The geometries and kinematic evolution of 3-D inversion thrust structures have

Transcription

1 15 Yamada, Y., and K. McClay, 2004, 3-D Analog modeling of inversion thrust structures, in K. R. McClay, ed., Thrust tectonics and hydrocarbon systems: AAPG Memoir 82, p D Analog Modeling of Inversion Thrust Structures Yasuhiro Yamada 1 Fault Dynamics Research Group, Geology Department, Royal Holloway University of London, Egham, Surrey, United Kingdom Ken McClay Fault Dynamics Research Group, Geology Department, Royal Holloway University of London, Egham, Surrey, United Kingdom ABSTRACT The geometries and kinematic evolution of 3-D inversion thrust structures have been modeled using 3-D sandbox analogs of hanging-wall deformation above footwall blocks with both concave-up and convex-up listric geometries. Extension over 3-D concave-up listric detachments produced characteristic rollover anticlines and crestal-collapse graben systems that parallel the along-strike, sinusoidal or cuspate plan geometries of the detachment breakaways. Three-dimensional inversion by horizontal contraction produced asymmetric, thrust-fault-bounded, hanging-wall inversion anticlines with curved axial traces that also follow the plan-view shape of the extensional breakaways. The main detachments were reactivated during the inversion, and new, steep thrust segments propagated upward from the detachment breakaways. Shallow to moderately dipping hanging-wall back thrusts also propagated outward from the crestal-collapse graben systems. The periclinal inversion anticlines exhibited two plunge culminations above the most concave sections of the main detachment surface. Extension above a 3-D convex-up, listric detachment surface produced a hanging-wall syncline together with a narrow, complex crestal-collapse graben system. In plan view, the axes of the hanging-wall syncline and the crestal-collapse graben formed parallel to the sinusoidal detachment breakaway. Inversion of this system produced a broad, thrust-fault-bounded anticline that shows along-strike plunge culminations. The main detachment surface was reactivated, and a moderately dipping thrust propagated upward through the syninversion strata. Segmented, hanging-wall back thrusts formed subparallel to the main detachment breakaway. Vertical and horizontal sections through the completed models were used to construct 3-D synoptic models for these inversion systems. The results of the analog experiments compare well with published examples of 3-D inversion structures from petroleum basins in the North Sea, Indonesia, and Argentina. 1 Present address: Dept. of Civil and Earth Resources Engineering, Kyoto University, Kyoto, Japan. 276

2 3-D Analog Modeling of Inversion Thrust Structures 277 INTRODUCTION Inversion structures develop where there is a change in the stress field, from extension to contraction or vice versa (Williams et al., 1989). This transformation of the stress field usually causes reactivation of preexisting faults in the opposite mode as well as the development of new faults. Inversion-related structures are formed in both the hanging walls and footwalls of the reactivated faults. In most sedimentary basins, the common mode of inversion is positive inversion (Williams et al., 1989), in which preexisting extensional faults are reactivated as contractional faults (Figure 1). In this chapter, the term inversion is used for positive inversion that is, extension followed by contraction. Inversion structures have been recognized in many sedimentary basins in a variety of tectonic environments (Lowell, 1985; Cooper and Williams, 1989; Buchanan and Buchanan, 1995). In particular, inversion has been invoked to explain basement involvement in some fold-thrust belts, such as in Papua New Guinea, (Buchanan and Warburton, 1996), in the sub-andean Neuquen Basin, Argentina (Manceda and Figueroa, 1995), and in the Syrian Arc Fold Belt, Egypt (Ayyad, 1997). Inversion produces thrust structures and related hanging-wall anticlines, which are important trap-forming systems in many hydrocarbon basins, including the southern North Sea (Ziegler, 1987), Indonesia (Ginger et al., 1993; Phillips et al., 1997), and in the Neuquen and Cuyo Basins of Argentina (Manceda and Figueroa, 1995; Uliana et al., 1995). The aims of the research presented in this chapter were to develop geometric and kinematic models of 3-D inversion structures using scaled physical models. In particular, this study focused on the inversion of 3-D listric fault systems to produce structural templates for seismic interpretation of inversion systems. Previous experimental studies of inverted extensional fault systems have been highly successful in developing templates for the kinematic evolution of 2-D inversion structures (Koopman et al., 1987; McClay, 1989; Buchanan, 1991; McClay and Buchanan, 1992; Buchanan and Mc- Clay, 1991, 1992; Mitra, 1993; Eisenstadt and Withjack, 1995; McClay, 1995). In contrast, however, the nature of 3-D inversion structures is poorly understood. The research presented here developed from an earlier investigation by Buchanan (1991), who conducted a comprehensive series of 2-D inversion experiments. EXPERIMENTAL PROCEDURE A comprehensive suite of sandbox models has been used to simulate the development of inversion above both concave-up and convex-up 3-D listric fault systems. Each experiment was run at least three times to check reproducibility and to permit vertical sectioning in two directions as well as horizontal sectioning. The deformation rig (Figure 2) was designed to model the inversion structures of various 3-D listric fault geometries. These were defined by rigid footwall blocks where the fault surfaces were sinusoidal or cuspate in plan view and either concave-up or convex-up in cross section (Figure 3). Two synchronized electric motors were used to move the rigid footwall blocks at a constant displacement rate of cm 1. For extension, the footwall block was moved out from underneath the sand pack, and the motion was reversed for the inversion phase of the experiment (Figure 2). The detachment surface was composed of a series of rayon strips parallel to the movement direction. In this manner, space problems and local complexities resulting from the 3-D geometries of the footwall blocks were minimized. The deformation rig had initial dimensions of cm. Dry, cohesionless quartz sand (average grain size 190 mm) was used as the modeling material. Quartz sand has been successfully used to model brittle, uppercrust deformation in a wide variety of tectonic environments. These include extension (Horsfield, 1977, 1980; McClay and Ellis, 1987a, b; Ellis and McClay, 1988; McClay, 1990a), strike-slip (Naylor et al., 1986; Richard et al., 1995), thrust faulting (Colletta et al., 1991; Huiqi et al., 1992; Lallemand et al., 1992), salt tectonics (Vendeville and Jackson, 1992a, b; Jackson and Vendeville, 1994; Alsop, 1996), and inversion tectonics (Koopman et al., 1987; McClay, 1989; Buchanan and McClay, 1991, 1992; McClay and Buchanan, 1992; McClay, 1995). The models were constructed from alternating layers of white and colored sand that were mechanically sieved into the apparatus. Prerift strata consisted of blue, white, and black sand layers, whereas synextensional strata were red, black, and white sand layers. At the end of 10 cm of extension in the models, a uniform, 9-mmthick, green-colored postrift layer was added. During the contraction phase, synkinematic layers were added after each 1 cm of shortening, such that the inversion anticline was just covered by syninversion strata. The scaling ratio of the models to natural structures is approximately 10 5, thus, 1 cm in the model scales to 1 km in nature (McClay, 1990b). Three different 3-D listric fault geometries were analyzed in this study. Series I experiments used a fault block that was concave-up in cross section and had a sinusoidal fault trace in plan view (Figure 3a). Series II experiments had a footwall fault block that was concave-up in cross section but had a plan view shape consisting of two distinct, spoon-shaped concave elements (Figure 3b). Series III experiments had a convex-up listric fault in cross section and a sinusoidal trace in plan view (Figure 3c). During the experiments, the free top surface of the models was photographed

