2.a. Data

This study is mainly based on two-dimensional (2D) seismic reflection data that constrain major faults and several seismic horizons. The 2D survey of the Linhe Depression covers c. 4900 km, and was carried out over several years. The spaces between the seismic lines are from 4 × 16 km to 6 × 20 km (Fig. 3). Nine seismic profiles that are perpendicular to the strike of the major faults were selected to describe the geometry and architecture of different sags. One cross-section (S–T) that is parallel to the strike of the major faults was selected to describe the geometry of transfer zones. The direction changed to cut across the major fault (F₇) at the SE end of the cross-section (S–T). These seismic profiles display data down to 5–7 s two-way travel time (TWT). A total of 14 wells were drilled for oil and gas exploration in the Linhe depression. Seven wells that drilled down to the basement were selected to calibrate the stratigraphic sequence on the seismic reflection (Fig. 3).

Fig. 3. Structural map of lowermost Neogene seismic reflections in the Linhe Depression (for the location of the map, see Fig. 1b). A–B–C and D–E–F indicate the locations of the regional sections in combination with the outcrop and seismic sections; J–J', K–K', M–M', N–N', O–O', H–I and other lines only indicate the locations of seismic sections. F₁, F₂, F₃, F₄, TZ1, TZ2, TZ3 and TZ4 and (1)–(4) as in Figure 1c.

The outcrop data around the Linhe Depression were collected through the field investigation and measurement. Three field sections (A–B, D–E and G–H) were conducted in the study area. Thin-sections of different rock types were collected to determine the rock composition, and fossil specimens were found to correlate the stratigraphy sequence. Dips or dip angles of fold limbs and fault plane or striae were measured to construct an equal-area, lower-hemisphere stereoplot in the vicinity of cross-sections.

2.b. Methodology

A tectonic stratigraphic sequence and structural map were compiled according to the field investigation and seismic interpretation of the Linhe Depression and its adjacent regions (Figs 1, 4). Twenty-four major faults (F₁–F₂₄) were recognized to represent the boundary faults of the basin or normal faults within the depression. Four reverse faults (f₁–f₄) were identified in the front of the Langshan Belt through field investigation.

Fig. 4. Generalized stratigraphy of the Linhe Depression and its adjacent regions showing the main tectonic evolution (modified after Liu, Reference Liu1998). K₁l, K₁g, E₂w, E₃l, N₁w and N₂wl represent the Lisangou, Guyang, Wulate, Linhe, Wuyuan and Wulantuke formations, respectively. K₁l ¹ and K₁l ² denote the first and second members of the Lisangou Formation during the Cretaceous Era.

On the basis of striae measurements in the fault gouge or on fault planes and dips of fold limbs in the vicinity of the y–y' and G–H cross-sections, the equal-area, lower-hemisphere stereoplot was adapted to analyse the shortening direction (Fig. 5). According to the measurement of stratigraphic rocks, occurrence and fault azimuth/dip, the sections of A–B, D–E and G–H were compiled. In order to show the whole architecture of different sags, these three field sections were combined with seismic profiles (B–C, E–F and H–I) to form three framework sections (Fig. 6), which were used to provide insights into the relative timing of the deformation and syntectonic sequence for stepwise reconstruction, together with the fault throws in the other seismic profiles (Figs 7, 8). Residual anomalies in the gravity map around the Hanghou Sag were interpreted to show the morphological changes in the Central Fault Zone (Fig. 9). By means of the reconstruction of sequential schematic sections, two negative inversion models were built to indicate the different evolution between the Jilantai Sag and Hanghou Sag from Early Cretaceous time to the present (Figs 10, 11). In order to verify the tectonic evolution of the two sags, four analogue models were devised (Figs 12, 13) to investigate the validity of the evolution process as first proposed in Figures 10 and 11. Finally, we compared the simulation results with nature (Figs 14–20) to establish the new model for the Linhe Depression (Figs 21, 22).

Fig. 5. (a) Simplified geological map of the fold–thrust belt from the north part of the Langshan to the Setengshan. Original mapping conducted at a scale of 1:200 000 from NMBGMR (1991). Note that north is directed towards the bottom of the map (for the location of the map, see Fig. 1c). (b) Equal-area, lower-hemisphere stereoplot of data collected in the vicinity of the y–y' and G–H cross-sections. Black lines represent fault planes, blue dots represent striae measurements in the fault gouge or on fault planes, and the green squares represent the poles to the axial planes (modified from Darby & Ritts, Reference Darby and Ritts2007). The average shortening direction is NW–SE (155–335°).

Fig. 6. Cross-sections through the Linhe Depression showing the main tectonic units and their deformations (for the locations of the sections, see Fig. 1c). Fault labels as in Figure 1c. Ar – Archean; Pt – Proterozoic; J_1-2 – Middle–Lower Jurassic; K₁l ¹ – first member of Lisangou Formation; K₁l ² – second member of Lisangou Formation; K₁g – Cretaceous Guyang Formation; E_2-3 – Eocene–Oligocene; N – Neogene; Q – Quaternary.

Fig. 7. Seismic sections through the Jilantai Sag showing the subsidiary structural units and their deformations (for the locations of the sections, see Fig. 1c). Abbreviations as for Figure 6. E₂w – Eocene Wulate Formation; E₃l – Oligocene Linhe Formation. Fault labels as in Figure 1c.

Fig. 8. Seismic sections through the Hanghou Sag showing the secondary structural units and their deformations (for the locations of sections, see Fig. 1c). Abbreviations as in Figures 6 and 7. Fault labels as in Figure 1c.

Fig. 9. (a) Residual anomalies of gravity around the Hanghou Sag. (b) Interpretation of seismic section across the transfer zones (for the location of section S-T, see Figs. 9a and 3). Abbreviations as in Figures 6 and 7. N₁w – Miocene Wuyuan Formation; N₂wl – Pliocene Wulantuke Formation. Fault labels as in Figure 1c.

