Crustal Poisson ' s Ratio Off Eastern Taiwan From OBS Data Modeling

One vertical component and two horizontal components of 27 ocean­ bottom seismometer (OBS) data sets along four profiles off eastern Taiwan were obtained using a large air-gun array of the RIV Maurice Ewing dur­ ing summer/fall 1995. From the horizontal components of the OBS data, we identify most refractions converted in the sediment and reflections con­ verted in the crust respectively as the first and later arrivals. These con­ verted arrivals are subsequently used for travel-time modeling of the sedi­ mentary and crustal Poisson's ratios. The overall results show differences of Poisson's ratios in the accretionary prism (0.28-0.3), the continental crust (0.25-0.29) and the oceanic slab (0.23-0.33), and these are associated with the different structural compositions and/or tectonic settings. OBS data modeling of profile EW9509-1, which covers the Ryukyu arc-trench system in the north-south direction, reveals relatively low Poisson's ratios in the northern portion of the Yaeyama accretionary prism (0.28) and in the south­ ern portion of the Ryukyu Arc basement (0.25). Anomalous Poisson's ra­ tios within the accretionary prism and the continental crust may be attrib­ uted to the fractures of the forearc compression beneath the Nanao Basin. From the east-west profiles of EW9509-14 and EW9509-16 across three forearc basins, we also find anomalously high Poisson's ratios (0.27-0.29) for the Ryukyu Arc basement beneath the Nanao Basin. High Poisson's ratios of the overlying plate suggest the presence of fluid-filled micro-cracks, which may be related to the northward movement of the Luzon Arc and the Ganga Ridge. Poisson's ratio decreases northwestward along profile EW9509-23 that is consistent with the increasing crustal thickness north­ westward in the P-wave modeling below the Huatung Basin.


INTRODUCTION
The complexity of sedimentary and crustal structures off eastern Taiwan has been attrib uted to the oblique convergence of the Philippine Sea Plate (PSP) and the Eurasian Plate (EP), as illustrated by the tectonic provinces and the sea floor in Fig. L Off northeastern Taiwan, the northward subduction of the PSP beneath the Ryukyu Arc and the southward extension of the Okinawa Trough led to the forearc compression of the Ryukyu Arc basement and the northern Yaeyama accretionary prism (Wang et al. 2001).Off southeastern Taiwan, the northwestward collision of the Luzon Arc and the EP continent generated the collision front along the east coast, whereas the northward subduction of the Gagua Ridge (Schnurle et al. 1998) resulted in the fracture zones beneath the Huatung Basin (Kao et al. 2000).Understanding the sedimen tary and crustal properties off eastern Taiwan is critical.For example, relocations and mecha nisms of earthquakes in this seismogenic zone can be enhanced on the basis of velocity models of P-and S-waves.
Various geophysical properties have been investigated to understand the complexity of the sedimentation and the crustal structures off eastern Taiwan.Among them, P-wave veloci ties from deep seismic acquisitions (Hagen et al. 1988;Cheng et al. 1996) and from earth quake tomography (Hsu 2001) have been derived to image the crust with higher resolution than gravity and magnetism.However, due to limitations with respect to instrumentation and data quality in the marine environment, these P-wave velocity models have neither delineated the specific boundaries of the tectonic structures nor determined the structural composition.
On the other hand, the Poisson's ratio (or the ratio of S-wave velocity over P-wave velocity) of the various tectonic structures can be used to identify the structural composition, fracture and fluid saturation, including gas hydrates (Tinivella and Accaino 2000).The reasons for using anomalous Poisson's ratio to investigate the lithosphere are numerous.The Poisson's ratio of the continental crust or a crust older than 140 Ma is generally uniform and less than 0. 28 (Holbrook et al. 1988) compared with the irregular and highly variable Poisson's ratio of the young oceanic crust.Shaw (1 994) and Kodaira et al. (1996) indicated that the Poisson's ratio of an oceanic crust less than 15Ma can reach the highest value of 0.33.On the other hand, due to variations in structural composition and porosity, the Poisson's ratio of an oceanic crust can vary from 0.25 to 0.32 (Bratt and Solomon 1984;Christensen 1996).Since most oceanic sediments are less consolidated, their Poisson's ratios might be as high as 0.39-0.49(Iwasaki et al. 1994;Christeson et al. 1997).Hence, models of the Poisson's ratio in this study can be used to understand the structural compositions and the tectonic settings of the lithosphere off eastern Taiwan.On the basis of travel-time modeling and fractured models, Shaw (1994) pro posed an anomalously low Poisson's ratio (0.24) for highly fractured crusts but high Poisson's ratios (0.28-0.3) for less fractured crusts.Alternatively, high Poisson's ratio and S-wave anisot ropy were evidence of fluid-filled micro-cracks (Mjelde et al. 1995).It is followed that highly fractured rocks and/or pore fluids resulting from the subduction and the arc-continent collision off eastern Taiwan can be identified from the anomalous Poisson's ratios in this study.The aim of this paper is to investigate these anomalous Poisson's ratios enclosed by the well constrained interfaces for confirming tectonic boundaries and for identifying fracture zones