3 278 Yamada and McClay FIGURE 1. Schematic diagrams of (a) extension, and (b) positive inversion structure, showing synextension and syninversion growth wedges. after every 10 mm of displacement of the footwall block, in both the extension and inversion phases. Synkinematic sediments were added incrementally during both the extension and inversion phases. The completed models were serially sectioned in three orientations: two vertical and one horizontal. The scaled physical models described in this chapter have important limitations that must be borne in mind

4 3-D Analog Modeling of Inversion Thrust Structures 279 FIGURE 2. Experimental apparatus. Deformation rig configurations for extension and inversion by horizontal contraction. when applying the results to natural inversion structures. The main limitations are that pore-fluid pressures (which are important during inversion tectonics, e.g. Sibson, 1985, 1995), isotatic and flexural responses, and thermal effects, cannot be simulated in the physical models (cf. Buchanan, 1991; McClay, 1996). Despite these limitations, the scaled analog models produce geometries and kinematic pathways that can be directly compared with natural inversion systems and, as such, can provide valuable insights into inversion tectonics and the formation of basement-involved thrust systems.

5 280 Yamada and McClay FIGURE 3. Schematic diagrams of the three 3-D footwall block geometries used in this study. (a) Series I sinusoidal breakaway geometry with a listric, concave-up detachment surface; (b) Series II cuspate breakaway geometry with a listric, concave-up detachment surface; and (c) Series III sinusoidal breakaway geometry with a listric, convex-up detachment surface. EXPERIMENTAL RESULTS Series I Experiments Extension Series I experiments involved 10 cm of extension above a sinusoidal, concave-up listric detachment system followed by 10 cm of horizontal contraction. To illustrate the characteristic geometries of the hangingwall deformation above a concave-up listric detachment, the results of an extension-only experiment were analyzed for Figure 4. During extension above a concaveup listric detachment fault, the hanging wall formed a classic rollover anticline together with a crestal-collapse graben system (Figure 4). In 3-D, the rollover anticline showed little variation along strike, with a well-developed wedge of synextension sediments forming a large halfgraben structure together with planar antithetic and convex-up synthetic faults in the crestal-collapse graben (Figure 4). Greatest growth of the synextensional layers occurred where the main detachment surface was strongly concave in plan view, whereas least growth occurred above the salient in the main detachment surface (Figure 4). After 10 cm of extension, the upper surface of the model showed strongly segmented crestal-collapse graben faults in a sinusoidal pattern that was subparallel to the main detachment breakaway (Figure 5a). The dominant trends of the extensional faults in the crestalcollapse graben were at a high angle to the extension direction, except for local small, extensional faults around the concave embayment in the main detachment surface (Figure 5a). Small, low-displacement reverse faults also formed above the salient in the main detachment surface (Figures 4 and 5a). Broadly similar rollover anticlines and crestal-collapse graben systems have been well described from 2-D models of concaveup listric detachment systems (McClay and Ellis, 1987a; Ellis and McClay, 1988; and McClay 1989, 1990a). Inversion After 10 cm of extension, the Series I models were inverted by horizontal contraction (Figure 2). The main detachment fault was reactivated as a thrust fault and propagated upward, initially through the postextension layer and then through the syninversion layers as they were progressively added to the model (Figure 5b f ). As was revealed in serial cross sections through the inverted model, this fault formed a new, high-angle thrust that cut through the postextension and syninversion layers and propagated at 608, which is the initial cutoff angle of the detachment breakaway (Figure 6). The main characteristics of the inversion structure were the steep reverse fault of the reactivated detachment surface, an asymmetric anticline in the hanging wall above this thrust, and segmented, hanging-wall-vergent back thrusts that propagated upward out of the original crestal-collapse graben system. The inversion anticline was strongly asymmetric, with a gently dipping backlimb and a steeply dipping forelimb (Figure 6). Progressive evolution of the inversion structures, as was revealed by successive plan views of the upper surface of the model, showed the rapid reactivation and upward propagation of the main detachment surface together with the formation of the asymmetric hangingwall anticline (Figure 5b, c). Segmented, hanging-wallvergent back thrusts formed early in the inversion history and became more linked as the amount of contraction

