Review and New Insights on Foreland Tectonics in Western Taiwan

Transcription

1 International Geology Review, Vol. 48, 2006, p Copyright 2006 by V. H. Winston & Son, Inc. All rights reserved. Review and New Insights on Foreland Tectonics in Western Taiwan KENN-MING YANG, 1 SHIUH-TSANN HUANG, JONG-CHANG WU, HSIN-HSIU TING, AND WEN-WEI MEI Exploration and Development Research Institute, Chinese Petroleum Corporation, 1 Ta Yuan, Wen Shan, Miaoli, Taiwan Abstract Taiwan is located on the boundary between the Eurasian and Philippine Sea plates. As a result, foreland tectonics in western Taiwan can be divided into two domains: pre-orogenic extensional structures and those of the outer part of the fold-and-thrust belt that mingled with syn-orogenic normal fault reactivation. This paper proposes a synthetic model for foreland tectonics in western Taiwan, and advances possible mechanisms by which pre-existing normal faults might have affected the evolving thrust tectonics in foreland areas of western Taiwan. The sedimentary basins of pre-orogenic extensional tectonics are of two types Paleogene and Neogene which reflect two stages of continental rifting. Results from several studies have been synthesized to provide a tectonic map displaying the regional distribution of tectonic settings at different stages and the trends of normal faults in the basins. The similarity of the en echelon patterns of arrangement for both the Neogene and Paleogene tectonic and structural settings, as shown by the tectonic map, strongly suggests that the entire foreland area was influenced by regional dextral shear. We also provide a detailed description of structures in each tectonic setting, and propose a tectonic evolution model for Cenozoic basin architecture in western Taiwan. Among the pre-orogenic sedimentary basins, the Neogene ones, in which normal faults extend to the frontal areas of the fold-and-thrust belt in western Taiwan, open northeastward. Structural analysis of the thrust fault geometry indicates that, during development of the fold-and-thrust belt on the rifted continental margin in western Taiwan, the pre-existing normal faults in northwestern Taiwan were reactivated to form inversion structures of various types on different scales, depending on the angle between the strike of the normal faults and the direction of maximum compressive stress field. In southwestern Taiwan, where normal fault reactivation is absent from the eastern part of the foreland areas, pre-existing normal faults interacted with developing low-angle thrusts in the inner part of the fold-and-thrust belt. Normal fault reactivation, regardless of how it occurs, thus plays an important role in forming the deformation front of the fold-and-thrust belt. Based on this view, we propose that the orocline or tectonic arc of the island has been influenced more by normal fault reactivation than by the morphology of basement highs. Introduction ACCORDING TO the definition given by the Glossary of Geology (Jackson, 1997), foreland refers to the areas in the outer part of a mountain-building belt, and is characterized by nonmetamorphic deformation. A foreland basin system, as defined by DeCelles and Giles (1996), extends from the frontal areas of a mountain-building belt to the margin of craton and includes wedge-top, foreland basin, forebulge, and backbulge. In this paper, we follow the above definitions and regard the foreland areas in western Taiwan as equivalent to the areas covering the onshore fold-and-thrust belt (Western Foothills 1 Corresponding author; @cpc.com.tw in Fig. 1), the coastal plain outcropping with alluvial and terrace deposits, and the offshore Tungyintao, Nanjihtao, and Penghu basins (Fig. 1). The foldand-thrust belt is included because its tectonic development has been strongly influenced by preexisting normal faults. As for the offshore areas, structural cross sections (Fig. 2) (Sun, 1985; Shiao and Teng, 1991; Chou, 1999; Chou and Yu, 2002) through the Taiwan Strait show that the backbulge of DeCelles and Giles (1996) foreland basin system is almost equivalent to the linear zone connecting the Tungyintao, Nanjihtao, and Penghu basins; therefore, the entire Taiwan Strait can be considered as the western part of the foreland. Taiwan is located on the convergent plate boundary between the Eurasian and Philippine Sea plates; /06/898/ $

2 FORELAND TECTONICS, WESTERN TAIWAN 911 FIG. 1. Tectonic map of Taiwan and its adjacent areas. The map of the syn-orogenic lithotectonic belts is compiled and modified from Ho (1982), Ernst et al. (1985), and CPC (1994). In the text of this article, the onshore foreland areas cover the alluvium and terrace gravel in the coastal plain, the outer fold-and-thrust belt (OFATB), and the inner foldand-thrust belt (IFATB) in the Western Foothills. The map of the pre-orogenic Cenozoic extensional basins and their coeval basement highs and uplifts is from Yang et al. (1996a). The basic framework for division of the Paleogene and Neogene settings is from Sun (1982), and the detailed structures in the basins are from Chow et al. (1991), Hsiao et al. (1991a, 1991b), Yang et al. (1991, 1998a), Huang et al. (1993) and Fuh and Hu (1995). Note that, on this map, the Paleogene basin does not appear in the eastern part of the Taiwan Strait. This is because the Neogene settings are predominant in that area and almost overprint the Paleogene settings. Also note the en echelon arrangement for the major tectonic settings. Abbreviations: WF = Western Foothills; HR = Hsuehshan Range; BR = Backbone Range; BC = Basement Complex; LVFZ = Longitudinal Valley fault zone; CR = Coastal Range.

3 912 YANG ET AL. FIG. 2. Tectonic settings of the foreland basin system in the Taiwan Strait (from Chou, 1999). The area occupied by the backbulge of DeCelles and Giles (1996) foreland basin system is almost equivalent to the linear zone connecting the Tungyintao, Nanjihtao, and Penghu basins; therefore, the entire Taiwan Strait can be viewed as the western part of the foreland areas. its tectonic style is a concatenation of structures aligned from east to west: the island arc in the Coastal Range, the suture zone along the Longitudinal Valley fault zone, the inner mountain-building belt of high-grade metamorphism in the Basement Complex, the outer mountain-building belt of lowgrade metamorphism in the Backbone and Hsuehshan ranges, the imbricate fold-and-thrust belt in the Western Foothills, and the foreland basin (Fig. 1). The formation of such features originated from the orogeny caused by arc-continent collision which began in the Late Miocene or Early Pliocene (Ho, 1982; Ernst et al., 1985). The orogeny is still active, and the entire mountain-building belt is being gradually translated to the west (Suppe, 1981; Chi et al., 1981) and is cannibalizing the pre-orogenic terrane accumulated in the sedimentary basins located on the margin of the Eurasian plate (Covey, 1986; Teng, 1990). Numerous publications address various aspects of the orogeny, and yet the kinematics and

