AB STRACTThe re sponse of Tai pei ba sin upon earth quake ex ci ta tion was stud ied us ing re cords of re cent earth quakes. The strong-mo tion da ta base in cludes re cords ob tained at 32 sta tions of the Tai pei TSMIP net work from 83 deep and 142 shal low earth quakes (M > 4.0) that oc curred in 1992 -2004. The char ac ter is tics of fre quency-de pend ent site re sponse were ob tained as spec tral ra tios be tween the ac tual earth quake re cords (hor i zon tal com po nents) and those mod elled for a hy po thet i cal Very Hard Rock (VHR) con di tion. The mod els for VHR spec tra of Tai wan earth quakes had been re cently pro posed by Sokolov et al. (2005bSokolov et al. ( , 2006. Anal y sis of site re sponse char ac ter is tics and com par i son with sim ple 1D mod els of the soil col umn re sulted in the fol low ing con clu sions: (1) The spec tral ra tios through out the ba sin ob tained from deep earth quakes (depth > 35 km) ex hibit good agree ment with the the o ret i cal ra tios cal cu lated us ing the 1D mod els con structed us ing avail able geo log i cal and geotechnical data. (2) The spec tral ra tios ob tained from shal low earth quakes show in flu ence of: (a) sur face waves gen er ated when trav el ling from dis tant sources to the ba sin and (b) rel a tively low-fre quency (< 1 -2 Hz) waves gen er ated within the ba sin. (3) Some shal low earth quakes pro duce ex tremely high am pli fi ca tion at fre quen cies 0.3 -1 Hz within the ba sin that may be dan ger ous for high-rise build ings and high way bridges. (4) The ob tained re sults may be used in proba bil is tic seis mic microzonation of the ba sin when many pos si ble earth quakes lo cated at various distances are considered. 2D and 3D simulation is necessary to model the seismic influence from particularly large earthquakes. Sci., 20, 687-702, doi: 10.3319/TAO.2008.10.15.01(T) IN TRO DUC TIONSeis mic ity in the Tai wan area is very high, and many large earth quakes (M > 6.5) have oc curred in the re gion in his tor i cal and mod ern times. Some of these earth quakes, e.g., the re cent Chi-Chi earth quake of 21 Sep tem ber 1999, caused se vere dam age. Tai pei City is the cap i tal of Tai wan and is lo cated on a sed i ment-filled ba sin in the north ern part of the is land. The area has ex pe ri enced sev eral dam ag ing earth quakes, the most re cent of which oc curred on 31 March 2002. The re sults of pre vi ous re search on earth quake ground mo tion pe cu liar i ties in the Tai pei ba sin (e.g., Kuo et al. 1995;Wen et al. 1995;Loh et al. 1998; Wen and Peng 1998a, b;Sokolov and Jean 2002;Wang and Lee 2002;Chen 2003;Fletcher and Wen 2005) showed that large lat eral vari a tions in ground-mo tion char ac ter is tics (peak ground ac cel er ation, re sponse spec tra, dom i nant fre quen cies, etc.) are appar ent.Re cent needs of earth quake en gi neer ing re quire consideration of site ef fect in seis mic haz ard anal y sis. When ap plying the proba bil is tic ap proach for seis mic haz ard es tima tion in ur ban ter ri to ries, it is...

S U M M A R YThe Taipei basin in northern Taiwan is located in a high seismicity region and was affected by several earthquakes in the past (M L = 7.3 on 1909 April 15; M L = 6.8 on 1986 November 15; the Chi-Chi M L = 7.3 earthquake on 1999 September 21 and M L = 6.8 on 2002 March 31). The main characteristic of the Taipei basin is its complex shape with a deep western and shallow eastern part. The uppermost Sungshan formation with its low shear wave velocities (90-200 m s −1 ) is also a distinct feature of the basin. Based on the large data base of earthquake records obtained from the Taiwan Strong Motion Instrumentation Program network, many studies on ground motion within the Taipei basin exist. However, the influence of the various subsurface structures on the observed ground motions as well as the variability of ground motion with respect to earthquake location is not fully understood. We apply a 3-D finitedifference method to simulate wave propagation up to 1 Hz for a small earthquake close to the basin in order to resolve these open questions. By varying source and structural parameters, we explore the variability of ground motion.Our study includes a subsurface model that is based on recent studies on the basin structure and on the crustal structure of Taiwan. From our simulations we find a good fit between simulated and observed waveforms and peak ground accelerations for the considered small earthquake near the basin. We also explore the influence of fault plane orientation, hypocentre location, deep basin structure and soft soil surface layers of the Sungshan formation by varying the subsurface structure and earthquake position. Our studies reveal that the basin structure produces an amplification factor of about 4 compared to hard rock conditions. Additionally, the soft soil Sungshan formation produce amplification of a factor of 2. This results in a maximum amplification of the basin structure of about 8, which is in good comparison with amplification values larger than 5 found from the analysis of observed earthquakes. These values clearly exceed the amplification values of about 2-3 obtained when applying standard 1-D site effect analysis.Our simulations for different earthquake positions show that ground motion depends strongly on earthquake location and fault orientation. Therefore, the application of average values of spectral amplification obtained from the analysis of recorded data from distant earthquakes with different azimuths and fault planes may significantly underestimate future ground motions of possible earthquakes on known faults close to the Taipei basin. The simulation of a small earthquake near the Taipei basin presented in this study will help to set up adequate simulation parameters for a possible large earthquake close to the Taipei basin. Such a simulation of a scenario earthquake close to the Taipei basin would allow to significantly improve hazard assessment as no observations of strong earthquakes in the vicinity of the basin exist.