6 3-D Analog Modeling of Inversion Thrust Structures 281 FIGURE 4. Series I, Experiment L68 Extension. Serial vertical sections through a completed, concave-up, listric, extension-only model, showing a characteristic rollover anticline and crestal-collapse graben system. Note that reverse faults formed during extension (Section 30). The section numbers correspond to the distance in centimeters from a model s margin. increased (Figure 5 d f ). Small-displacement extensional faults formed on the crest of the inversion anticline and also on the steep, front limb of the anticline. In vertical cross sections, the inversion structures were remarkably similar along strike (Figure 6). The inversion anticline was slightly more asymmetric over the more concave elements of the main detachment surface and slightly tighter above the salient in the detachment surface. In all sections, back thrusts were seen to nucleate upward from the crestal-collapse graben system, and minor footwall shortcut thrusts developed in front of the main reactivated detachment surface (Figure 6). The geometry of the main inversion anticline was an arrowhead or harpoon structure, as has been classically described for typical inverted extensional half grabens (Badley et al., 1989; McClay, 1995). The inversion anticline characteristically had thin syninversion strata on the crest and thicker syninversion strata on the flanks (Figure 6). At the end of the inversion phase of deformation, the listric detachment fault remained just in net extension (<1 cm) where it was measured at the top of the pre-extension sequence (Figure 6). Horizontal sections through the completed models showed that the uplift was not uniform during inversion, with the main hanging-wall anticline showing a

7 282 Yamada and McClay FIGURE 5. Series I, Experiment I304 Inversion. Line diagrams of the top surface of the model: (a) at the end of 10 cm of extension (preexisting faults before inversion); (b) after 2 cm of horizontal contraction (initial stage of fault reactivation); (c) after 4 cm of horizontal contraction; (d) after 6 cm of horizontal contraction; (e) after 8 cm of horizontal contraction; and (f) end of the experiment after 10 cm of horizontal contraction. The dotted area represents the characteristic flat surface in the hanging wall, just next to the main thrust.

8 3-D Analog Modeling of Inversion Thrust Structures 283 FIGURE 6. Series I, Experiment I309. Serial vertical sections through the central sectors of a completed concave-up, listric inversion experiment, showing the characteristic harpoon structure, reactivated detachment fault and thrust fault, as well as the back-thrust system. Note that the amount of uplifting during inversion varies. The section numbers correspond to the distance in centimeters from a model s margin.

9 284 Yamada and McClay strongly sinusoidal axial trace and two marked plunge culminations (Figure 7). Cross section I304-5 was cut at the top of the footwall block and showed the complexity of the extensional faulting associated with both the crestal-collapse graben and with the strongly concave elements of the main detachment fault surface (Figure 7a and b). Contractional faults (except for the main detachment surface) were difficult to distinguish at this level (Figure 7b). Section I304-7, taken higher in the model, had a simpler fault pattern, with the main detachment thrust forming a clearly defined thrust footwall to the asymmetric inversion anticline. Segmented and linked back thrusts occurred on the right-hand side of the model, whereas convex-up, footwall shortcut thrusts formed beneath the main detachment thrust surface (Figure 7c and d). Series II Experiments Extension Series II experiments involved 10 cm of extension above a doubly cuspate concave-up listric detachment system, followed by 10 cm of horizontal contraction. As in Series I experiments, extension produced classic listricfault-bounded half grabens with classic rollover anticlines and crestal-collapse geometries. In cross-sectional view, they were almost identical to the vertical sections shown in Figure 4. In plan view, the extensional architecture was characterized by two well-developed crestalcollapse grabens associated with each concave element of the main detachment system (Figure 8a). A less well developed crestal-collapse graben formed above the sharp salient in the main detachment surface. The extensional faults defining these crestal-collapse systems were strongly segmented but overall were subparallel to the plan view trace of the main fault surface (Figure 8a). Small, low-displacement reverse faults also formed above the sharp salient in the main detachment surface (Figure 8a). Inversion After 10 cm of extension, the Series II models were inverted by 10 cm of horizontal contraction. At the early stages of inversion, the main detachment fault propagated upward at a constant angle of 608 (the same angle as the extensional breakaway) (Figure 8b). An asymmetric hanging-wall inversion anticline developed with a steep forelimb and a gently dipping backlimb. The axial trace of the anticline was parallel to the plan view of the main detachment surface. At 4 cm of contraction, shallowly dipping back thrusts developed on the upper surface of the model (Figure 8c). These increased in strike length with increased shortening and had traces that were relatively straight compared with the main detachment thrust fault (Figure 8d and e). Small-displacement shortcut thrusts developed in the footwall to the main detachment fault, after 6 cm of shortening (Figure 8d). At the end of 10 cm of contraction, the inversion was characterized by a major steep thrust (the reactivated detachment surface) with an asymmetric hanging-wall anticline and shallow to moderately dipping, hangingwall-vergent back thrusts on the gently dipping anticline limb (Figure 8f ). Small-displacement, slightly convex-up footwall shortcut faults developed in front of the main thrust fault. In cross section, the inversion profiles of Series II experiments (Figure 9) were almost identical to those of Series I experiments (Figure 6). In vertical cross sections, the inversion structures were remarkably similar along strike (Figure 9). The inversion anticline was slightly more asymmetric over the more concave elements of the main detachment surface and less well developed over the sharp salient in the detachment surface. As occurred in Series I experiments, back thrusts were seen to have nucleated upward from the crestalcollapse graben system. Minor, convex-up footwall shortcut thrusts developed in front of the main reactivated detachment fault (Figure 9). The geometry of the main inversion anticline was an arrowhead or harpoon structure, with thin syninversion strata on the crest and thicker syninversion strata on the flanks. At the end of the inversion phase of deformation, the listric detachment fault remained just in net extension (<1 cm), as was the case in Series I experiments (Figure 9). The two horizontal sections in Figure 10 show that the maximum inversion uplift was focused in the immediate hanging wall of the main detachment system s maximum concave-up portions. Section I314-5, taken at the level of the top of the footwall block, showed mainly the synextensional fault arrays, with complex patterns of crestal-collapse faulting above the main concave parts of the main detachment system (Figure 10a and b). Section I314-7, taken through the inversion anticline, showed two doubly plunging, anticlinal culminations focused on the most concave parts of the reactivated main detachment surface (Figure 10c and d). Small-displacement extensional faults developed on the crest of the anticlines. As is revealed in the vertical cross sections, these inversion anticlines were strongly asymmetric, with maximum uplift and maximum inversion focused in the immediate hanging wall of the main footwall vergent thrust surface (Figure 9). Series III Experiments Extension Series III experiments involved 10 cm of extension above a sinusoidal convex-up listric detachment system, followed by 10 cm of horizontal contraction. The extensional geometries produced by hanging-wall