4 FORELAND TECTONICS, WESTERN TAIWAN 913 deformation mechanisms in arc-continent collision are still open to conjecture. Several models have been proposed (Suppe, 1984; Teng, 1990; Sibuet and Hsu, 1996), and they all refer to different models for the basin architecture that existed prior to the arc-continent collision. In addition, the interpreted geometry of the geological structures of the basement existing in the pre-orogenic sedimentary basin has strongly influenced the studies on the spatial and temporal distribution of structure styles in the mountain-building belt (Meng, 1962; Chiu, 1971; Huang, 1987; Biq, 1992; Lu, 1994; Yang et al., 1997). An essential goal for geologists is to reconstruct the pre-orogenic geohistory of the terrane (Ernst et al., 1985); however, the terrane in the mountainbuilding belt underwent extensive deformation and erosion, and it is not easy to reconstruct the tectonic evolution based on incomplete stratigraphic records. The established timing of the stages for the tectonic evolution of the South China Sea and its surrounding continental margin areas has been referred to in reconstructions of the tectonic evolution of Taiwan and its adjacent areas (Yuan et al., 1989). Ernst et al. (1985) were the first to synthesize results of various studies on the lithotectonic belts in the central and eastern part of the mountainbuilding belt and to provide a first-order model of the tectonic evolution of the continental margin in Taiwan and its adjacent areas. They proposed that the Southeast Asian continental margin has been subjected to more than one phase of rifting, drifting, and crustal accretion since the Late Mesozoic. Their work provided some constraints and intriguing issues for all succeeding attempts to develop tectonic models of the Taiwan mountain-building belt. Another approach would be to study the basin architecture using the geological and geophysical data that has been acquired in the foreland of western Taiwan, which is relatively undeformed and unaltered by the orogeny. This would provide a valuable reference framework for studies of the geology of the mountain-building belt (Wang, 1987). The main purpose of this paper is to propose a synthesis for foreland tectonics in western Taiwan. We first present a review, with some comments, of previous studies of the extensional tectonic model for western Taiwan. Several debates and differing points of view about the interaction between preand syn-orogenic foreland tectonics then are brought into discussion, based on new insights stemming partly from the results of our own study. Much of the research in this area cites the results of studies and proprietary data in internal reports of the Chinese Petroleum Corporation (CPC), which has been exploring the foreland areas in western Taiwan for almost half a century, generating a vast amount of geological and geophysical data. Many documents reviewed in this paper include public and semi-public publications by CPC. Some unpublished internal reports by CPC are also cited in this paper, where the data are crucial for reconstructing the foreland tectonics. Briefly, the foreland tectonics of western Taiwan can be divided into two domains: the structures of the pre-orogenic extensional tectonics, and those of the outer part of the fold-and-thrust belt that mingled with normal fault reactivation of the synorogenic tectonics. Because the fold-and-thrust belt has been evolving on a rifted continental margin, pre-existing normal faults would have been reactivated to form inversion structures of various types on different scales, depending on the angle between the strike of normal faults and the direction of maximum compressive stress field (Etheridge, 1986; Welbon and Butler, 1992; Sassi et al., 1993). Pre-existing normal faults also would have altered the local maximum compressive stress field and trajectory of evolving thrust, and would have strongly affected the features of the low-angle thrusts (Wiltschko and Eastman, 1983; Kraig et al., 1987; Schedl and Wiltschko, 1987; Thomas, 1990). This paper proposes possible mechanisms through which pre-existing normal faults might have affected the evolving thrust tectonics in the foreland areas of western Taiwan. Geological Structures of the Pre-orogenic Extensional Tectonics Long before CPC started to explore offshore areas in western Taiwan in the early 1970s, investigation of the geological structures in the onshore Cenozoic extensional basin was carried out based on regional gravity and magnetic surveys, as well as on data from isolated drilled wells. Stach (1957) proposed several structural features for the pre- Miocene basement in the subsurface of onshore western Taiwan: (1) the depth of the basement increases toward the east in the offshore and onshore areas in western Taiwan; (2) a local basement high, which he named the Mesozoic Basement Shelf, exists in the Peikang area in the central part of Taiwan; (3) well bore data from the shelf and the

5 914 YANG ET AL. adjacent areas indicate that Lower Miocene strata unconformably overlie the Cretaceous; (4) to the south of an E-W trending, large-scale normal fault located at the southern edge of the shelf, the depth of the basement abruptly increases toward the south; and (5) the Late Miocene basement high shifted westward to the Penghu Islands after transgression in the Plio-Pleistocene. Stach (1957) thought that the basement high had never existed in the Peikang area until the Late Miocene. However, based on more well bore data, Schreiber (1965) suggested that the basement high actually had been forming since the beginning of the Miocene, and he renamed it the Peikang Shelf. He also speculated that the base of the Miocene sedimentary section in northwestern Taiwan is the Paleogene. This opinion was supported by Sun s (1965) evolutionary stratigraphic architecture, which showed that the Miocene strata in the north of the basement high had been onlapping southward on the shelf where the base is the Cretaceous. Meng (1968) was the first to propose a schematic tectonic outline for the Neogene sedimentary basin in offshore and onshore western Taiwan. Based on some exploration information and data from some geophysical and geological studies from the Penghu area (Pan, 1967; Huang, 1967), he proposed that offshore and onshore western Taiwan are separated into two Neogene sedimentary basins by an E-W trending, ridge-like basement high composed of Mesozoic rocks. The basement high was named the Peikang Massif (Meng, 1968), or the Penghu- Peikang Basement High (hereafter referred to as Peikang Basement High for short) (Meng, 1971), and the separated sedimentary basins were named the Northern and Southern basins, respectively. After CPC started to explore offshore western Taiwan in the early 1970s, Sun (1982) proposed a Cenozoic basin architecture for offshore and onshore western Taiwan (Fig. 1). He divided the Cenozoic basins into two groups, one Paleogene and the other Neogene (Fig. 1), separated by an Upper Eocene/Lower Miocene regional unconformity. This basin architecture was established based on seismic data and thickness of sediments; several isolated basins, tectonic basement highs/platforms between the basins, and major trends for the tectonic settings in different stages were delineated. Sun (1982) grouped the Paleogene basins into two types: half-grabens formed by normal faulting, including the Tungyintao, Nanjihtao, and Penghu basins, and those formed simply by subsidence accompanied by normal faulting. The Tungyintao, Nanjihtao, and Penghu basins are located in the western part of the Taiwan Strait and line up in a left-lateral en echelon NE-SW trend. Other Paleogene basins in the eastern part of the Taiwan Strait appear to lack the prominent half-graben type structures of those to the west. Sun (1982) further observed that the major trend for the other Paleogene structures, especially the normal faults, is also NE-SW. In comparison with the Paleogene tectonic settings, the trends for the tectonic settings of the Neogene basins and basement highs are not consistent. The major normal fault zone along the northwest side of the basins trends NE-SW, whereas the normal faults in the basins strike E-W to ENE- WSW. Also, the Neogene basins developed only in the eastern part of the Taiwan Strait, in contrast to the thin Neogene strata that were accumulated after slow subsidence in the western part of the Taiwan Strait. Moreover, every Neogene basin went through crustal subsidence (from Late Oligocene to Middle Miocene) and crustal extension (Middle Miocene to Pleistocene) (Sun, 1982). Subsequent CPC projects (Yuan et al., 1989; Hu et al., 1990; Chi et al., 1994) systematically analyzed various aspects of several distinct Cenozoic basins. The results provide very useful frameworks for further research on extensional tectonics in western Taiwan and supporting data on the spatial distribution of the basins and basement highs delineated by Sun (1982) (Chow et al., 1991; Hsiao et al, 1991a, 1991b; Yang et al, 1991; Huang et al., 1993; Fuh and Hu, 1995). In this review, we synthesize results of previous studies and provide a tectonic map (Fig. 1) to display the regional distribution of tectonic settings at different stages and the trends of normal faults in the basins. Below, we provide detailed descriptions of tectonics and structures in each tectonic setting in the Cenozoic basin architecture of western Taiwan. Penghu, Nanjihtao, and Tungyintao basins These basins were grouped by Sun (1982) into the Paleogene basin (half-graben) and they line up in a left-lateral en echelon NE-SW trend. For studies on the Nanjihtao and Tungyintao basins, only well bore data from the ridges on the southeast sides of the basins are available. The structure and stratigraphy of the Penghu Basin are documented by Lee et al. (1981) and Lee (1991), Hsiao et al. (1991a) and Chen et al. (1991).

