Vol. 45 No. 2 SCIENCE IN CHINA (Series D) February 2002

Determination of Euler parameters of plate and the inferences

ZANG Shaoxian (臧绍先)1, CHEN Qiyong (陈起永)1*, NING Jieyuan (宁杰远)1, SHEN Zhengkang (沈正康)2 & LIU Yonggang (刘永刚)1

1 Department of Geophysics, Peking University, Beijing 100871, China; 2 Department of Earth Sciences, University of California, Los Angeles, USA Correspondence should be addressed to Zang Shaoxian (email: [email protected])

Received July 20, 2001 Abstract Euler vectors of 12 plates, including (PH), relative to a randomly fixed (PA) were determined by inverting the 1122 data from NUVEL-1 global plate motion model, slip vectors along Philippine Sea plate boundary, and GPS observed velocities. Euler vectors of Philippine Sea plate relative to adjacent plates are also gained. Our results are well consistent with observed data and can satisfy the geological and geophysical con- straints along the Caroline(CR)-PH and PA-CR boundaries. Deformation of Philippine Sea plate is also discussed by using the plate motion Euler parameters.

Keywords: Philippine Sea plate, earthquake slip vectors, intraplate deformation, relative plate motion, plate motion parameters.

Euler parameters are the basic parameters describing the plate motion. The Philippine Sea plate, proposed as a distinct plate at the early stage of plate motion theory, plays a special role in the history of west Pacific and east Eurasia continent. Containing mainly convergent boundaries and lacking of accreting boundary, which is useful to determine relative motion rates, and transform boundary, which can determine relative motion azimuth more accu- rately than earthquake slip vectors, along the boundaries make it very hard to determine Euler vector of Philippine Sea plate. It is a difficulty and also a focus of . Before the availability of accurate Geodesy data, only slip vectors of thrust along the subducted boundary of Philippine Sea plate can be used to constrain its motion, but there exist at least three factors that cause uncertainties when using these earthquake slip vectors. First, the back-arc spreading and the deformation of overriding plate, such as the spreading of Mariana Trough and strike slip such as Philippine fault within overriding plate, make the azimuth of earthquake slip vectors differ from that predicted by plate motion. Second, Some small colliding zones around the Philippine Sea plate make earthquake focal mechanism more complicated. Fi- nally, it is uncertain whether the north boundary of Philippine Sea plate is adjacent to Eurasia plate or North America plate, or even a distinct . These problems must be carefully ana- lyzed before determining the Euler vector of Philippine Sea plate with slip

* Now at Department of Geological Sciences, Northwestern University, Evanston, IL 60208, USA 134 SCIENCE IN CHINA (Series D) Vol. 45 vectors. Many researchers have discussed or obtained Euler vectors of Philippine Sea plate with earthquake slip vectors[1 6], or discussed the constrain conditions[7]. Seno et al.[8] discussed the deficiencies in previous studies and obtained Euler vector of Philippine Sea plate by using 11 and 16 earthquake slip vectors at - along the PH-EU boundary and Izu- Bonin Trench along the PH-PA plate boundary respectively, constrained by NUVEL-1[9] EU-PA Euler vector. Seno’s result is now the best, but the data constraining motion of the Philippine Sea plate are still not enough, especially lacking the data of rates. Many new earthquake slip vectors have become available from Harvard CMT catalogue. The foundation of GPS observation network makes the data of rates describe the motion between geological blocks or plates available, which was used to study the deformation of continent[10,11]. It is possible to determine the Euler vector of Philippine Sea plate more accurately now. In this study, we obtain our PH-PA Euler vector with a global inversion by simultaneously inverting these new data, and the 1122 data from NUVEL-1 model.

1 Method of determining the Euler parameters of Philippine Sea plate

The Euler parameters of the Philippine Sea plate in NUVEL-1 model were taken from Seno et al.[12] and not gained by inversion, possibly because of lacking data describing motion of the Philippine Sea plate, especially the data of rates. So, there are total 12 plates in NUVEL-1 model and Euler parameters of 11 plates relative to Pacific plate are gained by inversion. In this study, we add the Philippine Sea plate to the 12 plates in NUVEL-1 model and gain Euler parameters of 13 plates constrained to consistency with global plate circuit closure. Suppose there exist m plates, we randomly choose the reference frame so that the mth plate is fixed. The 3m-3 independent components of other m-1 plates of a plate motion model then form a model vector M: M = (q 1 ,L ,q m -1 ,f1 ,L f m-1 ,w1 ,L ,w m-1 ) , q i ,f i ,w i are Euler pole’s latitude, longitude, and the rotation rate of ith plate.