10 3-D Analog Modeling of Inversion Thrust Structures 285 FIGURE 7. Series I, Experiment I304. Horizontal sections and interpretations: (a) Photograph of section 5 cut at 10 cm above base of the model at the position of the top of the footwall block; (b) line diagram interpretation of (a); (c) photograph of section 7 cut at 14 cm above the base of the model; and (d) line diagram interpretation of (c). Section 5 shows that most preexisting faults before inversion were preserved (cf. Figure 5a). The inversion anticline has two marked culminations (Section 7).

11 286 Yamada and McClay FIGURE 8. Series II, Experiment I313 Inversion. Line diagrams of the top surface of the model: (a) At the end of 10 cm of extension (preexisting faults before inversion); (b) after 2 cm of horizontal contraction; (c) after 4 cm of horizontal contraction; (d) After 6 cm of horizontal contraction; (e) after 8 cm of horizontal contraction; and (f) end of the experiment, after 10 cm of horizontal contraction. The dotted area represents the characteristic flat surface in the hanging wall, just next to the main thrust.

12 3-D Analog Modeling of Inversion Thrust Structures 287 FIGURE 9. Series II, Experiment I312. Serial vertical sections through the central sectors of a completed concave-up, listric inversion experiment, showing the characteristic harpoon structure, reactivated detachment fault and thrust fault, as well as the back-thrust system. Note the variation in hanging-wall uplift during inversion. The section numbers correspond to the distance in centimeters from a model s margin.

13 288 Yamada and McClay FIGURE 10. Series II, Experiment I314. Horizontal sections and interpretations: (a) Photograph of section cut at 10 cm above the base of the model at the position of the top of the footwall block; (b) line diagram interpretation of (a); (c) photograph of section cut at 14 cm above the base of the model; and (d) line diagram interpretation of (c). Section 5 shows that most preexisting faults before inversion were preserved (cf. Figure 8a). The inversion anticline has two marked culminations (Section 7).

14 3-D Analog Modeling of Inversion Thrust Structures 289 FIGURE 11. Series III, Experiment L72 Extension. Serial vertical sections through a completed, convex-up, listric, extension-only model, showing a characteristic hanging-wall syncline and a narrow, complex crestal-collapse graben system. deformation above a convex-up listric detachment were markedly different from those for a concave-up listric detachment. The experiment shown in Figure 11 was a model vertically sectioned, at the end of extension, to analyze the 3-D hanging-wall geometries. Above the convex-up listric detachment, a broad hanging-wall syncline of synextensional strata developed, together with a complex crestal-collapse graben system (Figure 11). Steep, planar antithetic faults defined the right-hand margin of the crestal-collapse structure, whereas shallowdipping, almost flat-lying synthetic faults define the lefthand margin of the crestal-collapse structure. Early-formed