124'
and fluid intrusions in the Ryukyu arc-trench system and in the Huatung Basin.

P-WAVE VELOCITY PROFILES FROM OBS DATA MODELING
In August and September 1995, twenty-seven ocean-bottom seismometers (OBS) deployed by the RIV Ocean Researcher I and four corresponding multi-channel seismic (MCS) lines collected by the RIV Maurice Ewing (Fig. 1) were implemented off eastern Taiwan (Liu et al. 1997).OBS and MCS data modeling of four P-wave velocity (V ) profiles (Mcintosh and p Nakamura 1998; Wang and Chiang 1998;Yang and Wang 1998;Wang et al. 2001) provided velocity and interface information of crustal structures with the highest resolution ever achieved in this region.The P-wave velocity of profile EW9509-1 (Wang et al. 2001) shows a suddenly decreasing thickness (from 7.5 km to 6 km) and an abruptly increasing angle (from 5 degrees to 25 degrees) of the subduction slab (Vp == 6.75-7.75km/sec) below the Nanao Basin.A southward dipping interface (Vp contour of 6.75 km/sec) of about 30 degrees within the Ryukyu Arc and the variation of the Moho (Vp contour of 7.75 km/sec) beneath the Ryukyu Arc may have resulted from the northward subduction of the PSP.The thickness of the Ryukyu Arc basement (Vp == 5.5-6.75 km/sec) and the subducting PSP crust (Vp == 6.75-7.75km/sec) along two profiles, i.e., EW9509-14 (Mcintosh and Nakamura 1998) and EW9509-16 (Wang and Chiang 1998), are 10-15 km and 8-12 km, respectively.The old sediment (Vp == 3-4.5 km/sec) and the upper crust (Vp == 4.5-5.5 km/sec) beneath the Hoping Basin is thickening westward, which may be attributed to either the westward convergence of the PSP or the collision of the Luzon Arc and the Ryukyu Arc.The thickness of the oceanic layer (Vp == 4. 5-7.75 km/sec) southeast of OBS 31 in the EW9509-23 profile (Yang and Wang 1998) is 5-7 km, whereas the crust northwest of OBS 31 is thickening northwestward.The velocity varia tion of the sediment and the upper crust below OBS stations 27 and 28 imply that the eastern boundary of the Luzon Arc is between stations 28 and 29.Low velocity zones in the sedimen tary layers and in the upper crust (Vp < 7 km/sec) are easily identified below OBS stations 30, 32 and 33.
The P-wave velocity profiles introduced above as well as the horizontal components of the OBS data acquired in the 1995 experiment have previously been used to construct the Poisson's ratios of three OBS profiles (EW9509-1, EW9509-16 and EW9509-23).Prelimi nary profiles of the Poisson's ratios were proposed by Wang (1997) and Wang et al. (1998) elsewhere to investigate fracture zones of the tectonic motions off eastern Taiwan.However, due to the inadequate constraints on their structural interfaces, those earlier results failed to determine the specific boundaries of the fracture zones.In this study, the converted boundaries of the shear waves can be properly evaluated from the well-constrained interfaces of four P wave velocity profiles.Reliable Poisson's ratios are thus obtained in this way and are used to identify fractures and porosities of various tectonic structures off eastern Taiwan.