Fig. 10. Sequential schematic sections showing the evolution of the Jixi Bulge and Jilantai Sag from Early Cretaceous time to the present. Abbreviations as in Figures 6 and 7. N – Miocene–Pliocene undifferentiated. Arrows show the direction of the stress applied to the section.

Fig. 11. Sequential schematic sections showing the evolution of the Langshan fold–thrust belt and Hanghou Sag from Early Cretaceous time to the present. Abbreviations as in Figures 6 and 7. N – Miocene–Pliocene undifferentiated. Arrows show the direction of the stress applied to the section.

Fig. 12. (a) Model 1 set-up showing a 3D view of the deformation rig and its components for modelling the evolution of the Jilantai Sag and Jixi Bulge during Early Cretaceous–Pliocene time. (b) Sketch of the filling process of Model 1 at different stages. (c) Model 2 set-up showing a 3D view of the deformation rig and its components for modelling the evolution of the Hanghou Sag from during Early Cretaceous–Pliocene time. (d) The sSketch of the filling process of Model 2 at different stages. The red solid arrows represent the direction of compression, and the blue dashed arrows represent the direction of extension. a, b, c are plastic blocks representing the morphological changes in the forebulge zone in the basement.

Fig. 13. (a) Model 3 set-up showing a 3D view of the deformation rig and its components for comparison with Model 1. (b) Sketch of the filling process of Model 3 at different stages. (c) Model 4 set-up showing a 3D view of the deformation rig and its components for comparison with Model 2. (d) Sketch of the filling process of Model 4 at different stages. Legend as for Figure 12.

Fig. 14. Top view (photographs and line drawings) of different stages of Model 1 to show the evolution of the Jixi Bulge and Jilantai Sag from the compression stages (A) to the extension stages (B). (C) Cross-section interpretations (photographs and line drawings) at the final stage of Model 1 showing the deformation styles (for the location of the section, see Fig. 14f). Cl and Cr represent the compression from the left and right sides, respectively; El and Er represent the extension from the left and right sides; T denotes the total thickness of the sand layers; # denotes the deformation phase of the simulation process; the fault label numbers represent the sequence of the faults developed in the experiment; red arrows represent the direction of compression; and blue arrows represent the direction of extension.

Fig. 15. Top view (photographs and line drawings) of different stages of Model 2 showing the evolution of the Hanghou Sag from the compression stages (A) to the extension stages (B). (C) Cross-section interpretations (photographs and line drawings) at the final stage of Model 2 showing the deformation styles (for the location of the section, see Fig. 15i). TZ5 and TZ6 represent the transfer zones and the fault label numbers represent the order of the faults developed in the experiment. Abbreviations as for Figure 14.

Fig. 16. 3D view (photographs and line drawings) of different stages of Model 3 to show the evolution of the Jixi Bulge and Jilantai Sag from (a) the compression stage to (b, c) the extension stages. Abbreviations as for Figure 14.

Fig. 17. 3D view (photographs and line drawings) of different stages of Model 4 to show the evolution of the Hanghou Sag from (a, b) the compression stages to (c) the extension stage (c). Abbreviations as for Figure 14.

Fig. 18. Comparison of negative patterns between the models (a, b) proposed by Williams et al. (Reference Williams, Powell, Cooper, Cooper and Williams1989) and (c, d) the seismic interpretation in the Jilantai Sag. (a) Eroded thrust system; (b) extensional fault cutting cross the pre-existing thrust system with no reuse of thrust faults; (c) thrust system formed in the first stage of compression; and (d) partial reactivation of a complex thrust system in the first stage of extension. S – synrift extensional sequence.

Fig. 19. (a) Cenozoic tectonic map of lowermost Neogene seismic reflections in the Jilantai Sag showing the main structural elements. (b) Line drawing of map view of Model 1 at the end of extension (at 10.5 cm extension). (c) Cenozoic tectonic map of lowermost Eocene seismic reflections in the Hanghou Sag showing the main structural elements. (d) Line drawing of map view of Model 2 at the phase of extension (at 6.0 cm extension). Fault names in (a) and (c) as in Figure 3, and in (b) and (d) as in Figures 14f and 15f, respectively. Unit names such as ①–③ in (c) as in Figure 1c. Er represents the extension from the right side; T denotes the total thickness of the sand layers; # denotes the deformation phase of the simulation process. TZ2, TZ3, TZ5 and TZ6 indicate the transfer zones, and (1)–(6) represent the surface topography of the modelling results.

Fig. 20. Comparison between modelling results and interpretation of seismic lines across the Linhe Depression. (a) Photograph and line drawing of a cross-section in Model 1 (for location see Fig. 14f); (b) interpreted seismic line across the Jilantai Sag in the Linhe Depression (for location see Fig. 1c); (c) photograph and line drawing of a cross-section in Model 2 (for location see Fig. 15i); (d) interpreted seismic line across the Hanghou Sag in the Linhe Depression (for location see Fig. 1c). Fault names in (a) and (c) as in Figures 14e and 15j, respectively. Fault names in (b) and (d) as in Figure 1c. The stratigraphic names as in Figures 7 and 8.

Fig. 21. Schematic kinematic model showing the differential compression during Early Cretaceous time based on field observation and seismic interpretation. 1, granite gneiss; 2, schist; 3, conglomerate; 4, sandstone; 5, siltstone; 6, mudstone; 7, Archean; 8, Proterozoic; 9, second member of Lisangou Formation; 10, first member of Lisangou Formation; 11, fault labels; 12, wells. ① – Bayanwulashan fold–thrust belt; ② – Helanshan fold–thrust belt; ③ – Jixi piggyback basin (wedge-top depozone); ④ – Jidong piggyback basin (wedge-top depozone); ⑤ – Jizhong foredeep depozone; ⑥ – Langshan fold–thrust belt; ⑦ – wedge-top depozone; ⑧ – foredeep depozone; ⑨ – forebulge depozone; ⑩ – back forebulge depozone; ⑪ – Dengkou transfer zone. The arrows show the direction of the stress to the Linhe Depression.