15 290 Yamada and McClay crestal-collapse extensional faults have been strongly rotated, with some antithetic faults overturned to give an apparent reverse offset within the prekinematic layers (Figure 11). The hanging-wall syncline and crestalcollapse graben system showed little change in structural style along the strike of the model (Figure 11). In plan view, at the end of extension, the axial trace of the hanging-wall syncline was parallel to the footwall breakaway fault system (Figure 12a). The crestal-collapse graben faults were highly segmented and curved subparallel to the main detachment breakaway. Inversion After 10 cm of extension, the Series III models were inverted by 10 cm of horizontal contraction. Initial contraction of the model reactivated the main detachment surface as a thrust fault that propagated upward at a moderate angle through the postextension and syninversion layers (Figure 12b). A broad, flat-topped hanging-wall anticline developed, initially with an axial trace parallel to the main detachment breakaway (Figure 12b). With increased shortening, the main footwall-vergent thrust continued to propagate outward and upward, but the axial trace of the hanging-wall anticline became progressively straighter (Figure 12c e). After about 6 cm of shortening, a hanging-wall-vergent thrust developed on the upper surface of the model. This had only a slightly curved trace (Figure 12d). At the end of inversion, the model consisted of a broad, nearly symmetric, hanging-wall anticlinal uplift bounded by moderately dipping, footwall-vergent and hanging-wallvergent thrusts (Figure 12f ). Serial, vertical cross sections through the completed model showed almost identical inversion geometries along strike (Figure 13). The inversion structure consisted of a broad, thrust-bound symmetric anticline with thin syninversion strata on the crest and thicker syninversion strata on the flanks (Figure 13). The main detachment surface was reactivated and propagated upward as a new footwall-vergent thrust, at a constant angle of 308, through both the postextension and the syninversion strata (Figure 13). Moderately dipping hinterlandvergent back thrusts propagated outward from the crestalcollapse graben in the prekinematic layers. Maximum hanging-wall uplift occurred in the regions above the strongly concave parts of the main detachment surface. Minor bypass thrusts developed in the hanging wall next to the main thrust fault, but, because the footwallvergent thrust ramp angle was only 308, no footwall shortcut thrusts formed (Figure 13). Horizontal sections through the completed model revealed structural styles that were significantly different from those in the Series I and II experiments. Section I316-5, taken at the level of the top of the footwall block, primarily showed the synextensional crestalcollapse-graben fault systems and the main hanging-wallvergent back-thrust system (Figure 14a, b). Horizontal section I316-7, which cut through the main inversion structure, showed two major plunge culminations within the broad inversion anticline. It is noteworthy that this inversion anticline had an axial trace that was subparallel to the main sinusoidal detachment surface but was more symmetric (Figure 14c, d) than the concaveup inversion systems (Figure 10). Inversion Growth Sequences Syninversion growth strata were mechanically sieved into the models after each centimeter of contraction. This simulated uniform background sedimentation, whereby the upper surface of each growth increment was horizontal. In the models, there was no syninversion erosion of the crests of the anticlines. Growthstrata patterns showed that there was differential uplift while inversion progressed, and recorded the relative rotational patterns of the limbs of the inversion folds. Serial vertical sections through the completed models revealed growth-strata geometries for each experimental series (Figures 6, 9, and 13). In vertical cross sections, Series I III experiments produced very similar growth-strata patterns (Figures 6, 9, and 13). Thinnest growth strata were found on the crests of the hanging-wall anticlines, whereas the thickest growth strata formed on the undeformed footwall block and on the very right-hand part of the hanging wall where there was no internal deformation. In the Series I and II experiments, the thickest accumulations of growth strata were found in the region of the hanging-wall salients where there was the least vertical uplift during inversion (Figures 6 and 9). Progressive limb rotation during the growth of the inversion anticlines was revealed by outwardly fanning growth wedges on all of the backlimbs of the inversion folds (Figures 6, 9, and 13). For the concave-up detachment geometries, front-limb growth structures were not well defined, but, where they were visible, these also showed fanning growth wedges indicating progressive limb rotation. The best frontlimb growth structures were found on the convex-up experiments where fanning growth wedges were well developed (Figure 13). Figure 15 shows selected thickness maps of syninversion strata for all three series of experiments. These vertical isopach maps show that, for all three series of experiments, the greatest syninversion uplift occurred in the hanging walls immediate to the concave embayments in the main detachment surface, regardless of whether the main fault surface was concave or convexup in cross section (Figure 15). The contours also show that, for the concave-up experiments (Series I and II), the crest and the axial trace of the inversion anticline

16 3-D Analog Modeling of Inversion Thrust Structures 291 FIGURE 12. Series III, Experiment I315 Inversion. Line diagrams of the top surface of the model: (a) At the end of 10 cm of extension (preexisting faults before inversion); (b) after 2 cm of horizontal contraction; (c) after 4 cm of horizontal contraction; (d) after 6 cm of horizontal contraction; (e) after 8 cm of horizontal contraction; and (f) end of the experiment, after 10 cm of horizontal contraction. The dotted area represents the characteristic flat surface in the hanging wall, just next to the main thrust.

17 292 Yamada and McClay FIGURE 13. Series III, Experiment I311. Serial vertical sections through the central sectors of a completed convex-up, listric inversion experiment, showing the characteristic broad harpoon structure, reactivated detachment fault and thrust fault, as well as the back-thrust system. Note the variation in hanging-wall uplift during inversion. The section numbers correspond to the distance in centimeters from a model s margin.

18 3-D Analog Modeling of Inversion Thrust Structures 293 FIGURE 14. Series III, Experiment I316. Horizontal sections and interpretations: (a) Photograph of a section cut at 10 cm above base of the model at the position of the top of the footwall block; (b) line diagram interpretation of (a); (c) photograph of a section cut at 14 cm above the base of the model; and (d) line diagram interpretation of (c). Section 5 shows that most preexisting faults before inversion were preserved (cf. Figure 12a). The inversion anticline has two marked culminations (Section 7).

19 294 Yamada and McClay FIGURE 15. Maps of syninversion stratigraphy, showing vertical thickness variations around the thrust-bound anticlines. Contours are of thickness in millimeters for the lowermost five layers of the syninversion sequence. (a) Series I, Experiment I309. (b) Series II, Experiment I312. (c) Series III, Experiment I311.