0 Suppose there exist N observed data and they form a vector d . s i and di (M ) are the

0 0 standard error and predicted value, respectively, of di , the ith component of d . And, suppose there exist n and N-n observed data of magnitudes and direction, respectively, of relative motion. The total residual error is 2 2 2 N é d 0 - d (M ) ù n éd 0 - d (M ) ù N é d 0 - d (M ) ù c 2 = å ê i i ú = å ê i i ú + å ê i i ú , (1) i=1 ëê s i ûú i=1 ëê s i ûú i=n+1 ëê s i ûú where c 2 achieves a stationary value when its derivatives with respect to the model parameters are zero. And, we suppose that the dispersion between the observed and predicted directions is 0 0 less than 20°, then we replace sin[ di - d i (M )] by di - d i (M ) and get No. 2 DETERMINATION OF EULER PARAMETERS OF PHILIPPINE SEA PLATE 135

N d (M ) ¶d (M ) N d 0 ¶d (M ) i i = i i , j = 1, L , 3m - 3 . (2) å 2 å 2 i =1 s i ¶M j i=1 s i ¶M j Construct a reasonably good starting model M * so that M - M * is small enough and we 2 can replace ¶d i (M ) / ¶M j by ¶d i (M *) / ¶M j . Neglecting terms of (M j - M j *) , finally, we get

N 3m-3 1 ¶d ¶d N 1 ¶d i i (M - M *) = i [d 0 - d (M *)] , j = 1, L , 3m - 3 . (3) å å 2 ¶M ¶M k k å 2 ¶M i i i=1 k =1 s i j k i=1 s i j Eq. (3) is a linear equation of M - M * . The matrix expression is AT V -1 AdM = AT V -1dd 0 , (4) where

0 0 ¶d i T 2 dM = M - M *, dd = d - d (M *), Aik = , Aik = Aki, Vik = s i d ik . (5) ¶M k If independent data are as many as model parameters (or more independent data), eq. (4) has the unique solution dM = ( AT V -1 A) -1 AT V -1dd 0 . (6) The new estimate of the model is given by M = dM + M * . This procedure is iterated until convergence is gained.

2 Data and process

Data used in this study include: (1) The 1122 data, consisting of spreading rates, azimuth, and earthquake slip vectors, from NUVEL-1 model. (2) 67 earthquake slip vectors exist along the Philippine Sea plate boundary. In order to make sure that the slip vectors represent the azimuth of interplate motion, we choose them according to three rules. First, the magnitudes of earthquake are big enough. We choose those with a scalar moment scale of 24 or more. Second, epicenters are located in subduction zone and on the top surface of Benioff Zone. Third, the depths of epicenters are between 15 and 60 km, the fault plane solutions are for thrust earthquakes, and the dips of the fault plane are close to the dip of subduc- tion zone. We discard the data in Luzu Trough-Philippine Trend when selecting earthquake slip vectors from Harvard CMT 1977 —1999, because the Philippine Archipelago is a complicated deforma- tion zone with a nearly N-S left-handed strike slip fault passing through it, thus making the rela- tive motion between EU and PH more complicated. Also, we do not use the data in because of the ~40 mm/a spreading rate at the back-arc of Mariana which changes the coupling status between PH and PA. We finally choose 32 earthquake slip vectors at Nankai Trough-Ryukyu Trench and 35 at Izu-Bonin Trench, including 25 from Seno et al.[8] Table 1-1 shows these earthquake slip vectors and their predicted values of this study. 136 SCIENCE IN CHINA (Series D) Vol. 45

(3) GPS data. Observed velocities of 4 GPS stations located in PH plate relative to EU plate are shown in table 1-2. Data of Guam and Palau stations are not used for inversion.