6 FORELAND TECTONICS, WESTERN TAIWAN 915 Five drilled wells penetrate to the Middle Eocene and reveal an unconformity between the Upper Eocene and Neogene strata, as well as several layers of intermediate to mafic igneous rocks. The structural map (Fig. 1), constructed based on seismic interpretation, shows that the basin, with its width increasing to the northeast, is segmented by transverse faults into three sub-basins and bounded by NE-SW trending major normal faults. The basin is a typical half-graben, and the slip along the main boundary fault initiated in the Middle Eocene (base of NN16) and formed wedge-shaped syn-rift deposits (Hsiao et al., 1991a). Seismic stratigraphy interpretation indicates that eustatic sea level change also influenced deposition in the basin (Hsiao et al., 1991a). The Nanjihtao Basin is an intact half-graben type basin (Sun, 1982; Chow et al., 1991), separated from the Penghu Uplift to the southwest by a NE-SW trending main boundary fault and bordered by the Taihsi Basin to the east and Kuanyin Uplift to the north, respectively. Only a few normal faults occur in the interior part of the basin (Chow et al., 1991); its syn-rift sequences are uptilting and thinning to the northwest. There is no well bore data to indicate the age of the syn-rift sequences; however, indirect correlation through seismic lines (Chow et al., 1991) implies that the oldest and the youngest syn-rift strata are the Paleocene and the Upper Eocene, respectively, and that the syn-rift sequences are separated from the overlying Neogene sequences by a regional unconformity. The Tungyintao Basin is bordered by the NE- SW trending Tungyintao Ridge. The map-view of the basin as described by Sun (1982) shows that its northern part does not form a closure; however, a later study by Hsiao et al. (1991b) indicates that the basin is composed of four sub-basins of half-graben type (Fig. 1) arranged in a left-lateral en echelon sequence (Fig. 1). Well bore data are not accessible; however, the ages of the syn-rift sequences in the basin can be inferred by indirect correlation through seismic lines to the drilled wells in the marginal ridge and outside of the basin (Tang and Chi, 1991). Proposed ages for the syn-rift sequences vary from the Early Paleocene (Hsiao et al., 1991b; Tang and Chi, 1991) or the Late Cretaceous (Lee et al., 1996) to the Late Eocene (Hsiao et al., 1991b; Tang and Chi, 1991) or the Early Oligocene (Lee et al., 1996). Tainan Basin This basin was regarded by Sun (1982) as a Neogene basin; its development began in the Late Oligocene and completely ended in the Late Pliocene. The Yuchu fault is the main boundary fault along its northern margin, and is the tectonic hinge line fault between the Mesozoic Basement Shelf and the deeper water area to the south. Elishewitz (1961) presented a structure map of the coastal plain based on CPC seismic data and proposed a major normal fault located to the south of the one proposed by Stach (1957). Elishewitz (1961) named the fault the Yuchu fault, or the A fault, and the one by Stach (1957) as the B fault. The A and B faults, as well as other normal faults occurring between these two major faults, strike E-W. Elishewitz (1961) speculated that the strike of the normal faults might change from E-W to NE-SW or N-S in the frontal area of the Foothills Belt to fit the regional trend of the mountain-building belt. Elishewitz s (1961) structure map was followed by later studies on subsurface structures in the coastal plain (Hsiao, 1970, 1971, 1974; Tang, 1977; Leu et al., 1985). Meng (1971) and Meng and Chou (1976) suggested that the A fault would extend offshore to the southwest and serve as the southern boundary fault of the Peikang Basement High. Since then, the A fault has been regarded as a long continuous normal fault (Tsao and Chang, 1988; Hu, 1988; Lee T. Y. et al., 1993). However, recent studies on the A fault onshore (Chow et al., 1986, 1987, 1988a) and offshore (Yuan et al., 1989; Yang et al., 1991) indicate that it is in fact a NE-SW trending fault zone that is composed of several segments of E-W trending normal faults in a left-lateral en echelon alignment (Fig. 1). In fact, the other normal faults in the interior part of the basin, like those in the onshore area, also strike E-W (Yuan et al., 1989; Yang et al., 1991). In general, the basin is bordered on the north by the Peikang Basement High and the A fault zone (Meng and Chou, 1976; Tsao and Chang, 1988; Hu, 1988; Yuan et al., 1989; Yang et al., 1991) and on the south by another basement high, the so-called Central Uplift, named by the CPC. It becomes wider to the east in the coastal plain of southwestern Taiwan (Fig. 1). The fan shape of the basin has some implications for the age of initial extension. The pivot of the opening of the basin is now at its southwestern tip. This suggests that rifting started in the northeasternmost area and propagated toward the southwest and that the onset of rifting in the offshore

7 916 YANG ET AL. area to the west cannot be earlier than the Late Miocene onset of rifting in the onshore area to the east (Tang, 1977; Chow et al., 1988b). The thickness of Pliocene sequences can be an indicator of the amount of extension. In addition, the space between normal faults can be an indicator of extensional ratio in a rifted basin (Colletta et al., 1988) i.e., narrower spacing between faults suggests more extended lithosphere. Eastward thickening of Pliocene strata and narrower average space between normal faults indicate greater extension toward the east, which is consistent with the eastward opening of the rifted basin. Profiles through the entire width of the basin display an asymmetrical sectional feature that is characterized by half-graben structures (Yang et al., 1991). The sense of the asymmetry is reversed along the strike of the basin (Yang et al., 1991). The largest dip slip of the main boundary fault on one side of the basin gradually diminishes toward the area where the asymmetry is reversed and forms a local structural high, the so-called low-relief accommodation zone of Rosendahl (1987). Therefore, the Tainan Basin actually is composed of two or three half-graben sub-basins (Yang et al., 1991) (Fig. 1), according to the concept of the subdivision of rift basins (Rosendahl, 1987; Scott and Rosendahl, 1989; Morley, 1990). Peikang Basement High Although the basement high has long been regarded as the tectonic setting that has separated the basins to the north and south since the Paleogene, the structural features of the pre-miocene strata are more complex than expected, and are not consistent with those of the basement high itself. Based on gravity and magnetic surveys, Hu (1979) postulated that several Paleogene basement highs and a depocenter coexist beneath the Peikang Basement High and along its eastern margin. Petrologic and geochronological studies by Yuan et al. (1985) on PK-1 and PC-1 wells confirmed that Paleogene igneous rocks and their related sedimentary rocks do exist beneath the basement high and its eastern margin. The structural features of the strata underlying the pre-miocene unconformity in the Peikang area have been delineated based on well bore and seismic data (Chi et al., 1988; Yang et al., 1998a). The pre-miocene strata are faulted and form a series of E-W trending horsts and grabens (Chi et al., 1988; Yang et al., 1998a). The ages of the strata underlying the pre-miocene unconformity vary and depend on structural settings; a drilled well either penetrates through a pre-miocene graben and meets the Paleogene underlying the unconformity, or it penetrates Cretaceous strata exposed on the horst underlying the unconformity. Therefore, it may be concluded that a fault contact occurs between the Paleogene and the Cretaceous underneath the Peikang Basement High; normal faulting during the Paleogene caused the Cretaceous strata to be exposed on the uptilted corner of faulted block, whereas the Paleogene strata were accumulated on the downthrown side of a normal fault, and such tectonics results in a pattern of alternatively exposed Eocene Paleocene Cretaceous underlying the pre-miocene unconformity. The pre-miocene extensional structures extend to the east and turn NNE-SSW or even N-S in the eastern margin of the basement high (Fig. 1). The areas adjacent to the Peikang Basement High went through subsidence in the basins and coeval onlapping of the Miocene strata onto the basement high from the Late Oligocene or the Early Miocene to the Middle Miocene (Schreiber, 1965; Sun, 1965; Meng, 1971; Yuan et al., 1989; Yang et al., 1996a). Therefore, the basement high, except for a few highly standing horsts (Chi et al., 1988; Yang et al., 1998a) revealed by gravity and magnetic surveys (Hu, 1979), did not exist until the Middle Miocene. Taihsi Basin In the northern margin of the Peikang Basement High, the top of the basement or the pre-miocene unconformity, which is the base of the Neogene basin, gradually deepens to the north and becomes part of the Taihsi Basin. According to the well bore data (Sun, 1982), the Paleogene strata underlying the pre-miocene unconformity vary from Paleocene in the south to Eocene in the north in this southern part of the basin. Several different models of the Paleogene basin architecture in the southern part of the Taihsi Basin have been proposed. They are all based on different structural interpretations of seismic profiles and correlations between drilled wells (Lin et al., 1988; Chen et al., 1991; Lee, 1991). For our model, we used the latest processed network of seismic profiles, as well as data from five drilled wells in the interior part of the basin, to reconstruct and analyze the structural features of the Eocene basin and proposed that the basin forms a rhombshaped rift basin composed of two sub-basins