20 3-D Analog Modeling of Inversion Thrust Structures 295 were very close to the main thrust fault, whereas these features were significantly further into the hanging wall for the convex-up experiments (Figure 15). The doubly plunging nature of the hanging-wall inversion anticlines was also revealed by the thickness plots, in which major plunge culminations were developed above the concave embayments in the main detachment surface. Small, second-order culminations (elevations < 5 mm) also developed opposite the convex salients in the sigmoidally shaped main detachment surfaces (Figure 15a and c). DISCUSSION In this section, the results of the analog experiments are analyzed and discussed. Three-dimensional conceptual models for inversion thrust systems are developed, and the implications for hydrocarbon systems associated with inversion thrust structures are reviewed. 3-D Extension Geometries 3-D extension above Series I concave-up listric detachment faults with a sigmoidal plan view geometry produced a classic rollover anticline and crestal-collapse graben geometry (Figure 4). Vertical serial sections showed very little structural variation along strike. The vertical cross sections from the 3-D models are very similar to cross sections from 2-D dip-slip extensional models published by Ellis and McClay (1988). Therefore, we can expect to have problems if we attempt to identify complex 3-D extension structures solely from single, or widely spaced, 2-D seismic lines. Plan view maps of the experimental crestal-collapse graben systems show strongly segmented extensional faults that are subparallel to the geometry of the main extensional breakaway (Figure 5a). Similar rollover architectures and extensional fault patterns are found for the Series II experiments (Figure 8a). In both Series I and II experiments, complex 3-D, obliquely striking reverse faults developed around the along-strike convex elements of the detachment surface to accommodate the oblique-slip extension in these regions (e.g., Figures 5a and 8a). In contrast, Series III experiments produced radically different extensional structures. Hanging-wall deformation above a convex-up listric detachment fault produced a hanging-wall syncline together with a complex crestal-collapse graben system (Figure 11). As was the case in the Series I and II experiments, there was little along-strike change in the hanging-wall geometry. In plan view, the strongly segmented, crestal-collapse extensional faults mimic the geometry of the detachment breakaway (Figure 12a). The results of the extensional phases of deformation show that the profile section of the main detachment surface has the most dramatic effects on the fault and rollover architectures. However, plan-view fault geometries are largely controlled by the along-strike shape of the main detachment surface. Local oblique extensional faults form around concave elements in the main detachment surface. 3-D Inversion Geometries All three series of experiments produced similar 3-D inversion geometries. In each model, the dominant feature was a thrust-bounded harpoon anticlinal structure that displayed along-strike plunge variations as a result of variable uplift during inversion (Figures 5, 8, and 12). In each experiment, the main extensional detachment fault was reactivated, in contraction, such that a new thrust propagated upward at the same angle as that of the upper part of the extensional fault. The thrust maintained a constant angle through the postextension and synextension strata. Where the thrust dipped at 608, shortcut thrusts developed in the footwall (Series I and II experiments; Figures 5 and 9). In the Series II experiments, however, the thrust propagated upward at 308 and no footwall shortcuts developed (Figure 13). Low-angle to moderately dipping, hangingwall vergent back thrusts developed in all models (Figures 6, 9, and 13). Typically, these nucleate from the tips of antithetic extensional faults that bound the crestal-collapse grabens in the synextensional sequence. The hanging-wall inversion anticlines are consistently asymmetric, with inclined axial surfaces and a dominant vergence toward the footwall (Figures 6, 9, and 13). They are fault-propagation folds that formed as a new thrust fault propagated upward through the postextension and syninversion strata. Concave-up listric detachments (Series I and II) produce inversion anticlines that have maximum uplift above the edge of the footwall block, narrow forelimbs, and curved, gently dipping backlimbs (Figures 6 and 9). In contrast, convex-up listric detachments (Series III) produce inversion anticlines that have maximum uplift above the most convex part of the footwall detachment surface, wide forelimbs, and gently dipping backlimbs (Figure 13). In Series III experiments, the inversion anticlines are less asymmetric, broader, and have their crests further into the hanging wall, away from the detachment breakaway. In all experiments, syninversion growth strata thin over the anticlinal crests and thicken down both the backlimb and forelimbs of the structures. The growthstrata wedges show outward-fanning geometries, which indicate progressive limb rotation during development of the inversion anticlines (Figures 6, 9, and 13). The horizontal sections through the completed models revealed the geometries of the hanging-wall inversion anticlines in detail (Figures 7, 10, and 14). In all experiments, hanging-wall uplift varied along the

