 ± 35.6 110.5 ± 29.7 31.4 ± 21.3 60.4 ± 20.5 3.3 ± 18.4 60.4 ± 29.1 131.2 ± 23.2 82.9 ± 24.4 152.8 ± 25.9 22.6 ± 15.3 44.6 ± 15.8 9.9 ± 19.0 38.4 ± 17.0 50.9 ± 16.7 23.0 ± 24.4 30.1 ± 22.3 30.7 ± 19.3 15.2 ± 16.4 73.3 ± 19.4 56.9 ± 31.2 0.4 ± 16.3 16.3 ± 34.6 82.1 ± 33.0

107.9 15.2 26.0 28.5 12.4 18.5 21.3 5.9 13.5 17.2 21.5 2.3 17.1 23.6 166.5 81.0 47.9 6.8 252.4 11.6 39.3 7.3 49.9 208.3 215.5 35.0 15.1 36.5 54.4 132.1 26.8 2.6 59.1 186.3 93.9 12.9 114.3 23.1 96.3 54.2 7.9 57.5 27.0 90.3 167.5 18.1 11.7 44.6 197.4 151.5 200.1 165.4 18.8 48.3 15.3 86.6 142.7 169.2 183.1 15.9 13.4 25.5 31.6 21.5 2.8 16.2 5.9 24.7 22.6 168.7 10.8 129.9 83.4

297 85 135 154 142 146 260 211 228 225 235 86 123 141 290 310 271 103 289 159 235 323 259 278 298 264 122 115 263 286 25 349 278 288 283 154 260 198 263 310 175 248 129 292 289 268 125 198 291 290 287 296 120 272 105 264 310 287 293 195 253 135 140 285 290 130 131 158 213 274 206 315 340

9.4 5.2 4.8 8.8 4.8 6.5 4.4 5.8 5.0 5.5 4.5 6.4 7.4 6.2 4.6 4.9 4.9 5.1 4.4 3.9 5.3 9.5 4.8 6.6 4.6 5.3 5.9 6.1 6.9 5.4 5.7 5.5 7.9 5.3 5.9 6.1 13.1 5.5 8.4 6.1 9.7 5.7 6.4 6.0 15.7 6.4 5.9 9.4 10.3 8.4 11.6 9.5 6.6 7.2 5.8 9.3 7.4 7.7 8.7 5.2 5.4 6.3 6.0 5.3 7.3 7.3 5.7 5.7 6.0 9.9 5.2 12.6 9.9

7.5 4.4 4.3 7.1 4.3 6.2 3.9 5.0 4.4 5.1 3.9 6.2 6.9 5.1 4.4 4.3 4.4 4.6 3.9 3.3 4.6 8.0 4.3 5.5 4.0 4.8 5.4 5.8 5.7 4.6 4.9 5.1 7.6 4.4 4.8 5.0 8.8 4.9 6.6 5.6 7.4 5.2 5.8 5.1 11.8 5.8 5.3 7.4 9.8 7.0 9.4 7.7 6.4 6.1 5.2 7.4 5.5 6.6 6.8 4.6 4.7 5.8 5.1 4.9 6.4 6.4 5.1 5.0 5.5 8.3 4.4 9.7 8.6

15 15 34 44 63 103 60 70 59 67 56 100 89 58 53 58 66 52 60 55 63 74 60 65 63 61 58 40 69 58 57 51 36 77 73 83 85 58 67 77 75 57 68 72 100 90 102 81 162 87 99 92 47 98 100 96 77 84 80 69 71 61 68 63 90 96 65 66 73 91 65 108 73

ETG

3 - 12


Table 4. (continued) Station

Latitude, N

Longitude, E

DE, mm

DN, mm

DU, mm

S, mm

Azis, deg

a, mm

b, mm

Azia, deg

S038 S049 S164 S165 TECS WNTS

23.756 23.888 24.044 24.044 24.358 24.140

121.135 121.535 120.692 120.689 120.647 120.576

234.6 83.3 30.5 28.6 14.9 10.6

81.2 35.3 19.6 2.2 27.3 8.9

60.7 ± 55.9 37.0 ± 36.7 12.0 ± 38.9 38.1 ± 40.0 19.1 ± 14.3 14.6 ± 13.0

248.2 90.4 36.2 28.7 31.1 13.8

289 293 237 266 151 130

26.1 10.0 12.2 12.7 4.9 4.4

23.2 8.4 11.0 11.4 4.3 3.8

148 16 178 171 44 33

a DE, DN and DU, east, north and up components of the postseismic displacements, respectively; S, postseismic horizontal displacement; Azis, azimuth of S; a and b, semimajor and semiminor axes for the error ellipse of S, Azia, azimuth of a. The uncertainties quoted are one standard deviation.