Table 1-1 Data at PH-EU boundary: Nankai Trough-Ryukyu Trench No. 1a) 2 a) 3 a) 4 a) 5 a) 6 a) 7 a) 8 a) 9 10 Lat. (°N) 33.76 33.7 33.13 32.31 32.26 32.24 32.23 32.15 31.84 31.79 Lon. (°E) 137.2 136.05 135.84 131.83 131.78 132.21 131.78 131.98 131.8 131.63 Obs. value (°) -57 -54 -50 -48 -49 -54 -48 -48 -51.4 -53 s (°) 10 10 10 15 10 15 10 15 20 20 Pre. value (°) -44.6 -43.7 -44.5 -42.9 -42.9 -43.2 -43 -43.2 -43.5 -43.5 11 12 13 14 a) 15 16 17 18 19 20 21 31.79 31.73 31.29 31.14 30.59 30.59 29.78 29.77 29.48 29.28 29.26 131.31 131.59 131.29 131.33 131.24 131.07 130.51 130.63 130.64 130.31 129.86 -53.4 -41.2 -44.4 -61 -46.6 -59.8 -49.4 -41.7 -46.6 -44 -43.8 15 20 15 10 20 15 20 20 20 20 20 -43.3 -43.6 -44 -44.2 -44.9 -44.8 -45.5 -45.6 -46 -46 -45.8 22 23 24 25 26 27 a) 28 29 30 31 a) 32 29.24 29.2 28.51 27.99 27.52 27.48 27.42 26.61 24.88 24.3 24.24 130.31 130.31 130 129.39 128.44 128.57 128.48 127.81 125.33 125.21 125.23 -35 -45.8 -46.6 -41.1 -40 -48 -45.4 -51.4 -42.3 -43 -55.1 20 20 20 20 20 15 20 20 20 10 20 -46.1 -46.1 -46.8 -47 -47 -47.1 -47.1 -47.6 -47.9 -48.4 -48.5 (To be continued)

Table 1-1 Data at PA-PH boundary: Izu-Bonin Trench (continued) No. 1 a) 2 3 a) 4 5 6 a) 7 8 9 10 a) 11 Lat. (°N) 35.93 35.86 35.76 35.74 35.68 35.67 35.58 35.57 35.56 34.02 33.91 Lon. (°E) 140.08 141.35 140.7 141.08 140.65 140.64 140.61 140.65 140.54 141.63 141.35 Obs. value (°) -89 -64.6 -83 -75.5 -84.3 -80 -75.3 -68.6 -92.4 -86 -94.4 s (°) 15 20 15 20 20 15 20 20 20 10 20 Pre. value (°) -82.9 -80.9 -81.9 -81.3 -82 -82 -82 -82 -82.1 -80.1 -80.5 12 a) 13 14 a) 15 16 17 18 19 a) 20 21* 22 23 33.61 33.49 33.38 33.31 33.26 32.61 31.35 31.34 31.22 31.11 30.92 30.75 141.71 141.4 140.97 141.36 141.34 141.27 140.96 141.8 141.75 142.09 141.51 141.73 -78 -90.9 -80 -91.6 -75.6 -79.8 -82.3 -80 -89.2 -82 -93.8 -72 10 20 20 20 15 20 20 15 20 10 20 10 -79.8 -80.3 -81 -80.4 -80.4 -80.3 -80.6 -79.1 -79.2 -78.5 -79.5 -79.1 24 a) 25 26 27 28 a) 29 30 31 32 33* 34 a) 35 a) 30.71 30.69 30.63 30.61 29.34 29.31 29.3 28.97 28.12 27.84 27.72 23.71 141.82 141.62 141.63 141.62 142.23 141.98 142.02 142.05 142.41 142.76 142.08 143.2 -72 -86.3 -88.8 -90.2 -75 -90.4 -81.2 -87.5 -89.2 -78 -81 -70 15 15 20 20 10 20 20 20 20 15 15 10 -78.9 -79.2 -79.2 -79.2 -77.7 -78.2 -78.1 -77.9 -77 -76.2 -77.5 -73.4 a) from Seno et al.[8] No. 2 DETERMINATION OF EULER PARAMETERS OF PHILIPPINE SEA PLATE 137

Table 1-2 Velocities of GPS stations relative to Eurasia plate

GPS Rate (mm/a) Azimuth (°) Lat. (°N) Lon. (°E) Note station observed predicted observed predicted S063 22.67 121.47 68.60 71.51 315.4 312.02 1 S102 22.04 121.56 71.70 72.14 316.8 311.47 2 BTS3 20.44 121.96 80.70 73.58 298.6 310.05 2 Okin 20.43 136.08 59.77 61.48 300.1 300.56 3 Guam 13.59 144.87 22.52 66.03 301.0 287.83 4 Palau 7.34 134.48 100.42 80.22 277.9 294.45 4 1, Computed from Yu and Kuo[13] and Yu et al.[14]; 2, Yu and Kuo[13]; 3, Kato et al.[15]; 4, computed from Kato et al.[16] 3 Results

We obtain Euler vectors of other 12 plates relative to Pacific plate by inverting all the data above simultaneously, being constrained to consistency with global plate circuit closure. Results are shown in table 2. Results of NUVEL-1 model are also given for comparison. The results of EU-PH and PA-PH Euler vectors are shown in table 3. Fig. 1 shows the locations of PA-PH and EU-PH Euler poles of this and previous studies.