21 296 Yamada and McClay strike of the detachment surface and consistently was greatest above the most concave parts of the detachment surface. The stratal patterns on these horizontal sections graphically illustrate the along-strike plunge culminations generated by differential inversion uplift. Anticlinal axial traces follow the geometry of the detachment-surface footwall breakaway (Figures 7, 10, and 14). The analog model s results demonstrate that the 3-D inversion s structural architecture is controlled by the geometry of the main listric detachment surface. Maximum inversion uplift occurs where there previously was maximum extension-generated subsidence. Progressive inversion uplift is accomplished by translation of the hanging wall back up the reactivated extensional fault and the new thrust fault that propagates from the extensional breakaway. During inversion, progressive back-rotation of the hanging wall produces fanning, syninversion growth-strata patterns. Vertical sections through all three series of experiments show that, for each model, the cross-section structures are remarkably similar along strike (Figures 6, 9, and 13). Interpretations of complex 3-D inversion structures from single or widely spaced 2-D seismic lines would be extremely difficult, and therefore 3-D seismic data would be needed. 3-D Synoptic Models of Inversion Structures The results of the analog experiments presented in this chapter have been used to construct 3-D synoptic models for the hanging-wall extension and inversion geometries above each type of detachment surface (Figure 16). In each of the models, the main inversion features are controlled by the geometry of the main detachment surface. Maximum inversion and uplift characteristically occur above the strongly concave elements in the detachment surface, thus producing doubly plunging, anticlinal culminations in the hanging wall (Figure 16). These synoptic models may be used as templates for structural interpretation of 3-D inversion systems where good 3-D seismic data are poor or absent. Series I Synoptic Models Figure 16a summarizes the 3-D geometries of extension and inversion above a sinusoidal, concave-up, listric detachment fault system. The extension phase shows the main rollover anticline and the sigmoidally shaped, crestal-collapse graben system. Note also the oblique reverse faults adjacent to the convex salient in the main detachment surface. The inversion structure is a 3-D harpoon structure dominated by a steeply dipping main thrust fault and an asymmetric, doubly plunging anticline in its hanging wall (Figure 16a). Slightly sinusoidal back thrusts occur in the hanging wall of the inversion structure. Small-displacement shortcut thrusts occur in the footwall to the main thrust fault. Series II Synoptic Models Figure 16b summarizes the 3-D geometries of extension and inversion above a doubly cuspate, concaveup, listric detachment fault system. The basic elements are similar to the Series I synoptic models (Figure 16a). The extension phase shows the main rollover anticline and the segmented, cusp-shaped, crestal-collapse graben system. Note also the minor thrust system formed at the cusp in the 3-D master-fault geometry. The inversion structure is a 3-D harpoon dominated by a steeply dipping main thrust fault and an asymmetric, doubly plunging anticline in its immediate hanging wall (Figure 16b). Segmented and slightly sinusoidal back thrusts occur in the hanging wall of the inversion structure. Small-displacement shortcut thrusts also occur in the footwall to the main thrust fault. Series III Synoptic Models Figure 16c summarizes the 3-D geometries of extension and inversion above a sinusoidal, convex-up, listric detachment fault system. The extension phase shows the main, hanging-wall syncline and the segmented, sigmoidal, crestal-collapse graben system. Note the well-developed relay ramps between overlapping extensional fault segments. Complex fault systems, including reverse faults, form between the detachment surface and the crestal-collapse graben system. The inversion structure is a 3-D harpoon dominated by a moderately dipping main thrust fault and a doubly plunging, broad anticline in the hanging wall (Figure 16c). Segmented and slightly sinusoidal back thrusts occur in the hanging wall of the inversion structure. Footwall shortcut thrusts do not occur in this model. Analog Models Compared with Natural Inversion Structures: Implications for Hydrocarbon Systems 2-D analog models of inversion structures have been successfully compared with a wide variety of natural structures (Buchanan, 1991; Buchanan and Mc- Clay; 1991; Buchanan and McClay, 1992; McClay, 1995, 1996). In particular, these studies emphasized the complexities of 2-D inversion geometries produced by simple dip-slip extension followed by dip-slip inversion. Natural examples of 3-D inversion systems, however, have not been well described in the literature, despite their importance in generating structural traps for hydrocarbon systems in basins such as the southern North

22 3-D Analog Modeling of Inversion Thrust Structures 297 FIGURE D synoptic models derived from the experimental results, showing hanging-wall extension and inversion. (a) Series I experiments. (b) Series II experiments. (c) Series III experiments.

23 298 Yamada and McClay FIGURE 16. (cont.). Sea (Ziegler, 1987; Badley et al., 1989) in the West Natuna Sea, Indonesia (Dickerman, 1993, Ginger et al., 1993; Phillips et al., 1997), the Malay Basin (Tjia, 1999), and in the Neuquen and Cuyo Basins, Argentina (Vergani et al., 1995; Uliana et al., 1995). Yamada (1999) described a 3-D example of inversion structures from the Niigata-Akita back-arc basin in the Sea of Japan and concluded that the published seismic sections of Okamura et al. (1995) were similar in structural style to the 3-D analog models of concave-up listric faults (e.g., Figures 6 10). In particular, the bounding thrust structures appeared to have propagated upward at the same angle as that of the main extensional fault system (Yamada, 1999). Similar, steep thrust-fault systems have been observed in a number of inversion structures in the West Natuna Sea, Indonesia, where important structural traps are formed in the anticlinal hanging walls above the steep inversion thrust system (Ginger et al., 1993; Phillips et al., 1997). Published examples of 3-D time slices from the Sembilang field in the West Natuna Sea (Dickerman, 1993) also show a dome structure that has developed above a convex salient in the main fault system. This structure is very similar to that found adjacent to the convex salient in the main detachment surface the Series I experiments (Figure 15a). Maps of inversion structures in the Cuyo Basin, Mendoza province, Argentina (Uliana et al., 1995) show well-developed periclinal (doubly plunging) anticlines in the hanging walls of inverted half-graben bounding faults. These are generally asymmetric anticlines bounded by a steep thrust fault. They form important structural traps for hydrocarbon accumulations (Uliana et al., 1995). Similarly, in the San Jorge Basin, Santa Cruz and Chubut provinces, Argentina, the north south-trending San Bernardo Fold Belt is characterized by numerous doubly plunging inversion anticlines (Homovc et al., 1995; Figari et al., 1999). In particular, the Perales field is a doubly plunging anticline with maximum uplift in the hanging wall above the most concave part of the inverted extensional fault-thrust system (Homovc et al., 1995), and therefore is very similar in style to the Series I experiments described in this paper. In the southern part of the Norwegian sector of the North Sea, the Valhall and Hod fields are developed in doubly plunging anticlines that occur in the hanging wall of the steeply dipping Skrubbe fault zone (Farmer and Barkved, 1999). This is an inverted Lower Jurassic extensional fault where variable uplift and outer-arc extension during inversion has produced a complex hanging-wall anticline. The field is developed in the syninversion Cretaceous Chalk reservoir, whose deposition is controlled by the crestal extensional graben systems (Farmer and Barkved, 1999). As in the Series I models, maximum uplift in the inversion structures appears to be focused in the hanging-wall sectors above the more concave elements of the steep thrust system.