over 15 months is about 2.5 times larger than that accumulated in the first 3 months after the main shock. The afterslip on the decollement in the first 3 months and over 15 months contribute 68% and 80% of the total modeled moment release, respectively. It is worth noting that the slip on the lower decollement becomes more prominent over the longer time period.

[29] The afterslip moment inferred from the 15-month GPS observation is 4.7 1019 N m, 44% of which occurred in the first 3 months. In contrast, the seismic moment released by the aftershocks in the same period is approximately 2.0 1019 N m and 85% of that moment was released before the end of 1999 [Kao et al., 2002]. This indicates that although part of the GPS-observed moment

Figure 9. Total postseismic displacements during the 15-month period after the Chi-Chi earthquake, September 1999 to December 2000. Horizontal displacements are shown by vectors with their respective 95% confidence ellipses at the tip of each displacement vector. Vertical displacements are shown by vertical bars (uplift by a hollow bar and subsidence by a solid bar). The star denotes the epicenter of the main shock. CHF, the Changhua fault; CLPF, the Chelungpu fault; STF, the Shuantung fault.


ETG

3 - 13

may be due to aftershocks, there is still a large amount of postseismic deformation which is aseismic.

5. Conclusions

Figure 10. Model fault geometry. D, hinge depth; q, dip angle of lower fault segment.

[30] Postseismic displacements following the 1999 ChiChi earthquake were estimated from an analysis of the continuous and campaign-surveyed GPS data collected between September 1999 and December 2000 in the Taiwan area. As much as 252 mm of west-northwest directed horizontal displacement and 229 mm of uplift were detected on the hanging wall of the Chelungpu fault around the epicenter of the Chi-Chi main shock. Conversely, on the footwall there was only slight or insignificant subsidence and the horizontal displacements were small and directed to the southeast. In general, the pattern of postseismic defor-

Figure 11. Comparison between the observed and model predicted postseismic displacements. (a) The horizontal observation are shown by solid vectors and model predicted by dashed vectors. The 95% confidence ellipse of the observed displacement is indicated at the tip of each vector. (b) The vertical component depicted by arrows and model predicted by hollow bars. The surface projection of the optimal fault planes is noted by dashed rectangles. The star denotes the epicenter of the main shock. CLPF: the Chelungpu fault.

ETG

3 - 14


Figure 11. (continued)

Figure 12. The inverted afterslip distribution found in the 15-month period using optimal fault geometry. The hypocenter of the Chi-Chi main shock is depicted by a white star. See color version of this figure at back of this issue.


mation is similar to that of coseismic displacements. However, the postseismic displacements are much larger relative to the coseismic displacements in the southern part of the hanging wall than to the north where the maximum coseismic displacements occurred. The stations very close to the rupture show only small postseismic displacements. [31] On the basis of the four-segment fault model of Johnson and Segall (submitted manuscript, 2003), the postseismic GPS data was inverted to infer the deeper fault geometry and afterslip distribution. The optimal geometry of lower fault segment is a horizontal plane at a depth of 10.4 km, with 95% confidence interval 6.5 13.1 km, consistent with the results obtained for the data in first three months after the main shock [Hsu et al., 2002], which results indicated a nearly horizontal decollement some 8 – 12 km deep. The afterslip distribution over the 15-month period shows a similar pattern as that found for the first 3 months after the main shock. The maximum afterslip (459 mm) was found in the hypocentral region and in the northern part of the decollement. The afterslip moment inferred from the 15-month GPS data is 4.7 1019 N m, while the seismic moment due to aftershocks is about 2.0 1019 N m, which implies that a major part of postseismic deformation is aseismic. [32] As the time elapses after the Chi-Chi earthquake viscoelastic relaxation of the lower crust or the upper mantle may be required to explain the observed deformation. A dense continuous GPS array in central Taiwan has been in operation since 2001, and annual measurements on the campaigned-surveyed stations will continue to be carried out over the next few years. These GPS data should shed new light on the characteristics and mechanisms of postseismic deformation following a large thrusting earthquake. [33] Acknowledgments. We thank many colleagues at the Institute of Earth Sciences, Academia Sinica (H. H. Su, S. Y. Chen, S. Y. Ker, and F. C. Wang), and Land Survey Bureau, Ministry of Interior, who have maintained the operation of continuous GPS stations and participated in the field work of the postseismic GPS surveys. The generous provision of continuous GPS data by the CWB and MOI as well as precise ephemerides of GPS satellites by the IGS community is greatly appreciated. We are indebted to J. C. Savage, P. Segall, K. Johnson, N. Bechor, S. Williams, Y. Bock, J. Freymueller, and an anonymous reviewer for their constructive discussion and suggestions. This study was financially supported by Academia Sinica and the National Science Council of the Republic of China under grant NSC 91-2119-M-001-019. This is a contribution of the Institute of Earth Sciences, Academia Sinica, IESAS903.