S U M M A R YFinite difference (FD) simulation of elastic wave propagation is an important tool in geophysical research. As large-scale 3-D simulations are only feasible on supercomputers or clusters, and even then the simulations are limited to long periods compared to the model size, 2-D FD simulations are widespread. Whereas in generally 3-D heterogeneous structures it is not possible to infer the correct amplitude and waveform from 2-D simulations, in 2.5-D heterogeneous structures some inferences are possible. In particular, Vidale & Helmberger developed an approach that simulates 3-D waveforms using 2-D FD experiments only. However, their method requires a special FD source implementation technique that is based on a source definition which is not any longer used in nowadays FD codes. In this paper, we derive a conversion between 2-D and 3-D Green tensors that allows us to simulate 3-D displacement seismograms using 2-D FD simulations and the actual ray path determined in the geometrical optic limit. We give the conversion for a source of a certain seismic moment that is implemented by incrementing the components of the stress tensor.Therefore, we present a hybrid modelling procedure involving 2-D FD and kinematic raytracing techniques. The applicability is demonstrated by numerical experiments of elastic wave propagation for models of different complexity.

The August 17, 1999 Kocaeli earthquake ruptured the earth's surface along 145 km and produced a magnitude of M w ¼ 7.4. As expected for such an event Modified Mercalli intensities of typically IX and X in the vicinity of the fault were determined. Yet the observed accelerations at the five near-fault sites remained amazingly small with horizontal PGA values of 0.14 g to 0.4 g. We attempt to resolve the enigma by modeling surface ground motion with a finite-difference algorithm, utilizing two different rupture and slip histories derived from the strong-motion observations and 'translate' the computed horizontal motion in intensity values. We can show that (a) in a given simple crustal velocity model different slip distributions result in significantly different ground motion distributions in the vicinity of the fault even though both slip distributions fit the observed accelerometer data quite well. (b) Both slip distributions 'project' high ground motion into areas adjacent to the fault where no accelerometer was located. (c) Both slip distributions are not fully compatible with observed intensity observations around the fault, although this could be partly attributed to the lack of knowledge regarding to the crustal structure. In the light of our results it would thus be misleading if the few strong-motion observations around the Kocaeli earthquake fault were taken as typical or representative for the entire area and for potential future events.

Taipei, the capital of Taiwan, suffered from destructive earthquakes four times during the 20th century (M L = 7.3 on April 15, 1909; M L = 6.8 on November 15, 1986; the Chi-Chi M L = 7.3 earthquake on September 21, 1999; and M L = 6.8 on March 31, 2002). Analysis of recorded data shows a strong dependence of spectral amplification in the Taipei Basin on earthquake depth and azimuth. At low frequencies ( f < 3 Hz) significant larger amplifications are observed for shallow earthquakes as compared to intermediate depth events. The former ones also display strong azimuthal dependence. As structures with large response periods such as bridges and tall buildings are sensitive to these low frequencies the understanding of the associated wave effects within the basin and their role for site effect amplification is critical. The tool we employ is 3D finite-difference modeling of wave propagation of incident wave fronts. The available detailed model of the basin allows studying the wave effects. Modeling clearly reveals that basin edge effects as observed in data are related to surface wave generation at the basin edges with a high degree of azimuthal dependency. The reproduced site amplification effects are in qualitative agreement with the observations from strong motion data.