8 FORELAND TECTONICS, WESTERN TAIWAN 917 FIG. 3. Structural map of a Paleogene basin in the northern margin of the Peikang Basement High and the southern part of the Taihsi Basin (from Yang et al., 1998b). The location of the basin is shown in Figure 1. The basin forms a rhomb-shaped rift basin composed of two sub-basins defined by normal faults respectively striking nearly N-S and NE- SW. Contour lines represent the thickness of the wedge-shaped syn-rift deposits, and show that two depocenter maxima are present. defined by normal faults respectively striking nearly N-S and NE-SW (Yang et al., 1998b) (Fig. 3). The structures and sequences in this basin can be compared with those in the Penghu Basin (Lee, 1991). Normal faulting in both basins was initiated in the Early Middle Eocene, although the age of normal faulting initiation in the Penghu Basin may have been earlier. Drilled wells penetrated mafic igneous rocks and terrestrial lacustrine deposits in both basins. The syn-rift deposits in the Penghu Basin are thicker than those in the coeval basins to its east. According to basin subsidence analysis and extension rate calculations (Yang et al., 1998b), we speculate that the Eocene basin in the southern part of the Taihsi Basin may have encountered another older rifting. In the northern part of the Taihsi Basin, the Paleogene structural features are obscured by their greater burial depth and are dominated by the Neogene basin, in which the normal faults strike nearly E-W. Some of these faults were reactivated to become high-angle thrusts or strike-slip faults as a result of the compressive stress in the areas close to the front of the onshore fold-and-thrust belt (Chen and Tang, 1993; Huang et al., 1993) (Fig. 1). The arrangement of the normal faults forms a fan-shaped basin opening to the east. Kuanyin Uplift The uplift was originally recognized as another basement high on the northern side of the Neogene basin in northern Taiwan (Schreiber, 1965; Meng

9 918 YANG ET AL. and Chou, 1976; Hu, 1979) and was named the Kuanyin Shelf by Schreiber (1965). Sun (1982) regarded the uplift as a Paleogene basement high separating the Tungyintao and Nanjihtao basins. In fact, a more recent reconstruction of basin architecture based on a dense network of seismic lines indicates that both the Neogene Kuanyin Shelf and the Paleogene Kuanyin Uplift exist in the subsurface (Chen et al., 1994). Regional tectonic mode Analysis of regional patterns of basins and normal faults can help to decipher the tectonism of basin development (Bosworth, 1986; McClay and White, 1995). The main purposes of such analysis are to unravel the changes in the direction of extension at different stages, to understand how pre-existing structural features influence the basin formation, and to determine if regional-scale tectonism could encompass all the tectonic settings. Chen and Tang (1993) and Huang et al. (1993) proposed that, in the Taihsi Basin, the directions of extension for the Paleogene and the Neogene rifting are identical. In contrast, in the Tainan Basin, the segments of normal faults along the NW margin are arranged in a left-lateral en echelon pattern in a NE- SW striking zone, whereas in the interior parts, the Neogene basins are characterized by E-W trending normal faults (Fig. 1). Because geometric constraints require that the fault block slip must be bounded by transfer faults (Lister et al., 1986), the E-W trending main normal faults and the orthogonal transfer faults in the Tainan Basin can only permit N-S movement of the faulted blocks. However, the entire basin is opened widely to the northeast in the sense that the direction of the extension is NNW- SSE (Fig. 1). This inconsistency in strain patterns is mainly due to the left-lateral en echelon pattern of discrete normal faults along the boundaries of the basin. A similar structural pattern may also appear in the Taihsi Basin (Fig. 1). The trend of the Paleogene tectonic settings in the Tainan Basin can be inferred from the temporalspatial distribution of the strata overlying and underlying the unconformity that marks the end of the development of the basin during the Paleogene; the spatial distribution of the isochronous line of the strata overlying and underlying the regional unconformity indicates a NE-SW trend of the major tectonic settings in Paleogene time (Fig. 4; Yang et al., 1996a). This suggests that the pre-existing tectonic weak zones formed during the Paleogene extensional tectonics must have affected development of the Neogene extensional basin. The leftlateral en echelon arrangement of structural settings is observed not only for normal faults in the main boundary fault zone of the basin but also for those of the two sub-basins (Fig. 1) and for those of Paleogene basins in the western part of the Taiwan Strait (Fig. 1). This coincidence of the en echelon patterns existing in both Neogene and Paleogene tectonic and structural settings strongly suggests that the entire foreland area has been influenced by regional dextral shear (Fig. 1; Yang et al., 1996a). Regional stratigraphic architecture and basin evolution Sun (1982) used three stratigraphic profiles to show the stratigraphic architecture of the foreland area but did not give any detailed analysis. Yuan et al. (1989) attempted to analyze the differential uplift and subsidence in the Tainan Basin based on its stratigraphic character. In our recent study (Yang et al., 1996a), several stratigraphic columns from the drilled wells in offshore and onshore areas of western Taiwan were compiled to construct two stratigraphic profiles that are parallel with and normal to the front of the fold-and-thrust belt, and to demonstrate basin features across the foreland area (Fig. 5). The profile normal to the front of the fold-andthrust belt shows the typical asymmetrical shape of a foreland basin; i.e., the thickness of the Pliocene and Pleistocene strata increases dramatically from the central line of the Taiwan Strait toward the front of the fold-and-thrust belt (Fig. 5A). In contrast, the thickness of the Miocene strata is rather uniform through the entire section across the foreland basin. This implies that the foreland basin apparently initiated in the beginning of the Pliocene and that the western limit of the foreland basin does not extend beyond the central line of the strait. The profile parallel with the front of the fold-and-thrust belt clearly reveals that the Penghu Basement High, the extending part of the Peikang Basement High in the offshore area, separates two Neogene basins, the Tainan and Taihsi basins, which have evolved since the Late Oligocene (Fig. 5B). The basin and basement high features are distinctive because they show variations in thickness not only of the Pliocene and Pleistocene strata but also of the Miocene strata. Stratigraphic features reveal the two coevolving Neogene sedimentary basins and, furthermore, imply that the basins were still developing

10 FORELAND TECTONICS, WESTERN TAIWAN 919 FIG. 4. Stratigraphy architecture in the Tainan Basin as demonstrated by well bore data (from Yuan et al., 1989; Yang et al., 1996a). The age of the strata underlying and overlying the regional unconformity indicates that the post-rift strata (Upper Oligocene Upper Miocene) had been onlapping from the southeast to the northwest. This implies that the predominant tectonic settings of the Paleogene rift strike NE-SW.