References Carena, S., J. Suppe, and H. Kao, Active detachment of Taiwan illuminated by small earthquakes and its control of first-order topography, Geology, 30, 935 – 938, 2002. Central Geological Survey, Investigation report of 921 earthquake geology (in Chinese), 315 pp., Central Geol. Surv., Ministry of Econ. Affairs, Taiwan, 1999. Chang, C. H., Y. M. Wu, T. C. Shin, and C. Y. Wang, Relocation of the 1999 Chi-Chi earthquake in Taiwan, Terr. Atmos. Oceanic Sci., 11, 581 – 590, 2000. Chang, S. L., Subsurface geologic study of the Taichung basin, Petrol. Geol. Taiwan, 8, 21 – 45, 1971. Chen, K. C., B. S. Huang, J. H. Wang, and H. Y. Yen, Conjugate thrust faulting associated with the 1999 Chi-Chi, Taiwan, earthquake sequence, Geophys. Res. Lett., 29(8), 1277, doi:10.1029/2001GL014250, 2002. Chen, Y. G., W. S. Chen, J. C. Lee, Y. H. Lee, C. T. Lee, H. C. Chang, and C. H. Lo, Surface rupture 1999 Chi-Chi earthquake yields insights on active tectonics of central Taiwan, Bull. Seismol. Soc. Am., 91, 977 – 985, 2001.