Fig. 22. Schematic model showing the negative inversion kinematics from the differential compression during Early Cretaceous time to the SE-directed extension in the Cenozoic Era based on field observation and seismic interpretation. 1, granite gneiss; 2, schist; 3, conglomerate; 4, sandstone; 5, siltstone; 6, mudstone; 7, Archean; 8, Proterozoic; 9, second member of Cretaceous Lisangou Formation; 10, first member of Cretaceous Lisangou Formation; 11, Cretaceous Guyang Formation; 12, Eocene–Oligocene undifferentiated; 13, Miocene–Pliocene undifferentiated; 14 – transform zone; 15, fault labels; 16, wells. ① –Bayanwulashan fold–thrust belt; ② – Jixi Bulge; ③ – Jixi Subsag; ④ – Jizhong Subsag; ⑤– Jidong Subsag; ⑥ – Langshan fold–thrust belt; ⑦ – wedge-top depozone; ⑧ – Aolun Subsag; ⑨ – Central Fault Zone; ⑩ – Huanghe Subsag; ⑪ – Dengkou transfer zone; i – Song 5 fault zone; ii – Zhage fault zone; iii – Xinglong fault zone. The arrows show the direction of the extensional stress applied to the Linhe Depression.

4.a. Framework of Jixi Bulge and Jilantai Sag

The Jilantai Sag is located at the SW of the Linhe Depression, and is separated by the Chahaer Uplift from the Bayanhaote Basin in the SW and by the Dengkou transfer zone from the Hanghou Sag in the NE (Fig. 1c). A cross-section (A–B–C) was drawn through the Jixi Bulge and Jilantai Sag by combining the outcrop observation section and seismic profile interpretation (Fig. 6a). The most prominent structures on the cross-section and on the geological structure map are the normal faults dipping steeply to the SE and juxtaposely cutting through all the cover layers and basement to generate a composite half-graben in the Jilantai Sag. The entire half-graben dipping to the NW and mainly bounded by the Wulanaobao Fault (F₄) was subdivided into three subsags: Jixi, Jizhong and Jidong (Figs 6a, 7). The Cenozoic layers (E_2-3–N₂) thickened towards the boundary normal fault on the hanging wall of each half-graben, while the Guyang Formation (K₁ g) maintained almost constant thickness across all seismic sections (Fig. 7). These sedimentation characteristics indicate that these normal faults began to activate at the beginning of the deposition of the Wulate Formation (E₂ w). Inverted thrust faults are another important structure developed in the Jilantai Sag. The wedge shape of K₁ l ² on the footwall and the absence of an equivalent sequence on the hanging wall indicate that fault F₈ is a boundary fault that governs the syndeposition of K₁ l ² (Fig. 7a, b). Meanwhile, the lesser thickness of K₁ l ² in the hanging wall compared with the footwall of F₈ in the J–J' cross-section implies that the syndeposition of K₁ l ² was affected by the thrust fault. On the basis of seismic interpretation and field data from Figures 6 and 7, we find that the faults F₃ and F₅ thrust from the NW to the basin inland, and some faults (F₇, F₈ and F₉) thrust from the SE to the Jilantai Sag. We therefore deduce that the compressional stress originated from both directions because of the Bayanwulashan and Helanshan shortening during Early Cretaceous time (Liu, Reference Liu1998).

The displacement of the layers of the Lisangou and Guyang formations along the fault plane indicates that F₈ is a negative inverted fault and that the intensity of the inversion differs along the fault strike. The interpretation of section K–K' shows that the extension displacement is larger than the contraction displacement, thereby causing a complete negative inversion (Fig. 7a). In contrast, interpretation of sections B–C and J–J' reveals that the extension displacement is less than the contraction displacement, thereby causing a partial negative inversion (Fig. 7b, c). Fault F₇ has the same inversion feature as fault F₈. The thickness of K₁ l ¹ on the footwall of fault F₇ is larger than the equivalent hanging-wall succession where K₂ l ² is eroded in all sections (Fig. 7). However, F₉ can only be identified to invert negatively in section J–J' (Fig. 7c). The inverted fault F₅ differs from the other inverted faults as it dips to the NW and is cut by the master boundary fault F₄ in section B–C. It is the only seismic profile that cuts through the northwestern boundary of the Jilantai Sag (Fig. 7b). The thickness of the Lisangou Formation on the footwall of F₅ is much larger than that of the equivalent sequence on the hanging wall. The maintenance of an almost-constant thickness by the Guyang Formation indicates that the compression event occurred mainly during the deposition of the Lisangou Formation. The section (A–B) compiled from field observations indicates that the cover layers on the basement of the Jixi Bulge comprise the Guyang and Lisangou formations, along with a considerably thin Eocene sequence (Fig. 6a). Archean metamorphic rocks thrust over the Lower Cretaceous sedimentary rocks along the boundary fault (F₃) that was reactivated during the Holocene Epoch by a dextral strike-slip fault (Dong et al. Reference Dong, Zhang, Zheng, Yu, Lei, Yang, Liu and Gong2018). The conglomerate composition of local sediment sources and the coarsening towards the steep thrust fault suggest that they were synorogenic with respect to the Bayanwulashan fold–thrust belt during Early Cretaceous time. During the Cenozoic Era, the Jixi Bulge was subjected to erosion due to the subsidence of the Jilantai Sag along the master normal Wulanaobao Fault (F₄). The frameworks of the Jixi Bulge and Jilantai Sag indicated that they experienced contractional deformation during Early Cretaceous time, followed by extensional deformation during the Cenozoic Era. The current section view therefore shows a negative inversion structure.