PHASE IDENTIFICATION OF OBS DATA
One vertical component and two horizontal components of 27 OBS data sets were re �orded off eastern Taiwan in the 1995 experiment.Although not for EW9509-16 profile, OBS stations on the other three profiles were also equipped with hydrophones to receive the com pressional component of the OBS data.Data processing prior to the identification of the arriv als includes station relocation, the generation of formatted OBS data and signal enhancement (Christeson 1995;Henkart 2000).Polarization or separation of multi-component OBS data for the enhancement of the converted shear waves (Digranes et al. 1998) is not implemented because P arrivals in the vertical and compressional components appear much earlier than the converted arrivals in the horizontal components of most OBS data.To demonstrate the pick ing of converted arrivals, this paper presents one of the horizontal components of the OBS data from stations 8, 17, 22 and 33 and the vertical component of OBSs 8 and 33.These four OBS stations, at or near the intersection of the seismic profiles, received signals from their own profile at 2-D acquisition.
In comparison with the vertical component, the horizontal components of the OBS data show clear and first arrivals of the converted shear waves.Figure 2 shows the vertical and horizontal components of the OBS data from station 8 located on the Nanao Basin along EW9509-1.observed from the horizontal component of OBS 8.The absence of converted reflections in the sediment is prevalent in the horizontal components of all OBS data, and this may be attrib uted to interference with the P-waves in the shallow depth and multiples in the fine strata.
The horizontal component of OBS 22 on the Nanao Basin along EW9509-14 (Fig. 3a) similarly displays clear first arrivals of converted refractions through the crust (Sg).Uncer tainties about later arrivals of converted shear waves in the data of OBSs 22 and 8 are about O.
1 sec, which is higher than those in the data of other stations.However, these later and con verted arrivals can be identified by overlapping th� calculated arrivals on the horizontal com ponents of the data.Crustal Poisson's ratios beneath the Nanao Basin are thus constrained by these identified later arrivals on the horizontal components.