11 920 YANG ET AL. FIG. 5. Stratigraphy profiles (A) normal to and (B) parallel with the front of the fold-and-thrust belt demonstrate basin features across the foreland area (from Yang et al., 1996a). The locations of the profiles are shown in Figure 1. The profile normal to the front of the fold-and-thrust belt shows the typical asymmetrical shape of a foreland basin, and shows that the thickness of the Pliocene and Pleistocene strata increases dramatically from the central line of the Taiwan Strait toward the front of the fold-and-thrust belt. In contrast, the thickness of the Miocene strata is rather uniform through the entire section across the foreland basin. The profile parallel with the front of the fold-and-thrust belt clearly reveals that the Penghu Basement High, the extending part of the Peikang Basement High in the offshore area, separates two Neogene basins, the Tainan and the Taihsi basins, which have been evolving since the Late Oligocene.

12 FORELAND TECTONICS, WESTERN TAIWAN 921 FIG. 6. Stratigraphy profile in the southern margin of the Peikang Basement High, showing the variation in age of the Miocene strata underlying the unconformity between the Miocene and the Pliocene (from Chow et al., 1988b). after the eastern part of the basins became the foreland basin. It is very important to clarify the timing of the unconformities and normal faulting initiation throughout western Taiwan; this would provide key constraints for any reconstruction of the tectonic evolution of the basins. The ages of the pre-neogene strata underlying the basal unconformity of the Neogene basins vary greatly from Cretaceous to Eocene throughout offshore western Taiwan. Nonetheless, based on the spatial distribution of the ages of the Paleogene units underlying the unconformity, the strata become younger from south to north, and the magnitude of uplift during the Late Eocene and the Early Oligocene in western Taiwan increases from north to south. This conclusion is in agreement with what is revealed by the sense of angular unconformity as observed on seismic profiles in onshore southwest Taiwan (Chow et al., 1987; Yuan et al., 1989). On the other hand, Neogene strata overlying the basal unconformity become younger from both northern and southern margins of the basement high to the crest of the basement high, and the ages of the strata range from Late Oligocene in the margins of the basement high to Middle Miocene on the crest of the basement high (Fig. 5B). This indicates that subsidence happened in the marginal areas of the Penghu Basement High after the Late Oligocene and formed two Neogene basins and the basement high (Meng, 1971). In addition, onlapping strata overlying the basal unconformity in the central areas of the Tainan basin are younger, although buried deeper, than those in the southern margin of the basin (Fig. 5B), indicating that another younger tectonic movement altered the basin features (Sun, 1982). The younger tectonic movement is related to the unconformity between the Miocene and the Pliocene that occurs in the marginal areas of the basin and changes into conformity in the basin center (Fig. 5B). In the onshore area on the northern side of the basin, the time gap represented by the unconformity between the Miocene and Pliocene strata becomes smaller toward the south (Fig. 6; Tang, 1977; Chow et al., 1988b). Chow et al. (1988b) indicated that this unconformity changes into a correlative conformable surface south of the A fault. The unconformity represents the record of the latest extensional tectonics, initiated in the Middle or Late Miocene (Sun, 1982; Yang et al., 1991). Sun (1982) suggested that both margins of the basin were uplifted during this extensional tectonics with continuous subsidence in the basin center. Yang et al. (1991) further proposed that the uplifted basin margins and the subsiding basin center were tectonically due to asymmetrical rifting. The timing of the normal fault initiation of the latest extensional tectonics needs to be addressed. Across some major faults of very large displacements, both the Pliocene and Miocene strata are characterized by growth structures. This might appear to indicate that normal faulting started in the

13 922 YANG ET AL. FIG. 7. (A) Stratal correlation between drilled wells on both sides of the main boundary fault of the Tainan Basin and (B) model for the formation of a normal faulting-related unconformity (from Yang et al., 1996a). The locations of the drilled wells are shown in Figures 1 and 5. Note that the Miocene is a continuous succession of the well P in the footwall of the main boundary fault, whereas Miocene strata unconformably underlie Pleistocene units of well K in the hanging wall. Early Miocene. However, the increase in stratal thickness across the fault plane is only apparent, because on the footwall of the major normal fault, there is an unconformity representing a large time gap between the Miocene and Pliocene strata, whereas on the hanging wall, the Miocene and Pliocene strata are much less unconformable and are even locally conformable. The youngest strata underlying the unconformity on the footwall are much older than those on the hanging wall and the time gap represented by the unconformity decreases sharply or even becomes zero across the main boundary fault in the hanging wall. The stratigraphic records from two drilled wells on both sides of the major normal fault (Fig. 7A) support the argument that there was no obvious change in sedimentary facies across the major normal fault until the end of Middle Miocene when uplifting started on the footwall and resulted in the unconformity between the Miocene and the Pliocene. This indicates that the major normal faulting did not start until after the end of the Middle Miocene. This analysis supports previous speculations about the timing of initiation of the main boundary fault between the Peikang Basement High and the Tainan Basin (Stach, 1957) and of those normal faults in the interior parts of the basin (Tang, 1977; Sun, 1982). Therefore, we propose that the thickening of the Miocene strata on the hanging wall is due to the uplift of the footwall block and the coeval subsidence of the hanging wall block during normal faulting (Fig. 7B). The regional unconformity between the Paleogene and Neogene strata (Sun, 1982) is a good indicator for demarcating the stages of tectonic evolution in western Taiwan. Chiu (1976) first recognized that the unconformity in onshore western Taiwan might represent a tectonic event influencing areas much larger than the whole island. Later studies on the sea-floor spreading of the South China Sea have determined the stages, as well as their timing, for the tectonic evolution of the South China Sea and its surrounding continental margin areas (Taylor and Hayes, 1980, 1983, Holloway, 1982; Ernst et al., 1985; Ru and Pigott, 1986; Wu, 1988) (Fig. 8). The