4.b. Framework of Hanghou Sag

According to the fault patterns and deposition characteristics, the Hanghou Sag was subdivided into the Aolun Subsag, Central Fault Zone and Huanghe Subsag (Fig. 8). The prominent architecture on the cross-sections and on the structural map (in yellow in Fig. 1) corresponds to the Central Fault Zone, which separates the Aolun and Huanghe subsags. The Central Fault Zone is segmented into three parts by the Song 5, Zhage and Xinglong fault zones from the SW to the NE (Fig. 9). Each fault zone was bounded by two normal faults to form a horst at the centre. The displacement of the horst boundary faults decreased dramatically from the SW to NE (Fig. 8). The thickness of the Palaeogene and Neogene strata increases to over 10 km towards the master fault of the Langshan Fault (F₂), and the Guyang Formation remains almost constant on the cross-sections. The extension therefore seems to have initiated during Eocene time. The thickness of the Lisangou Formation on the central horst was less than that on the Aolun and Huanghe subsags. The first member of the Lisangou Formation was truncated by the Guyang Formation at the location where the thrust faults (f₅) were also truncated on the cross-sections N–N' (Fig. 8c). The thickness of the Lisangou Formation on the hanging wall was less than that on the footwall of the thrust fault (f₅). This difference indicates that the compression event occurred during the deposition of the Lisangou Formation.

The outcrop section (D–E) well exposed across the Langshan area reveals that three NW-dipping (30–50°) thrust faults (f₁, f₂ and f₃) separated the Jurassic or Cretaceous strata from Proterozoic metamorphic rocks to form the fold–thrust belt with forwards propagation to the Hanghou Sag (Fig. 6b). The thrust fault and Cretaceous units were cut through by the boundary normal fault (F₂) at the pediment of the Langshan to cause the inversion of the foreland basin (Fig. 6b). The architecture of the Langshan fold–thrust belt and Hanghou Sag illustrates that they experienced compression deformation originating from the NW direction from Early Jurassic time to the Lisangou Formation, followed by extension deformation during the Cenozoic Era. The current profile view therefore shows a negative inversion basin structure.

4.c. Framework of Wuyuan Sag

The strike of the sag and the trending of the fault system in the Wuyuan Sag switched to the E–W direction from the NE–SW direction. The prominent architecture on the cross-sections and the geological structure map is one in which the whole sag is one half-graben governed by the master Sertengshan Fault (F₁) dipping to the south (Fig. 8d). Some synthetic and antithetic subsidiary normal faults developed on the hanging wall of the master fault.

The dramatic thickening of the Neogene sequence (N₁ w and N₁ wl) towards the hanging wall of the master fault suggests that the sag experienced intensive extension at that time. The Jurassic strata on the footwall of the Sertengshan Fault (F₁) on the G–H–I cross-section were separated by a thrust fault (f₄) from the Proterozoic metasedimentary rocks and intrusion rocks on the hanging wall to form the foreland basin structure (Fig. 6c). Given the absence of cover layers on top of the Jurassic strata, the end of the contraction age since Early Jurassic time is difficult to establish.

4.d. Characteristics of transfer zones

Residual anomalies can be used to analyse the differences of texture and composition in the basements (Shi et al. Reference Shi, Shi, Wu, Gao, Yang, Zhang, Song and Chen2019). The locations with normal anomalies indicate where the igneous rocks of higher density are distributed, while the negative anomalies indicate where the sedimentary rocks of lower density develop (Shi et al. Reference Shi, Shi, Wu, Gao, Yang, Zhang, Song and Chen2019). The characteristics of residual anomalies around the Hanghou Sag show some normal anomalies distributed along the Central Fault Zone and the Langshan fold–thrust belt (Fig. 9a). The locations with normal anomalies confirmed by well data (S5, S2, L1 and Z1) or field investigations consist of granite gneiss or ultrabasic rocks of Archean–Proterozoic age (Sun et al. Reference Sun, Zhang, Hu, Ren, Wang and Zhang2018). The transfer zones of the Hanghou Sag developed in the locations where the trending of the normal anomalies changed or discontinued abruptly. A negative flower structure existing among these transfer zones indicates that they originated from the SE extension together with the formation of Hanghou Sag during Eocene–Pliocene time (Fig. 9b). The transfer zone of TZ1 separated the Jilantai Sag from the Aolun Subsag. The transfer zones of TZ2 and TZ3 did not affect the thickness of the cover layers. Because of the absence of seismic lines across the transfer zone of TZ4, it is not possible to describe its deformation characteristics. As a result of the trending change of the boundary faults and the anomalies, we deduced that a transfer zone (TZ4) should exist between the Hanghou and Wuyuan sags.

5.a. Evolution of Jilantai Sag and Jixi Bulge

We present a reconstruction that represents the evolution of both the Jilantai Sag and Jixi Bulge from the initiation of crustal shortening during Early Cretaceous time to the active extension in the Cenozoic era along the Wulanaobao Fault (F₄) on the basis of the geological features along the A–B–C section (Fig. 10).

The first stage in the reconstruction is largely based on the foreland basin strata (K₁ l ²) preserved in the field area of the Bayanwulashan fold–thrust belt and the deformation patterns on the seismic reflections of the Jilantai Sag. The key features of this stage include the deposition of two piggyback basins close to the boundary faults (F₃ and F₈) and the foreland depozone in the central part, controlled by faults F₅ and F₇ (Fig. 10a). The relatively coarse-grained conglomerate of the Lisangou Formation currently exposed along the boundary of the thrust fault (F₃) suggests that it was the initial boundary of the Jixi piggyback basin. The second member of the Lisangou Formation (K₁ l ²) present on the footwall but absent from the hanging wall of fault F₈ indicates the existence of a piggyback basin (bounded by fault F₈) in the east of the foreland depozone at this stage (Fig. 10a). The second member of the Lisangou Formation (K₁ l ²) on section J–J' was preserved on the footwall of fault F₉, thereby indicating the existence of another piggyback basin (bounded by fault F₉) in the east of the Jidong piggyback basin at this stage (Fig. 7c).