SE SE
In this paper, we apply the trapezoidal profiles determined from grids of the velocity and the interface in which the Poisson's ratio is constant in each trapezoid (Zelt and Smith 1992).
In general, the V, I VP ratio increases as the Poisson's ratio decreases in the deeper section.By considering the P-wave velocity model and the trapezoidal model of the Poisson's ratio, the S wave velocity model can be defined.
Since the interface of the P-S conversion is difficult to determine in advance, a layer stripping forward modeling of the converted travel-t ime is implemented by adjusting the Poisson's ratios (within the ranges previously mentioned) and the converted interfaces from the upper sediment to the lower crust gradually.We find that most of the P-S conversions of the first arrivals clearly observed in our OBS data.take place at the sedimentary interfaces (Kodaira et al. 1996).These first arr ivals also travel as refractions with S-wave velocity within the sedimentary layers and the upper crust as illustrated by the dashed lines in Figs.6a and 7a.
Two factors may explain the P-S conversion at the sedimentary interfaces for first and re   2000).Therefore, we take into account the converted interfaces of the sediment and the upper crust for forward modeling most of the first and clear arrivals of the converted waves in the horizontal component of the OBS data.The Poisson's ratios of the sediment below the OBS stations can then be well constrained by the clear arrivals of these converted refractions.Fig ures 6b and 7b demonstrate the good travel-time fits and the low RMS errors (0.059 and 0.084 sec) of the first arrivals along EW9509-l and EW9509-14, respectively.The P-S conversion at the sedimentary interfaces for the first arrivals also suggests that reliable interfaces of the sediment in the P-wave velocity model are essential for modeling the Poisson's ratios in the sediment.
To determine the crustal Poisson's ratio, the arrivals of the P-S conversions at the crustal interfaces are similarly required for travel-time modeling.Most of the later arrivals traveling as shear waves in the horizontal components of the OBS data in this study are found as crustal reflections.The prevalence of the crustal reflections as the later arrivals can be justified from the converted and reflected rays of eight OBSs along EW9509-1 in Fig. 8a.The lack of refrac tions converted at the crustal interfaces may be due to the weak energy of the converted waves traveling the long distance or because the travel time of the converted waves is longer than the recording time of the OBS data.
Complete sets of converted interfaces with various Poisson's ratios in the allowable ranges are tested in order to calculate the corresponding travel times.Calculated arrivals superim posed on the horizontal components of the OBS data then may help us to identify the proper phases of the converted signals on OBS data.Finally, we adjust the Poisson's ratios to better fit the travel times of the specified converted phases until the total RMS error of the travel times is less than 0.1 sec as shown in Fig. 8b.The maximum error of 0.1 sec is set in the travel time modeling because it is about the maximum value of uncertainty in the travel-time picking.) is the lowest of all Poisson's ratios (0.33-0.35) of the lower sediment in the rest of the EW9509-14 profile (Fig. 11 ).This low Poisson's ratio may also be attributed to more consoli dated sediments and thus good sedimentary subsidence in the Nanao Basin .Crustal Poisson's ratios in Fig. 11 also show a sharp contrast of the Poisson's ratios (from 0.25 eastward increas ing to 0.32 in the shaded trapezoids) in the upper crust beneath the eastern portion of the Nanao Basin (below OBSs 23 and 24) and anomalously high Poisson's ratios (0.27 for the middle crust and 0.24 for the lower crust in the shaded trapezoids) in the crust beneath the Nanao Basin (below OBSs 22�24).The converted shear rays along EW9509-16 in Fig. 12a show that the coverage of shear rays (and thus constraints of the Poisson's ratios) in the sediment and in the crust is restricted below OBS stations and OBSs 16-18, respectively.The travel-time fits of all arrivals except for the easternmost arrivals are generally good with a total RM S error of 0.078 sec (Fig. 12b).

POISSON'S RATIO AND ITS TECTONIC IMPLICATIONS
As shown in Fig. 13, the Poisson's ratios of the sediment (0.43-0.45) in the northern edge of the Nanao Basin (below OBSs 16 and 17) are higher than the Poisson's ratios of the sediment Poisson's ratio of the oceanic crust in the Huatung Basin and along EW9509-l may be af fected by the Tai tung Canyon fault zone from the Gagua Ridge (Hsu et al. 1998;Kao et al. 2000), whereas those beneath the Nanao Basin may be inferred from the northward movement of the Luzon Arc and the Gagua Ridge as mentioned before.However, along EW9509-23, the average Poisson's ratio of 0.3 in the oceanic crust is much higher than those along other profiles.
Velocity anisotropy of the shear waves (Mjelde et al. 1995) beneath the Huatung Basin may be associated with the difference of the Poisson's ratios in the oceanic crust along EW9509-1 and EW9509-23.

CONCLUSIONS
In this study, most of the first arrivals in the horizontal components of the OBS data are  In this paper, sedimentary subsidence in the Nanao Basin and in the Ryukyu Trench is inferred from low Poisson's ratio (0.31) of the lower sediment.The tectonic boundaries among the accretionary prism, the continental crust and the oceanic slab in the Ryukyu forearc region are further identified from differences of the Poisson's ratios which are 0.28-0.3,0.25-0.29 and 0.23-0.33,respectively.As for tectonic motions beneath the Nanao Basin, since the north ward subduction of the PSP results in forearc compression and a fracture zone near the Nanao Basin, relatively low Poisson's ratios in the northern portion of the Yaeyama accretionary prism (0.28) and in the southern portion of the Ryukyu Arc basement (0.25) are found along EW9509-l.Furthermore, since the northward movement of the Luzon Arc and the Gauga and 31 may also be related to the arc-continent collision.