Table 2 Comparison of 12 plates’ Euler vectors of this study and NUVEL-1(Pacific fixed) Lat. of pole (°N) Lon. of pole (°E) Rotation rate (°/Ma) No. Plate name ours NUVEL-1 ours NUVEL-1 ours NUVEL-1 01 N. America 48.88 48.71 -78.07 -78.17 0.781 0.783 02 Cocos 36.90 36.82 -108.55 -108.63 2.085 2.089 03 Nazca 55.65 55.58 -89.82 -90.10 1.422 1.422 04 Antarctica 64.41 64.32 -83.33 -83.98 0.910 0.909 05 S. America 55.37 55.00 -85.25 -85.75 0.668 0.666 06 Eurasia 61.22 61.07 -85.78 -85.82 0.897 0.899 07 Africa 59.24 59.16 -72.74 -73.17 0.971 0.970 08 Caribbean 55.24 54.19 -80.99 -80.80 0.860 0.853 09 Australia 59.95 60.08 1.91 1.74 1.127 1.123 10 Arabia 59.60 59.66 -33.21 -33.19 1.164 1.162 11 India 60.43 60.49 -30.46 -30.40 1.157 1.154 12a) Philippine 2.65 -1.24 -44.35 -45.81 0.951 1.000 a) This Euler vector in NUVEL-1is taken from Seno et al.[12] , the one here is from Seno et al.[8]

Table 3 Some Euler vectors and comparison with those of Seno et al.[8] Lat. of pole (°N) Lon. of pole (°E) Rotation rates (°/Ma) Plate pair ours Seno et al.[8] ours Seno et al.[8] ours Seno et al.[8] PA-PH -2.65 1.24 135.65 134.19 0.951 1.000 EU-PH 47.16 48.23 160.19 156.97 1.012 1.085 CR-PH 5.40 6.24 134.60 134.09 0.500 0.700 PA-CR -11.39 -10.13 136.77 134.43 0.461 0.309 138 SCIENCE IN CHINA (Series D) Vol. 45

Fig. 1. Euler poles of EU-PH and PA-PH from different researchers. Solid line ellipses are the 95% confidence error ellipses gained by using a global inversion in this study and dashed line ellipses are the 68% confidence error ellipses of best-fitting poles from Seno et al.[8]. , This study; , Chase[2]; , Minster&Jordan[3]; , Karig[4]; , Ranken

et al.[5]; ╋, Huchon[6]; , Seno[1]; , Seno et al[8] best- fitting pole; , Seno et al.[8]; , This study using only earthquake slip vectors. Plate name abbreviations: EU, Eurasia; PH, Philippine Sea; PA, Pacific; NA, North Amer- ica; CR, Caroline; AU, Australia.

There are many micro plates besides the above main 13 plates not included in current global plate motion models, among which the (CR) has the most important influ- ence on our Philippine Sea plate’s Euler vector. The boundary constraints, divergence at Sorol Trough and convergence at Mussau Trench of PA-CR boundary were proposed by Weissel and Anderson[7] according to seafloor geological data and are a widely used constraint condition of PA-CR boundary currently. We obtain our PA-PH Euler vector directly by a global inversion and then adjust CR-PH Euler vector, and get a set of PA-CR Euler vectors satisfying the PA-CR boundary constraints. Our results show that a PA-CR Euler vector satisfying the PA-CR boundary constraint cannot be obtained when the CR-PH Euler vector’s rotation rate is bigger than 0.5°/Ma. We select the PA-CR Euler vector indicating the most obvious spreading at Sorol Trough and the corresponding CR-PH Euler vector. Table 3 shows PA-PH, EU-PH, CR-PH Euler vector and the PA-CR Euler vector satisfying the PA-CR boundary constraint. The results of Seno et al[8] are also shown.