The second stage of the deformation comprised the forwards propagation of the thrust belt and the continued development of the piggyback basin. The absence of K₁ l ¹ in the SE part of the section implies that the uplifting magnitude in the SE direction was much larger than that in the NW direction. This condition caused erosion on the top of the Jidong piggyback basin and deposition of K₁ l ¹ on the top of the central foreland depozone (Fig. 10b). The superimposed deposition during the two contraction stages resulted in the largest thickness of the Lisangou Formation in the central foreland depozone. As well as the forwards thrust faults, some backwards thrust faults also developed on the hanging wall of the master thrust faults.

The uniformly depressed subsidence dominated Stage 3 deformation (Fig. 10c). The thickness of the Guyang Formation remained almost constant across the whole area with clear reflection marks in the seismic profiles. These characteristics suggest that the depressed subsidence homogeneously occurred after the contraction. The absence of Upper Cretaceous and Paleocene strata between Guyang Formation (K₁ g) and Wulate Formation (E₂ w) indicates the possible erosion after Stage 3 deposition. The original thickness of the Guyang Formation should therefore be greater than that currently preserved in the outcrop and seismic interpretation.

After Stage 3 and erosion, the foreland deposits were subjected to extension from the SE direction and perpendicular to the plane of the thrust faults to generate the half-graben architecture superimposed on the contraction structure in the Jilantai Sag at Stage 4. The master normal fault (F₄) cut off the front part of the Jixi piggyback basin along with the thrust faults (F₅) and controlled the Eocene and Oligocene strata to thicken towards the hanging wall of the boundary fault. In the absence of sedimentary marks preserved in the crystalline basement on the footwall of the master normal fault (F₄), the exact locations of the original thrust faults (F₅) on the footwall of the master normal fault (F₄) cannot be identified.

Nevertheless, we can explain why the thickness of the Lisangou Formation between F₄ and F₅ is less than that in their footwall. A partially inverted pattern is the most important structure developed along the pre-existing thrust faults (F₇, F₈) to generate new secondary structure units, such as the Jixi, Jizhong and Jidong subsags (Fig. 10d). The displacement of the late extension was less than that of the previous compression along the planes of the faults (F₇ and F₈). This difference indicates a partial negative inversion structure (Fig. 7).

The last stage (Stage 5) of deformation in the restoration process is the progressive extension from the SE direction based on the first phase extension (Fig. 10e). Numerous subsidiary normal faults with minor displacement in the cover layers developed at this stage. Some of the normal faults, such as F₆, F₇ and F₉, cut through the basement with a relatively large displacement (Fig. 7). The inversion intensity along F₈ increased from the SW to the NE, and thereby caused a complete negative inversion structure pattern in the K–K' cross-section (Fig. 7a). Strike-slip displacement occurred along the Bayanwulashan Fault (F₃) and Langshan Fault (F₂) and resulted in at least four phases of surface rupture during Holocene time (Rao et al. Reference Rao, Chen, Hu, Yu and Qiu2016; Dong et al. Reference Dong, Zhang, Zheng, Yu, Lei, Yang, Liu and Gong2018).

5.b. Evolution of Hanghou Sag and Langshan fold–thrust belt

We present a reconstruction that represents the evolution of the Langshan fold–thrust belt together with the Hanghou Sag from the beginning of crustal shortening during Early Cretaceous time through the intensive extension in the Cenozoic Era along the Langshan Fault (F₂), according to the comprehensive features of the cross-section (Fig. 11). The Hanghou Sag and Langshan frontal fold–thrust belt experienced the same evolutionary stages as those that occurred in the Jilantai Sag and Jixi Bulge. However, an obvious difference in the contractional stages is that the direction of compression only originated from the NW direction to generate an associated foreland basin system, as defined by DeCelles & Giles (Reference DeCelles and Giles1996). The wedge top, foredeep, forebulge and back bulge depozone developed progressively from the Langshan fold–thrust belt to the Huanghe Subsag. The thrust faults (f₅ and f₆) developed at the shortening stages were preserved in sections E–F and N–N' (Figs 8c, 11a). After uniform subsidence in Stage 3 and erosion during Late Cretaceous – Paleocene time, the associated foreland basin system was subjected to two extension stages from the SE direction. The master boundary normal fault (F₂) cut off the foredeep depozone along the pediment of the Langshan fold–thrust belt to generate the Langshan Fault (F₂). The Neogene strata dramatically thickened towards the hanging wall of the Langshan Fault (F₂), thereby indicating that the extension intensity in the second extensional phase (Stage 5) was much higher than that in the first extension phase (Stage 4). The Central Fault Zone developed along the rugged forebulge under the Cretaceous layers and segmented into three secondary units along the NE-trending region (Figs 1c, 9a). Conjugate normal faults developed in the last extensional stage (Stage 5) to form grabens and horsts in the cover layers (Fig. 11e).

6.a. Set-up and kinematics

As the direction of compression applied to the Jilantai Sag originating from both the NW and SE direction was different from that of the Hanghou Sag originating only from the NW Langshan mountain to the Linhe Depression, only one model to simulate their simultaneous development is difficult to devise. In order to illustrate the tectonic evolution of the Linhe Depression, we developed four analogue models and used them to show the validity of the evolution model as first proposed in Figures 10 and 11.