14 FORELAND TECTONICS, WESTERN TAIWAN 923 FIG. 8. Periods of tectonic evolution in the areas surrounding the South China Sea Basin (from Ru and Pigott, 1986). The passive continental margin surrounding the South China Sea Basin has experienced three stages of continental rifting. The implications of these events for the interpretation of tectonics in the foreland areas in western Taiwan are discussed in the text. regional unconformity, which is distributed throughout the areas that surround the South China Sea and cover offshore and onshore Taiwan, has been defined, in terms of Falvey s (1974) concept, as a break-up unconformity representing the transitional stage from continental rifting to sea-floor spreading of the South China Sea (Holloway, 1982). According to the timing of tectonic evolution shown in Figure 8, the Tainan Basin and the areas to its north were subject to extensional tectonics from the Late Eocene to the Middle Oligocene, which eventually caused regional uplifting throughout western Taiwan, resulting in erosion of all Paleogene strata in the basins (Lee T. Y. et al., 1993; Chen and Tang, 1993; Tzeng et al., 1996; Lin et al., 2003) and giving rise to the regional unconformity. The other unconformity, representing the time gap from the Middle or Late Miocene to the Late Pliocene, can be correlated to that in the surrounding areas of the South China Sea and represents the latest stage of extensional tectonics in these areas (Fig. 8). According to such a basic framework of tectonic evolution, the Upper Oligocene and the Miocene strata represent post-rift sequences and the subsidence resulting from the post-rift thermal event of the crust which occurred without any prominent normal faulting. On the other hand, the Pliocene strata can be viewed as the next syn-rift sequence (Yuan at al., 1989; Yang et al., 1991; Lin et al., 2003), and the subsidence and uplifting were clearly related to normal faulting. As for the Central Uplift, it is a smaller basement high in the southern margin of the Tainan Basin and is located between the basin and the continental slope (Fig. 1). Its structural features can be recognized as the relics of normal faulting in the previous extensional tectonics. These features also influenced the deposition of transgressive basal sands in the Upper Oligocene (Yuan et al., 1989). Nonetheless, some studies on the sedimentary facies in the southern part of the Tainan Basin (Lee et al., 1991; Shaw et al., 1991, Chen, 1993) suggested that the normal faulting was very active in that area from the Late Oligocene to the Early Miocene. Lin et al. (2003) suggested that the normal faulting associated with rapid subsidence was active during the previous Late Paleogene syn-rift tectonics. They viewed

15 924 YANG ET AL. the Upper Oligocene as the earliest deposits of the post-rift sequences. In addition, a local unconformity exists within the Lower Miocene in the Central Uplift (Lee T. Y. et al., 1993; Tzeng et al., 1996; Lin et al., 2003). This local unconformity may have resulted from the prior tectonic effect of the shift in spreading in the South China Sea (Lee T. Y. et al., 1993). The character of the tectonism that allowed for accumulation of Pliocene strata in the basins in western Taiwan is another issue that needs to be discussed. The normal faulting and its associated uplifting and subsidence that are generally thought to have been initiated in the Middle or Late Miocene may have commenced at the end of Miocene and may have resulted from the lithospheric flexuring due to tectonic loading by the mountain-building belt to the east (Lee T. Y. et al., 1993). Lin et al. (2003) suggested that the normal faulting and its associated subsidence, which began in the Middle or Late Miocene were extensional features renewed in the Cenozoic extensional tectonic evolution of the surrounding areas of the South China Sea. They also suggested that the unconformity between the Miocene and the Pliocene strata in the marginal areas is the basal unconformity for the foreland basin (Chou and Yu, 2002) i.e., the Tainan Basin has been evolving into the foreland basin since the beginning of the Pliocene. The above interpretations of the unconformity actually are in contradiction to some subsidence analyses on the sequences in the Foothills Belt (Chou et al., 1994) and the Peikang Basement High (this study). According to these analyses, at the end of Miocene the distal part of the foreland basin was either in the present Foothills Belt or in the eastern part of the Peikang Basement High. This suggests that the unconformity and its following rapid subsidence through the Pliocene have no recognized connection with the tectonic loading by the mountain-building belt to the east. After considering the above analyses, we favor a model of tectonic evolution for the southern part of the foreland areas in western Taiwan, especially in the Tainan Basin and the Peikang-Penghu Basement High (Fig. 9), which also has been proposed by Yuan et al. (1989) and Yang et al. (1996a). From the Late Eocene to the Oligocene the foreland area in western Taiwan encountered differential uplifting and formed a large-scale regional unconformity. During this period, the m thick Paleogene strata were eroded on the Peikang-Penghu Basement High (Fuh, 2000), and the greater uplift to the south resulted in the complete erosion of Paleogene strata in that area. From the beginning of the Late Oligocene, the foreland area in western Taiwan encountered differential subsidence that formed two Neogene basins now separated by the Peikang- Penghu Basement High (Meng, 1971; Sun, 1982). The latest stage of extensional tectonics began in the Middle or the Late Miocene and caused uplifting in the margins of the basins and subsidence in the basin centers (Tang, 1977; Sun, 1982). In the onshore areas, extensional tectonics was interrupted during the Pleistocene by the orogeny to the east, while in the offshore areas it continued uninterrupted to the present. The uplifted margins of the basins started to subside and accumulate deposits in the Pliocene. Inasmuch as the Tainan Basin is located on the outer shelf of the Eurasian continental margin, the Pliocene strata in the basin have been prograding toward the outer shelf and continental slope, and the basal crust of the basin has been downtilting toward the South China Sea (Sun, 1982; Yu and Lin, 1991). Syn-orogenic Geological Structures in the Outer Fold-and-Thrust Belt and Coastal Plain Offshore Areas The structural map of the pre-orogenic extensional tectonics (Fig. 1) shows that, in general, both Neogene basins, the Tainan Basin to the south and the Taihsi Basin to the north, opened widely northeastward to the front of the fold-and-thrust belt. The structural features on map view imply that the extensional basins were probably connected to each other so that their formation would have caused a greatly thinned crust and an irregular tectonic setting to the east. Several models have been proposed to investigate the effects of the extensional tectonic settings on the complexity with which compressive structures developed during later orogeny. Several major differences among these models basically stem from different presumptions and concepts behind each model construction. Concept of mega-shear and wrench system for orocline development The inconsistency in the trend of some thrust faults in the outer part of the fold-and-thrust belt has long attracted the attention of local geologists (Biq, 1958; Meng, 1962, 1965). Several characteristic features of the geological structures in northwestern Taiwan have been described (Meng, 1965),

16 FORELAND TECTONICS, WESTERN TAIWAN 925 FIG. 9. Tectonic evolution of the Tainan Basin and the adjacent areas to its north (from Yuan et al., 1989 and Yang et al., 1996a). Detailed explanation is given in the text. including the en echelon arrangement of fold axes, two sets of intersectional thrusts, and the NE-SW trending thrust obliquely cutting off the fold structure with an upthrust block on the southeast side of the thrust. Meng (1962, 1967) was the first to consider the structural features as a part of a mega-shear of regional scale that was induced by compression and left-lateral shearing of the meta-

17 926 YANG ET AL. FIG. 10. Model of wrench faulting in forming two different sets of thrust faults in northwestern Taiwan (from Chiu, 1971). According to this model, the structural features in northwestern Taiwan have been formed mainly by two major factors, shear couple induced in the area adjacent to the basement highs, and horizontal compressive stress resulting in two sets of conjugate strike-slip fracture of pure shear. morphic crustal block imposed on the Foothills Belt. He also discussed the effect of the Peikang Basement High on the formation of the orocline or the tectonic arc. Chiu (1970, 1971) first proposed that the convex-northwestward salient of the fold-and-thrust belt in northwestern Taiwan developed in the area where thick sediments were accumulated between the Peikang Basement High and the Kuanyin Uplift. He also proposed that, during the development of compressional tectonics, shear deformation would have been induced in the fold-and-thrust belt between the basement highs and formed an en echelon arrangement of folds (Fig. 10; Chiu, 1971). Subsequently, detailed structural settings in the extensional tectonics that were compiled and constructed based on well bore and seismic data of CPC s exploration in 1980 in the offshore area of northwestern Taiwan revealed that the major structural settings are a set of E-W trending normal faults (cf., Fig. 1). Huang (1986, 1987) applied the model of wrench faulting to interpret the structural characteristics along the strike of the reactivated major normal faults, and he delineated shear strain due to inconsistency between the strike of pre-existing normal faults and the trend of regional compressive stress. Biq (1992) proposed an alternative model that is very similar to Meng s model (1962, 1967) for the tectonics in the onshore foreland areas. In Biq s (1992) tectonic model, the tectonic settings in the fold-and-thrust belt and the foreland areas, as well as the normal faults in the Tainan Basin, are regarded as parts of a regional wrench system. He