24 3-D Analog Modeling of Inversion Thrust Structures 299 The discussion above highlights the marked lack of detailed analyses of natural 3-D inversion thrustbounded anticlines, despite their importance in generating structural traps for hydrocarbons in many basins. To validate and further improve the experiments, additional detailed 3-D seismic studies of inverted structures are needed for comparison with the analog models presented in this chapter. CONCLUSIONS Extension and inversion above concave-up and convex-up listric detachment surfaces were modeled using scaled sandbox experiments. Extension over 3-D concave-up listric detachments produced characteristic rollover anticlines and crestal-collapse graben systems. In 3-D, the crestal-collapse-graben fault systems are strongly segmented and parallel the along-strike, sinusoidal or cuspate plan geometries of the detachment breakaways. Three-dimensional inversion by horizontal contraction produced asymmetric, thrust-fault-bounded, hanging-wall inversion anticlines. The anticlinal axial traces mimic the plan-view shape of the extensional breakaways. The main extensional detachment faults were reactivated during the inversion. New, steep thrust segments propagated upward at a constant angle from the detachment breakaways. Shallow to moderately dipping, hanging-wall vergent back thrusts also propagated outward from the crestal-collapse graben system. The model inversion anticlines were periclinal in form, with two plunge culminations that correspond to positions of maximum 3-D concavity in the geometry of the footwall blocks. In contrast, extension above a 3-D, convex-up, listric detachment surface produced a characteristic hangingwall syncline together with a narrow, complex crestalcollapse graben system. In plan view, the axial trace of the hanging-wall syncline and the faults of the crestalcollapse graben system parallel the sinusoidal shape of the detachment breakaway. Inversion of this system by horizontal contraction produced a broad, thrust faultbounded anticline. The main detachment surface was reactivated, and a moderately dipping thrust propagated upward through the syninversion strata. The hangingwall anticline was less asymmetric compared with the concave-up systems, but it also showed similar alongstrike plunge culminations. Segmented, hanging-wall back thrusts also formed with plan view geometries subparallel to the main detachment breakaway. In all models, syninversion growth strata thin across the crest of the anticlines and are thickest either in the footwall to the main thrust fault or at the extremities of the backlimb of the inversion anticline. Vertical and horizontal sections through the models were used to construct 3-D synoptic models for these inversion systems. These may be used as templates for seismic interpretation where there is poor resolution or insufficient coverage to permit a full 3-D analysis. The analog-model results demonstrate that the 3-D inversion thrust architecture is controlled by the geometry of the main listric detachment surface that is, by basement control. Maximum inversion uplift occurs where there previously was maximum extensiongenerated subsidence. Progressive inversion uplift is accomplished by translation of the hanging wall back up the reactivated extensional fault and along the new thrust fault that propagates from the extensional breakaway. During inversion, progressive back rotation of the hanging wall produces fanning, syninversion growthstrata patterns. The results of the analog-model experiments compare well with published examples of 3-D inversion thrust structures from petroleum basins in Indonesia, Japan, Argentina, and the North Sea. ACKNOWLEDGMENTS This work is based on Y. Yamada s Ph.D. research, which was supported by the Fault Dynamics Project sponsored by ARCO British Limited, PETROBRAS U.K. Ltd., BP Exploration, Conoco (U.K.) Limited, Mobil North Sea Limited, and Sun Oil Britain. Y. Yamada also acknowledges funding from JNOC and JAPEX. K. McClay gratefully acknowledges support from BP Exploration. Brian Adams and Howard Moore are thanked for constructing and maintaining the deformation apparatus. T. Dooley and P. Whitehouse are thanked for constructive reviews. Fault Dynamics Publication No REFERENCES CITED Alsop, G. I., 1996, Physical modelling of fold and fracture geometries associated with salt diapirism, in G. I. Alsop, D. J. Brundell, and I. Davison, eds., Salt tectonics: Geological Society of London Special Publication 100, p Ayyad, M. H., 1997, Inverted basins in northeast Egypt; Geology and hydrocarbon prospectivities: Ph.D. Thesis, Univ. of Cairo, 299 p. Badley, M. E., J. D. Price, and L. C. Backshall, 1989, Inversion, reactivated faults and related structures: seismic examples from the southern North Sea, in M. A. Cooper and G. D. Williams, eds., Inversion tectonics: Geological Society of London Special Publication 44, p Buchanan, P. G., 1991, Geometries and kinematic analysis of inversion tectonics from analogue model studies: Ph.D. Thesis, University of London, 416 p. Buchanan, J. G., and P. G. Buchanan, 1995, Basin inversion: Geological Society of London Special Publication 88, 596 p.