The set-up designed for Model 1 to simulate the evolution of the Jilantai Sag and Jixi Bulge comprised a sandbox measuring 85 cm long and 60 cm wide. The sandbox was driven by two motors from the left- and right-hand sides (Fig. 12a). A thin elastic rubber sheet (thickness of 1 mm) was placed at the bottom of Model 1 and fixed on a steel plate to initiate extension from the right side after compression. Two plastic plates measuring 60 × 10 × 1 cm were placed on the rubber sheet close to the right side to simulate the local granite gneiss located on the hanging wall of faults F₇ and F₈ in the basement. Another plastic plate measuring 60 × 13 × 1 cm was placed on the steel plate close to the left side to represent the basement granite gneiss at the front of the Bayanwulashan fold–thrust belt. In the shortening stage, the rubber sheet was first fixed on the bottom by two heavy steel blocks at each ends to avoid its movement along the basement, that is, we did not extend the rubber sheet and kept it in a stationary state at that time. Because the shortening amount in the SE of the seismic section (Fig. 6a) is little larger than that in the NW, two mobile sidewalls were made to push inwards at different rates of 0.2 mm/min and 0.3 mm/min for the left and right motors, respectively. Three layers of loose sand (total thickness 5 cm) were sieved at the bottom of the model to simulate the metamorphic sedimentary layers (Table 1 and Fig. 12b). The deformation of the model surface was covered by an additional layer of white sand (thickness of 1 cm) to represent K₁ l ² after the initial 7.5 cm bulk compression from both sides. Two motors were stopped after 15.0 cm of total shortening, and another 0.7-cm-thick white sand layer was sieved on the depressed area of the deformation model to represent the deposition of K₁ l ¹ after the small amount of erosion of the upper part of the deformation layers on the top of the right side. Thereafter, a 0.8-cm-thick grey sand layer was sieved uniformly on the surface of Model 1 to represent the deposition of the Guyang Formation (K₁ g) (Fig. 12b). Following the subsidence, the elastic rubber sheet was attached to the right mobile sidewall, and Model 1 was pulled only from the right side at rate of 0.2 mm/min to simulate the first extensional stage. After 6.0 cm of extension from the right side, Model 1 was driven from the right side to continue the second extensional stage simulation until 10.5 cm of total displacement was reached.

Table 1. Kinematics of models presented in this study

The set up for Model 2 to simulate the evolution of the Hanghou Sag comprised a sandbox measuring 70 cm long and 60 cm wide based on the natural scale. The sandbox was driven by two motors from the left- and right-hand sides (Fig. 12c). A thin elastic rubber sheet measuring 60 × 45 × 1 cm was placed at the bottom. Its left end was fixed to the steel plate and the right end was attached to the right movable wall to initiate extension after compression. A plastic plate measuring 60 × 10 × 1 cm was placed on the steel plate close to the left side to represent granite gneiss rocks at the front of the Langshan fold–thrust belt. Three small plastic blocks measuring 15 × 7 × 1 cm, 10 × 5 × 1 cm and 7 × 4 × 1 cm were placed on the rubber sheet surface to simulate the rugged forebulge in the Hanghou Sag (Fig. 9a). At the compressional stage, Model 2 was only pushed from the left side to simulate the foreland basin structure during Early Cretaceous time (Table 1 and Fig. 12c). After 6.0 cm of contraction and subsidence post-erosion, Model 2 was extended to a total of 12.0 cm at rate of 0.2 mm/min to simulate the two stages of extension, including five phases during the Cenozoic Era.

The total thicknesses of the loose sand layers involved in the deformation of Models 1 and 2 were 9.0 and 10.3 cm, respectively (Fig. 12b, d). During the deformation, all top surfaces of the models were photographed every 1 cm of displacement. At the final stage, all models covered by a 1.0-cm-thick layer of sand were impregnated with water, and serial vertical sections were sliced and photographed at 1 cm intervals.

Model 3 was designed to simulate the evolution of the Jilantai Sag and Jixi Bulge in order to compare this with Model 1. Three layers of loose sand (total thickness 5 cm) were sieved at the bottom of the model to simulate the basement layers (Table 1). In the shortening stage, the sandbox was compressed by two motors from the left- and right-hand sides (Fig. 13a). The deformation of the model surface was covered by an additional layer of sand (thickness 2 cm) after the 9.0 cm bulk compression from both sides. Another three layers of loose sand (total thickness 3 cm) were sieved on top of the model after the small amount of erosion of the upper part of the deformation layers to represent the deposition of subsidence. Following the subsidence, the elastic rubber sheet was attached to both the right and the left mobile sidewall, and Model 3 was pulled from two sides at rate of 0.2 mm/min to simulate the extensional stage until 10 cm of total displacement was reached.

Model 4 was designed to simulate the evolution of the Hanghou Sag in order to compare with Model 2. The set-up is similar to that for Model 2, except for the absence of the three small plastic blocks (Fig. 13b). At the first compressional stage, four layers of loose sand (total thickness 6 cm) were sieved at the bottom of the model to simulate the basement layers, and Model 4 was only pushed from the left-hand side to simulate the foreland basin structure during Early Cretaceous time (Table 1). The deformation of the model surface was covered by an additional layer of sand (thickness 1 cm) after the 7.0 cm bulk compression. After a total of 14.0 cm of contraction in the second compression stage, Model 4 was extended to a total of 15.0 cm at rate of 0.2 mm/min to simulate the extension deformation during the Cenozoic Era.

Models 3 and 4 were conducted in order to compare with Models 1 and 2; we therefore did not cut sections after simulation.

6.b. Scaling

The two models presented here were scaled to the Linhe Depression geometrically, kinematically and dynamically. Geometric similarity was achieved using a thickness ratio of approximately 1.0 × 10⁻⁵, where 1 cm of loose quartz sand in the models simulated 1.0 km of sediments in nature. Dry quartz sand has an average angle of internal friction of 36°, a cohesive strength of 1.05 kPa and deforms according to Navier–Coulomb failure, making it a suitable material for simulating the brittle deformation of sediments in the upper crust (McClay, Reference McClay, Knipe and Rutter1990). Quartz sand has a well-rounded grain shape and a size of c. 0.2 mm (see sand properties in Table 2).