18 FORELAND TECTONICS, WESTERN TAIWAN 927 FIG. 11. Model of the mega-shear system in the onshore areas of western Taiwan (from Biq, 1992) and a modeled stress field under the influence of the Peikang Basement High (from Jeng et al., 1996). Note that the normal faults in the main boundary fault zone of the Tainan Basin (Yuan et al., 1989; Yang et al., 1991) were regarded in the model as a part of the mega-shear system, although the timing is different from that of the thrust faulting in the onshore areas. Also note that the modeled stress trajectories point at normal and converge toward the front of the basement high. Symbols: solid triangles = maximum stress field; open triangles = minimum stress field. suggested that the tectonic settings as a whole resulted from the so-called tectonic escape that is due to the existence of the Peikang Basement High and the Kuanyin Uplift (Fig. 11). Biq s model (1992) suggested that several ENE-WSW trending thrust faults are characterized by the left-lateral strike-slip component, in contradistinction with the previous models (Chiu, 1971; Huang, 1987), in which these thrust faults were characterized by a right-lateral strike-slip component. The model of regional wrench system and tectonic escape (Biq, 1992) was followed in interpreting nearly all the faults on the island (Hsu et al., 1997; Lee et al., 1998) and in inventing several physical models (Lu, 1994, 1995; Lu and Malavielle, 1994; Lu et al., 1998, 2002) to investigate the effects of the asymmetric indenter and/or basement highs on the development of the fold-and-thrust belt during oblique collision between the Eurasian and the Philippine Sea plates. Numerical models (Jeng et al., 1996; Hu and Angelier, 1996; Hu et al., 1996) were also proposed to investigate the effects of the Peikang Basement High on the induced stress field during the development of the fold-and-thrust belt (Fig. 11). These authors (Jeng et al., 1996; Hu and Angelier, 1996) were strongly influenced by Biq s (1992) tectonic model as they attempted to provide geological interpretations for their numerical model results. Careful examination of the results either from physical models (Lu, 1994, 1995; Lu and Malavieille, 1994) or from numerical models (Hu and Angelier, 1996; Jeng et al., 1996) reveals that some conclusive points do not favor the regional wrench

19 928 YANG ET AL. system model for the interpretation of the foreland tectonics. The physical models (Lu, 1994, 1995; Lu and Malavieille, 1994) of the asymmetric indenter obliquely colliding with the Eurasia plate show that the thrust faulting in northwestern Taiwan should have been accompanied by right-lateral rather than the left-lateral (Biq, 1992) strike-slip. Furthermore, the numerical models (Hu and Angelier, 1996; Jeng et al., 1996; Hu et al., 1997) show that the trajectories of maximum stress converge toward, and are perpendicular to, the front of the Peikang Basement High (Fig. 11). Model results suggest that the chance of forming large-scale wrench faults is much less than that of forming conjugate strike-slip faults of pure shear, and that one set of the conjugate strike-slip faults would be right-lateral, striking ENE-WSW. The numerical simulations by Hu and Angelier (1996), Hu et al. (1997), and Jeng et al. (1996) are based on the presumption that the Peikang Basement High was a relatively rigid block with its relatively steeper frontal slope facing eastward. This presumption ignores the facts that the morphology of the basement high today has resulted mainly from tectonic loading by the mountain-building belt to the east (Chou and Yu, 2002; Lin and Watts, 2002; Fig. 2) and that prior to the orogeny, the basement high was formed by differential subsidence and normal faulting on both sides of the basement high itself, without any evident tilting down to the east. Some other physical models (Lin and Huang, 1998) were performed based on the boundary condition of the less steep slope of the basement high. The model results indicate that the S-shaped fold-and-thrust belt developed right in the frontal areas of the basement high and was dominated by thrust faults. Some NW-SE trending strike-slip faults did appear in the model; these are not due to the regional mega-shear as proposed by Meng (1962) or to the wrench system as proposed by Biq (1992), but to inconsistencies in the magnitude of thrust slip as affected by the obstacle presented by the basement high (Lin and Huang. 1998). In our opinion, the existence of basement highs may have affected the development of the fold-andthrust belt in such a way that different thicknesses of the pre-orogenic sediments would have resulted in salient and recess features (Chiu, 1971; Lin and Huang, 1998; Lu et al., 1998, 2002). On the other hand, the results from the physical and numerical models (Hu and Angelier, 1996; Jeng et al., 1996; Hu et al., 1997; Lin and Huang, 1998; Lu et al., 1998, 2002) strongly imply that the existence of basement highs would not have caused any regional mega-shear tectonics and associated tectonic escape. Some paleo-stress analyses based on field studies in offshore and onshore western Taiwan (Suppe, 1985; Angelier et al., 1986; Rocher et al., 1996; Lacombe et al., 1997, 1999, 2001, 2003; Lin and Huang, 1997; Mouthereau et al., 2002) also indicate that the syn-orogenic stress field has been induced mainly by the geometry of colliding settings in the eastern part of the island, as predicted by the numerical models (Hu et al., 1996, 1997). Because the existence of the basement highs is not sufficient to explain all of the structural features in the onshore foreland areas, the effects of the pre-existing normal faults on development of the fold-andthrust belt are needed to explain the complexity of geological structures in the outer part of the foldand-thrust belt. Normal fault reactivation in the outer part of the fold-and-thrust belt Bonilla (1975) first documented the active thrust faults that have originated from reactivated normal faults in western Taiwan. Before this, the CPC geologists who had been working on the exploration of subsurface structures in the coastal plain had noticed a problem regarding how the ENE-WSW or E-W trending normal faults intersect with the structural setting striking NNE-SSW in the Foothills Belt. They speculated that the strike of the normal faults is parallel with that of the structural settings in the frontal areas and in the subsurface portion of the Foothills Belt (Elishewitz, 1961; Hsiao, 1970, 1971, 1974). Such speculation gave rise to ambiguous interpretations of the structural characteristics of the Changhua fault, the surface frontal fault of the fold-and-thrust belt. Although field geology and gravity surveys indicated that the fault is a thrust (Chuang et al., 1969; Hu and Chen, 1969; Hu, 1985), consistent with the general structural features in the Foothills Belt, seismic interpretations in the earlier days suggested that the frontal fault was a normal fault (Hsiao, 1968; Chang, 1971; Wang, 1974). Interpretation based on digitally processed seismic data reveals a high-angle normal fault underneath the low-angle Changhua thrust fault (Chen, 1978). The co-existence of a high-angle normal fault and a low-angle thrust fault is also observed in subsurface structural settings of some gas fields in northwestern Taiwan. The subsurface normal faults