4 Discussion and conclusion

4.1 Comparison of the observed data and the predicted results We discuss the consistency of the observed data and the predicted values at the Philippine Sea plate boundaries. The results of Seno et al.[8] are also included for comparison. First, we com- pare the azimuths of earthquake slip vectors. Fig. 2(a) and (b) indicate the observed values and the predicted values of earthquake slip vectors at Nankai Trough-Ryukyu Trench of PH-EU boundary, and Izu-Bonin Trench of PA-PH boundary, respectively. The predicted values are consistent with observed data better than Seno et al.’s[8] predicted values, especially at the PA-PH boundary. We think that Seno’s adjusting the PA-PH Euler pole inside the 68% confidence error ellipse in order No. 2 DETERMINATION OF EULER PARAMETERS OF PHILIPPINE SEA PLATE 139 to gain a PA-CR Euler vector satisfying the PA-CR constraint made the consistency of observed and predicted earthquake slip vectors reduce at the PA-PH boundary.

Fig. 2. Comparison of the predicted and observed values of earthquake slip vectors. (a) Nankai Trough-Ryukyu Trench at PH- EU boundary; (b) Izu-Bonin Trench at PA-PH boundary; +, observed data; , predicted value; , Seno et al.[8]

Three earthquakes in the bottom right corner of fig. 2(a) lie away from the gathering of other earthquakes and have azimuths bigger than 56°N. Among the 3 earthquakes, the one with the low- est latitude is selected by us from Harvard CMT. This earthquake, with 30.59°N latitude, 131.07°E longitude, 33km depth, 208° fault strike, 25° dip and 88° slide angle, cannot be discarded with sufficient reasons. The other two earthquakes are used by Seno et al.[8] The one with the highest latitude is at the location of epicenter (33.76°N, 137.20°E). The depth of the earthquake is 21 km and the fault strike is 267.5° which differs significantly from the trench strike. The another earth- quake with the location (31.14°N, 131.33°E) and a depth of 8 km, is too shallow and may not represent the interplate motion. If we get rid of the latter 2 data, the consistency of our predicted and observed values will be better in fig. 2(a). The predicted and observed values of GPS data, indicated by solid and dashed bars with ar- rows respectively, are shown in fig. 3. Among the 4 GPS stations, the misfit between observed and predicted values of BTS3 is relatively high. This may be due to BTS3’s closing to left-handed strike slip deformation zone between and and thus may be affected by the defor- mation (Zang et al[17]). Other station’s predicted and observed velocities well coincide, especially Okino Torishima station, which has a (59.77 mm/a, -59.9°N) observed velocity and a (61.48 140 SCIENCE IN CHINA (Series D) Vol. 45 mm/a, -59.4°) predicted velocity (fig. 3 and table1-2). The consistency between observed and predicted value is better than Seno et al.’s[8] which has a slight misfit[16] bigger than ours. So, we give here a revised model of a 13-plate global model for NUVEL-1.

Fig. 3. Comparison of observed data and predicted values. Solid dots indicate locations of data. Dashed bar along PH-EU and PA-PH boundaries represent the azimuth of earthquake slip vector while the bold solid bar represents the predicted relative mo- tion azimuth at the two boundaries. Solid bar with arrow and dashed bar with arrow are the predicted and observed velocity vectors of GPS station relative to the Eurasia plate, respectively. Scale on top-left corner measures magnitudes of velocity vectors only.

4.2 Deformation inside the Philippine Sea plate and along the plate boundaries Fig. 3 shows the observed and the predicted results of this study. The consistency of the pre- dicted and the observed results is quite good at Nankai Trough-Ryukyu Trench and Izu-Bonin Trench, so is that of the predicted and observed results at GPS stations S063, S102, and Okino No. 2 DETERMINATION OF EULER PARAMETERS OF PHILIPPINE SEA PLATE 141

Torishima. The consistency shows that the above areas have a uniform motion as a rigid plate. At , Mariana Trough and Yap-Palau Trench, the predicted and observed results show poor consistency that is why we do not use data in these areas for inversion. The GPS station Guam lies between Mariana Trench and Mariana Trough. It shows signifi- cant misfit between the observed and predicted azimuths as well as rate relative to Eurasia plate. The misfit of the rate is about 40 mm/a and the direction is nearly E-W (fig. 3) that is consistent with the spreading rate of Mariana Trough given by Hussong and Uyeda[18]. The misfit between the predicted and observed results is possibly caused by the spreading of Mariana Trough, which is obviously the main deformation in this area. The misfit between predicted and observed results at Palau station is also obvious. This may be due to Palau’s closing to the incipient subduction boundary whose subduction makes the ob- served results differ significantly from the results predicted by plate motion theory. In Philippine Trench, the predicted and observed results also show poor consistency which may be caused by the moving of the sinistral Philippine Fault in Philippine Archipelago and the decoupling between the Philippine Sea plate and the Philippine Archipelago[1], or by the affection of South China Sea secondary-plate and the interplate deformation of the Philippine Archipel- ago[17]. The observed and predicted data at Nankai Trough-Ryukyu Trench are consistent very well. Thus, we believe that the extension of at Ryukyu Trench is not obvious.