Table 2. Scaling parameters of the analogue models

Obtaining kinematic similarity was approached by simulating a sequence of events in the models that closely followed the interpreted evolution history of the Linhe Depression. On the basis of our interpretation of the seismic and field data, along with the tectonostratigraphic relations of different sedimentary formations, a five-stage evolution was therefore proposed and mimicked in the models. In the first stage, Models 1 and 3 were compressed in the NW–SE direction to initiate orthogonal contraction in the Jilantai Sag, while Models 2 and 4 were compressed only from the NW direction in the Hanghou Sag. The second shortening stage was followed by two inverted stages resulting from the southeastern extension of Models 1 and 2 after a uniform subsidence.

Dynamic similarity was fulfilled by simulating the physical properties of the sedimentary units of the Linhe Depression with appropriate modelling materials. In this regard, the intrinsic material properties, such as cohesion (τ ₀) and coefficient of internal friction (μ), needed to be approximated in both the models and in nature (Koyi & Peterson, Reference Koyi and Peterson1993; Koyi, Reference Koyi1997). The angle of the internal friction of rocks in the upper crust (< 10 km) was averaged as 40° (Brace & Kohlstedt, Reference Brace and Kohlstedt1980), thereby resulting in a coefficient of internal friction (μ) of 0.84. The angle of internal friction of the uncompacted loose sand used in the models was 36°, resulting in a coefficient of internal friction of 0.73 (Koyi & Vendeville, Reference Koyi and Vendeville2003; Yu & Koyi, Reference Yu and Koyi2016, Reference Yu and Koyi2017), which is considerably close to that of the rocks in the upper crust. Cohesion (τ ₀) was scaled by the equality between the non-dimensional shear strength in the models and in nature:

(1) $${\left( {{{\rho gl} \over {{\tau _0}}}} \right)_m} = {\left( {{{\rho gl} \over {{\tau _0}}}} \right)_n}$$

where ρ is density, l is length, g is the acceleration due to gravity, and subscripts m and n denote the model and nature, respectively. The non-dimensionalized ratio was calculated for the models and for nature by using the shear strength of sedimentary rocks ranging from 1 to 10 MPa (Hoshino et al. Reference Hoshino, Koide, Lnami, Lwarmura and Mitsui1972). For clastic sediments, the shear strength and density were 10 MPa and 2550 kg m⁻³, respectively (Yu & Koyi, Reference Yu and Koyi2016).The cohesion of the loose sand was approximately 100–140 Pa (Yu & Koyi, Reference Yu and Koyi2016) and its density was 1550 kg m⁻³ (Table 2). These characteristics resulted in non-dimensional shear strengths (Equation (1)) of 11–15 for the model and 25 for nature. These two ratios, which are within the same order of magnitude, suggest that our models achieved approximate dynamic similarity to their prototypes.

Following the scaling procedure (Ramberg, Reference Ramberg1981; Weijermas & Schmeling, 1986), the scaling stress (σ*) is given by:

(2) $${\sigma ^*} = {{{\sigma _{\rm{m}}}} \over {{\sigma _{\rm{n}}}}} = {\rho ^{\rm{*}}}{g^{\rm{*}}}{l^{\rm{*}}} = {{{\rho _m}} \over {{\rho _n}}}\,{{{g_m}} \over {{g_n}}}\,{{{l_m}} \over {{l_n}}}$$

where σ* ∼ 0.54 × 10⁻⁶; all scaling parameters used in our models are reported in Table 2.

6.c. Model results

7.a. Negative inversion structure analysis

The mechanism responsible for the development of negative inversion structures in some parts of the Linhe Depression during the Mesozoic–Cenozoic eras remains debatable. This gap is mainly because of the lack of evidence of inversion within the Linhe Depression itself, with inversion only observed in certain areas of the Jilantai and Hanghou sags. The deformation patterns preserved in the Jilantai Sag present a typical negative inversion structure related to the model proposed by Williams et al. (Reference Williams, Powell, Cooper, Cooper and Williams1989) (Fig. 18a, b). The thrust system formed in the first stage of compression in the Jilantai Sag is similar to the eroded thrust system. The thrust faults F₇' and F₈' (Fig. 18c) are similar to faults f₁ and f₃ (Fig. 18a). The partial reactivation of F₈ generated in the two extensional stages is related to fault f₄ in the complex thrust system (Fig. 18b, d). The normal fault (F₇) cutting across the tip of thrust fault (F₇') in the seismic profile of the Jilantai Sag (Fig. 18d) is related to fault f₃ in the complex thrust system (Fig. 18b). Newly formed normal faults (f₁₂ and f₉) in Model 1 located where there were thrust planes (f₄ and f₃) (Fig. 14g); even if they do not reactivate them in these two examples, the normal faults are located at the tip of the thrusts and do not cut them apart from as shown in cross-section E–F (Fig. 14g). However, there is still a link between the localization of the normal faults and the pre-existing thrust faults.

According to Coulomb’s failure criterion, the initial dip angle of a fracture is equal to 45°–(ϕ/2). If the internal friction angle (ϕ) of clastic rocks in the Linhe Depression is close to 30° then, according to Anderson’s theory of faulting (Anderson, Reference Anderson1951), the dip angle of a thrust fault should be close to 30° and that of a normal fault should be approximately 60°. When the dip angle of a thrust fault increases to approximately 60° as founded in a thick-skinned compression (Jagger & McClay, Reference Jagger and McClay2018), extensional faults will reactivate with the reuse of thrust faults, thereby causing a partial or complete inversion structure, such as faults F₇ and F₈ in the seismic sections (Fig. 18d). Otherwise, normal faults cut across the pre-existing thrust system with no reuse of thrust faults, such as faults f₇ and f₁₂ in the simulation result of Model 1 (Fig. 14g). The similarity between our modelling results and the seismic interpretation illustrates that the Linhe Depression experienced compression during the Mesozoic Era, and then extension during the Cenozoic Era, ultimately leading to a negative inversion structure.