20 FORELAND TECTONICS, WESTERN TAIWAN 929 FIG. 12. Tectonic map and the study areas showing the surface structures of northwestern Taiwan (from Yang et al., 1994). The normal faults in the offshore areas strike ENE and are nearly parallel to the transverse faults and the fold axes of some anticlines in the westernmost part of the onshore area. Abbreviations: CHK = Chuhuangkeng anticline; CN = Chunan anticline; CS = Chinshui anticline; CT = Chiting anticline; CTH = Chingtsaohu anticline; CTU = Chutung thrust; FTK = Futoukeng fault; HCN = Hsincheng thrust; HCU = Hsinchu fault; LCK = Luchukeng thrust; LK = Lungkang fault; PS = Paoshan anticline; PST = Paishatun anticline; SH = Sanhu anticline; TCS = Tiehchenshan anticline; THP = Touhuanping fault; YHS = Yunghoshan anticline. strike mostly ENE-WSW or E-W and either cut through intact fold structures (Kuan, 1971; Hsiao, 1982) or exist in the footwall of low-angle thrust faults (Hu and Chiu, 1984). Hsiao (1982) suggested that the subsurface normal faults that cut through intact fold structures are the pre-existing normal faults. Namson (1981, 1984) speculated that some ENE-WSW trending high-angle thrust faults on his balanced cross sections through the Foothills Belt in northwestern Taiwan originated from the pre-existing normal fault inversion. Suppe (1984) further used seismic profiles to propose balanced structures for normal faults and suggested that some normal faults have been reactivated and turned into highangle thrust faults throughout the offshore areas of northwestern Taiwan. The mapped normal fault structures in the offshore areas of northwestern Taiwan (Huang et al., 1986; Yuan et al., 1989; Huang et al., 1993) reveal the tectonic relationship of the ENE-WSW or E- W trending normal faults with the structural settings in the onshore areas (Fig. 12). Some major reactivated normal faults can be connected with major thrust faults that are striking in a similar trend (Huang et al., 1993). Because the strikes of the preexisting normal faults are oblique to that of the foldand-thrust belt, the inversion along the reactivated normal fault would be accompanied with strike-slip components (Huang et al., 1993; Lee C. I. et al.,

21 930 YANG ET AL. 1993). This also implies a possibility that the preexisting normal faults are transcurrent faults in the fold-and-thrust belt, as speculated by Hsiao (1982) (Huang et al., 1993; Lee C. I. et al., 1993). The concept proposed here suggests that the formation of the two sets of structure with different trends in the outer part of the fold-and thrust belt basically has been affected by the pre-existing normal faults. This differs greatly from previous concepts (Meng, 1962, 1965; Chiu, 1971; Huang, 1987; Biq, 1992) and assumes that the transcurrent faults and/or those structures whose strikes deviate from the main strike of the fold-and-thrust belt originated from the pre-existing tectonic discontinuities and were not caused by wrench faulting or by regional megashear. In northwestern Taiwan, transcurrent structures may be manifested not only by reactivated normal faults but also by the 3-D geometry of a single thrust fault. In a study by Hung and Wiltschko (1993) on the 3-D geometry of the San-I thrust fault, the bent fault trace on the surface was interpreted as reflecting the variation from lateral ramp to frontal ramp of the fault plane, suggesting that the geometry was formed during the development of the thrust fault itself and may not be related to any pre-existing structure. Yang et al. (1994, 1996c, 1997) integrated the concept of normal fault reactivation, the characteristics of the 3-D geometry of a thrust fault (Hung and Wiltschko, 1993), and the regional structural map of offshore and onshore northwestern Taiwan (Fig. 12) to investigate the geometric and kinematic relationships between the thrust faults that strike ENE- WSW and those that strike NNE-SSW, respectively. In general, the former are regarded as high-angle thrust faults that originated from reactivation of the pre-existing normal faults, and the latter as lowangle thrusts arranged in an imbricate thrust system in the fold-and-thrust belt (Yang et al., 1994). The formation of the ENE-WSW trending high-angle thrust faults came earlier than that of the low-angle thrust faults (Yang et al., 1994, 1996c, 1997); the development sequence for the thrust faults of different trends can be inferred from the spatial distribution of pre-existing normal faults, reactivated normal faults, and low-angle thrusts, which are located in order from west to east (Fig. 12). The E- W trending high-angle faults or reactivated normal faults were used as northern lateral ramps for the later-developing low-angle thrust faults on the northeastern end of the low-angle thrust faults, where they played the role of tear fault on the southwestern end of the low-angle thrust faults (Fig, 13A; Yang et al., 1996c, 1997). Some high-angle thrusts and normal faults appear in the footwalls of lowangle thrust faults (Fig, 13b; Yang et al., 1996c, 1997). Some pre-existing normal faults and the associated transfer faults have been reactivated together and formed another type of connection between high-angle thrust faults (Yang et al., 1996c). The normal fault reactivation is highly selective and the resultant high-angle thrusts are consistently dipping to the southeast and are accompanied with the right lateral strike-slip component, which can be deduced by strain analysis based on the angular relationship between a high-angle thrust fault and its associated fold structure in the hanging wall (Yang et al., 1996b; Fig. 14). Yang et al. (1997) proposed a tectonic model for the structural development in the outer part of the fold-and-thrust belt (Fig. 15). According this model, in the earlier stage of the development of the foldand-thrust belt, the ENE-WSW trending normal faults were reactivated and formed high-angle thrust faults, or transverse faults, intersecting with the late-forming low-angle thrust faults. During the second stage, the NNE-SSW trending low-angle thrusts started to develop, initially formed small segments between the high-angle transverse faults, and propagated laterally toward the transverse faults. Inasmuch as the translation direction for the low-angle thrust faulting was not parallel with the trends of the transverse faults, the northeast end of the low-angle thrust faults developed into lateral ramps where they met with the transverse faults, whereas their southwestern ends were terminated by other transverse faults that served as tear faults (Fig. 13A). In the final stage, the low-angle thrusts continued to propagate toward the northwest, and climbed over the transverse faults on their northeast sides through the lateral ramps (Fig. 13B). This evolution model also implies that the development of low-angle thrust faulting passed from the northeast to the southwest. Lately, the model based on normal fault reactivation (Huang et al., 1993; Lee C. I. et al., 1993; Yang et al., 1994, 1996b, 1996c, 1997) and that based on mega-shear (Lu and Malavieille, 1994; Lu et al., 1998) have been reconciled in such a way that the reactivated normal faults in northwestern Taiwan are regarded as parts of the imbricate system, in which the reactivated normal faults share a common décollement of the fold-and-thrust belt (Mouthereau

22 FORELAND TECTONICS, WESTERN TAIWAN 931 FIG. 13. Seismic profiles demonstrating various connective relationships between high-angle transverse faults and low-angle thrust faults in northwestern Taiwan. The locations of the profiles are shown in Figure 12. A. Seismic profile running through a transverse fault demonstrates different structural features on both sides of this structure. On the southwest side of the transverse fault, a thrust fault is merged with the shallow part of the high-angle fault and forms its own lateral ramp, whereas another thrust fault on the NE side of the high-angle fault is cut by the high-angle fault. B. Seismic profile running through and into the inner part of fold-and-thrust belt demonstrates that a high-angle transverse fault or thrust fault is overthrusted by a low-angle thrust fault. Note that many normal faults exist in the footwalls of the low-angle thrust faults. Plio-Pleistocene formations: CL = Cholan Formation. Pliocene formations: CS = Chinshui Shale; KC = Kuechulin Formation. Miocene formations: SF = Shangfuchi Sandstone; HP = Hopai Formation; TL = Talu Formation; TL sh = Talu Shale Member, TL ss = Talu Sandstone Member.