Fig. 1 .
Fig. 1.Bathymetry, tectonic features and locations of the four OBS/MCS lines, namely EW9509-1, EW9509-l 4, EW9509-16 and EW9509-23, off east ern Taiwan.Solid circles indicate the location of the 27 OBS stations in the experiment.Contour interval of the bathymetry is 1,000 meters.

Fig. 2 .
Fig. 2. (a) Vertical component and (b) horizontal component of the OBS data and picked/identified arrivals (solid lines) from OBS station 8 along EW9509-1 in the Nanao Basin.

First
Fig. 3. (a) Horizontal component of the OBS data and picked/identified arrivals (solid lines) from OBS station 22 along EW9509-14 in the Nanao Basin.
Fig. 4. (a) Horizontal component of the OBS data and picked/identified arrivals (solid lines) from OBS station 17 along EW9509-16 at the eastern edge of the Hoping Basin.(b) Enlarged section at the western far-offset of (a) showing the converted arrivals reflected from the crustal interface (PcS) and the Moho (PmS).(c) Enlarged section at the eastern far-offset of (a)showing the converted arrivals reflected from the crustal interface (PcS).

Fig. 5 .
Fig. 5. (a) Vertical component and (b) horizontal component of the OBS data and picked/identified arrivals (solid lines) from OBS station 33 along EW9509-23 in the Huatung Basin.Prominent reflections from the crustal interface (PcS) can be seen from (b).

Fig. 6 .
Fig. 6.(a) Ray paths of converted shear waves traveling as the first and refracted arr ivals in the horizontal component of the OBS data along EW9509-1, and (b) observed (cross symbols) and predicted (small squares) arrivals of (a).Dashed rays (enclosed in small boxes) and solid rays in (a) corre spond to S waves and P waves, respectively.

Fig. 7 .
Fig. 7. (a) Ray paths of converted shear waves traveling as the first and refracted arrivals in the horizontal component of the OBS data along EW9509-14, and (b) observed (cross symbols) and predicted (small open squares) ar rivals of (a).Dashed and solid rays in (a) correspond to S waves and P waves, respectively.

Fig. 8 .Fig. 9 .Fig. 10 .
Fig. 8. (a) Ray paths of converted shear waves traveling as the later and reflected arrivals in the horizontal component of the OBS data along EW9509-1 and (b) observed (cross symbols) and predicted (small open squares) ar rivals of (a).Dashed and solid rays in (a) correspond to S waves and P waves, respectively.

Fig. 14 .
Fig. 14.(a) Ray paths of converted shear waves traveling along EW9509-23, and (b) observed (cross symbols) and predicted (small open squares) ar rivals of (a).Dashed and solid rays in (a) correspond to S waves and P waves, respectively.

Fig. 15 .
Fig. 15.Poisson's ratios in the trapezoidal model (dashed lines) of EW9509-23.Value in the triangular bracket is the average Poisson's ratio of the corre sponding layer.
These anomalously high Poisson's ratios in the crust be neath the Nanao Basin may reportedly imply either fluid-filled fractures (Mjelde et al. 1995) or less fractured rocks (Shaw 1994).However, since the Luzon Arc and Ryukyu Arc might have collided below OBSs 20 and 21 (Wang and Chiang 1998) and the Gagua Ridge may have subducted northward below OBSs 24 and 25 (Schnurle et al. 1998), the interpretation that there are fluid-filled micro-cracks beneath the Nanao Basin likely seems the most reasonable.