Acknowledgements Many thanks to Dr. T. Kato at Earthquake Institute of Tokyo University and Dr. Shui-Beih Yu at Earth Sciences Institute in Taiwan for their updated prints and GPS data. This work was supported by Project of Mechanism and Prediction of Continental Strong Earthquake (Grant No. 95-13-04-06). References

1. Seno, T., The instantaneous rotation vector of the Philippine Sea Plate relative to the Eurasia plate, Tectonophysics, 1977, 42: 209 226. 2. Chase, C. G., Plate kinematics: The Americas, East Africa, and the rest of the world, Earth Planet Sci. Lett., 1978, 37: 355 368. 3. Minster, J. B., Jordan, T. H., Rotation vectors for the Philippine and Rivera plates, Eos. Trans. AGU, 1978, 60: 958. 4. Karig, D. E., Basin genesis in the Philippine Sea, Initial Rep. Deep Sea Drill Proj., 1975, 31: 857 879. 5. Ranken, B., Cardwell, R. K., Karig, D. E., Kinematics of the Philippine Sea plate, Tectonics, 1984, 3: 555 575. 6. Huchon, P., Comments on “Kinematics of the Philippine Sea plate” (eds. Branken, R. K., Cardwell, Karig, D. E.) Tecton- ics, 1986, 5: 165 168. 7. Weissel, J. K., Anderson, R. N., Is there Caroline plate? Earth Planet Sci. Lett., 1978, 41: 143 158. 8. Seno, T., Seth, S., Alice, E. G., A model for the motion of the Philippine Sea plate consistent with NUVEL-1 and Geological Data, J. Geophys. Res., 1993, 98: (B10): 17941 17948. 9. DeMets, C., Gordon, R. G., Argus, D. F. et al., Current plate motions, Geophys. J. Intel., 1990, 101: 425 478. 10. Zhu, W. Y., Cheng, Z. Y., Xiong, Y. Q. et al., Preliminary results of measuring the crustal deformation in Qinghai-Xizang area using GPS technique, Scence in China, Ser. D, 1998, 41(3): 314 318. 11. Zhu, W. Y., Wang, X. Y., Cheng, Z. Y. et al., Crustal motion of Chinese mainland monitored by GPS, Science in China, Ser. D, 2000, 43(4): 394 400. 12. Seno, T., Moriyama, T., Stein, S. et al., Redetermination of the Philippine sea plate motion (abstract), EOS Trans. Am. Geophysics Res., 1987, 68: 1474. 142 SCIENCE IN CHINA (Series D) Vol. 45

13. Yu, S. B., Kuo, L. C., GPS observation of crustal deformation in the Taiwan-Luzon region, Geophysical Res. Lett., 1999, 26(7): 923 926. 14. Yu, S. B., Chen, H. Y., Kuo, L. C., Velocity field of GPS station in the Taiwan area, Tectonophysics, 1997, 274: 41 59. 15. Kato, T., Kotake, Y., Chachin, T. et al., An estimate of the Philippine Sea plate motion derived from the Global Position- ing System observation at Okino Torishima, Japan, J. Geod. Soc. Japan, 1996, 42: 233 243. 16. Kato, T., Kotake, Y., Nakao, S. et al., Initial results from WING, the continuous GPS network in the western Pacific area, Geophysical Res. Lett., 1998, 25(3): 369 372. 17. Zang, S. X., Chen, Q. Z., Huang, J. S., Distribution of earthquakes, stress state and interaction between Plates in southern Taiwan-Philippine area, Seismogeology (in Chinese), 1994, 16: 29 37. 18. Hussong, D. M., Uyeda, S., Tectonic processes and the history of the Mariana Arc: A synthesis of the results of deep sea drilling project Leg 60, Initial Rep. Deep Sea Drill Proj., 1981, 60: 909 929.