ETG

3 - 15

Dominguez, S., J. P. Avouace, and R. Michel, Horizontal coseismic deformation of the 1999 Chi-Chi earthquake measured from SPOT satellite images: Implications for the seismic cycle along the western foothills of central Taiwan, J. Geophys. Res., 108(B2), 2083, doi:10.1029/ 2001JB000951, 2003. Harris, R., and P. Segall, Detection of a locked zone at depth on the Parkfield, California, segment of the San Andreas fault, J. Geophys. Res., 92, 7945 – 7962, 1987. Hirata, N., S. Sakai, Z. S. Liaw, Y. B. Tsai, and S. B. Yu, Aftershock observations of the 1999 Chi-Chi, Taiwan earthquake, Bull. Earthquake Res. Inst., 75, 33 – 46, 2000. Hsu, Y. J., N. Bechor, P. Segall, S. B. Yu, L. C. Kao, and K. F. Ma, Rapid afterslip following the 1999 Chi-Chi, Taiwan, Earthquake, Geophys. Res. Lett., 29(16), 1754, doi:10.1029/2002GL014967, 2002. Johnson, K. M., Y. J. Hsu, P. Segall, and S. B. Yu, Fault geometry and slip distribution of the 1999 Chi-Chi, Taiwan earthquake imaged from inversion of GPS data, Geophys. Res. Lett., 28, 2285 – 2288, 2001. Kao, H., Y. H. Liu, W. T. Liang, and W. P. Chen, Source parameters of regional earthquakes in Taiwan: 1999 – 2000 including the Chi-Chi earthquake sequence, Terr. Atmos. Oceanic Sci., 13, 279 – 298, 2002. Langbein, J., and H. Johnson, Correlated errors in geodetic time series: Implications for time-dependent deformation, J. Geophys. Res., 102, 591 – 603, 1997. Lee, J. C., H. T. Chu, J. Angelier, Y. C. Chan, J. C. Hu, C. Y. Lu, and R. J. Rau, Structural characteristics of northern surface rupture of the 1999 Mw = 7.6 Chi-Chi, Taiwan earthquake, J. Struct. Geol., 24, 173 – 192, 2002. Loevenbruck, A., R. Cattin, X. Le Pichon, M. L. Courty, and S. B. Yu, Seismic cycle in Taiwan derived from GPS measurements, C. R. Acad. Sci., Ser. II, 333, 57 – 64, 2001. Mandlebrot, B., and J. Van Ness, Fractional Brownian motions, fractional noise, and applications, SIAM Rev., 10, 422 – 439, 1968. Mao, A., C. G. A. Harrison, and T. H. Dixon, Noise in GPS coordinate time series, J. Geophys. Res., 104, 2797 – 2816, 1999. Matthews, M., and P. Segall, Statistical inversion of crustal deformation data and estimation of the depth distribution of slip in the 1906 earthquake, J. Geophys. Res., 98, 12,153 – 12,163, 1993. Meng, C. Y., San-I overthrust, Petrol. Geol. Taiwan, 2, 1 – 20, 1963. Nikolaidis, R., Observation of geodetic and seismic deformation with the Global Positioning System, Ph.D. dissertation, 249 pp., Univ. of Calif. San Diego, San Diego, Calif., 2002. Okada, Y., Surface deformation due to shear and tensile faults in a halfspace, Bull. Seismol. Soc. Am., 75, 1135 – 1154, 1985. Rothacher, M., and L. Mervart (Eds.), Bernese GPS Software v.4.0, 418 pp., Astron. Inst., Univ. of Berne, Berne, Switzerland, 1996. Saastamoinen, I. I., Contribution to the theory of atmospheric refraction, Bull. Geod., 107, 13 – 24, 1973. Shin, T. C., Some seismological aspects of the 1999 Chi-Chi earthquake in Taiwan, Terr. Atmos. Oceanic Sci., 11, 555 – 566, 2000. Suppe, J., Imbricated structure of western foothills belt, south central Taiwan, Petrol. Geol. Taiwan, 17, 1 – 16, 1980. Wang, C. Y., C. L. Li, F. C. Su, M. T. Leu, M. S. Wu, S. H. Lai, and C. C. Chern, Structural mapping of the 1999 Chi-Chi earthquake fault, Taiwan, by seismic reflection methods, Terr. Atmos. Oceanic Sci., 13, 221 – 226, 1999. Williams, S., The effect of coloured noise on the uncertainties of rates estimated from geodetic time series, J. Geod., 76, 483 – 494, 2003. Yang, M., R. J. Rau, J. Y. Yu, and T. T. Yu, Geodetically observed surface displacements of the 1999 Chi-Chi, Taiwan, earthquake, Earth Planets Space, 52, 403 – 413, 2000. Yu, S. B., et al., Preseismic deformation and coseismic displacements associated with the 1999 Chi-Chi, Taiwan, earthquake, Bull. Seismol. Soc. Am., 91, 995 – 1012, 2001. Zhang, J., Continuous GPS measurements of crustal deformation in southern California, Ph.D. dissertation, Univ. of Calif. San Diego, San Diego, Calif., 1996. Zhang, J., Y. Bock, H. Johnson, P. Fang, S. Williams, J. Genrich, S. Wdowinski, and J. Behr, Southern California permanent GPS geodetic array: Error analysis of daily position estimates and site velocities, J. Geophys. Res., 102, 18,035 – 18,055, 1997.

H.-Y. Chen, L.-C. Kuo, and S.-B. Yu, Institute of Earth Sciences, Academia Sinica, Nankang, Taipei 115, Taiwan. ([email protected]. edu.tw) Y.-J. Hsu, Institute of Geophysics, National Central University, Chungli, Taoyuan 320, Taiwan. C.-C. Liu, Land Survey Bureau, Ministry of Interior, Taichung 408, Taiwan.


Figure 12. The inverted afterslip distribution found in the 15-month period using optimal fault geometry. The hypocenter of the Chi-Chi main shock is depicted by a white star.

ETG

3 - 14