The different textures of the basement rocks explain the multiple inverted patterns in the depression. Here, the results of Model 1 showed the formation of a piggyback basin in front of the plastic plate during the compression stages from both sides. The distance (L) between the initial thrust fault and the plastic plate can be calculated as:

(3) $$L = {t \over {{\rm{tan}}(45^\circ - \emptyset /2)}}$$

where t is the thickness of the sand layers and ϕ is the internal friction angle of the sand. In our experiments, the internal friction angle of the sand was c. 36° and the initial thickness of the sand layers in Model 1 was 5 cm. The distance (L) from the initial thrust fault f₁ to plastic plate 1 was therefore c. 10 cm and from fault f₁ to the right-hand side of the model c. 20 cm as the width of plastic plate 1 attached to the right-hand side reached 10 cm (Fig. 14a). When the shortening amount exceeded the space between plastic plates 1 and 2, the new thrust fault (f₄) began to develop in front of plastic plate 2 (Fig. 14c). The distance from f₄ to plastic plate 2 can be calculated from Equation (3). In our models, we used plastic plates to represent the Archean granite gneiss rocks in the basement. We also put loose sand around the plastic plates to represent Proterozoic metamorphic sedimentary rocks in the basement to represent the different textures of the basement. If we do not incorporate the differences in textures of the basement, then the modelling results will generate imbricate thrust faults with a small equal distance between each other (Buiter, Reference Buiter2012); this uniform condition of the basement is dissimilar to the seismic interpretation in the Jilantai Sag. The tectonic framework formed in the contraction stages controlled the areas where the normal faults developed during the extensional stages. If the model was extended from two sides, a graben-and-horst pair will generate in the both left and right segments as shown in Model 3.

The results of Model 2 showed that the Central Fault Zone, including the Song 5, Zhage and Xinglong fault zones, initially developed along the boundary of the pre-existing forebulge (the locations of the small plastic blocks) in the middle part of the Hanghou Sag. Two grabens at each side of the central horst developed following the horst during the extension stages. If we had not inserted plastic blocks in the basement to represent the forebulge, then the modelling results would show a homogeneous deformation in the cover layers as Model 4 indicated, or something like the conjugate fault sets due to the Poisson effect (Zwaan et al. Reference Zwaan, Schreurs and Buiter2019).

The location of the forebulge, showing high-gravity anomalies (Fig. 9), indicated the different textures in the basement that controlled the cover deformation to generate the Central Fault Zone in the Hanghou Sag.

7.b. Comparison between modelling results and nature

Models 1 and 2 produced similar tectonic evolutions for the Jilantai and Hanghou sags. The modelling results at the extension stage were compared with the structural map of lowermost Eocene seismic reflections in the Linhe Depression.

The map view of Model 1 results broadly showed the similarities in the distributions and configurations of the third-order structural units to those developed on lowermost Eocene seismic reflections in the Jilantai Sag (Fig. 19a, b). Faults F₄, F₅, F₇ and F₈, which are the boundary faults of the Jixi, Jizhong and Jidong subsags, respectively, were similar to faults f₆, f₈, f₁₂ and f₁₁ in the trending and dipping regions developed in the experiment conducted at the 10.5 cm extension. The inversion patterns of faults F₅, F₇ and F₈ in the seismic sections (Fig. 20b) were broadly similar to faults f₈, f₁₂ and f₁₁ in the simulation sections (Fig. 20a).

The map view of Model 2 results at the 6.0 cm extension also showed similarities in the distributions and configurations of the structural units to those developed on lowermost Eocene seismic reflections in the Hanghou Sag (Fig. 19c, d). The modelling results reproduced the development process for the Aolun Subsag, Central Fault Zone and Huanghe Subsag in the Hanghou Sag through extension from the SE side. The segmented characteristics of the Central Fault Zone were similar to the simulation results in the top view. The transfer zones of TZ2 and TZ3 in nature were similar to those of TZ5 and TZ6, which developed along the gaps between two basement plastic blocks in the modelling results at the 6.0 cm extension. In Model 4, the experiment conducted without any pre-existing plastic block in the basement, the modelling results for the same scenario would generate distributed deformation (Fig. 17). In Model 2, the separated plastic blocks in the basement were used to mimic a rugged forebulge at the centre of the Hanghou Sag in the foreland compression stages. We can therefore reasonably consider that the pre-existing rugged forebulge was responsible for the formation of the Central Fault Zone and that the gaps among them could easily have caused the transfer zones to offset the segments of the Song 5, Zhage and Xinglong fault zones (Fig. 19c, d).

The distances between the adjacent units in the modelling results did not exactly coincide with the structural map in nature because of the complexity of the inversion modelling process. Nevertheless, their similar configurations suggested that the Linhe Depression experienced the same tectonic evolution as that in the experiments.

The photographs and interpretations of the cross-sections at the final stages taken from Models 1 and 2 were used to compare the internal geometry of deformation with the seismic interpretation (Fig. 20). In the section of Model 1, thrust fault f₁ and its back thrust fault f₃, which developed during the contractional stages, were cut by normal faults f₁₁ and f₁₀ to generate a partial inversion structure (Fig. 20a). This inversion pattern was very similar to the features of inverted faults F₈ and its back thrust fault in the seismic profile (Fig. 20b). Thrust fault f₅, which developed at the shortening stages in Model 1, was cut through by normal faults f₆, f₇ and f₈ in the later extension (Fig. 20a). The reverse evidence of fault f₅, which was clearly preserved in layers 4 and 5, could be compared with that of fault F₅ in the seismic profile (Fig. 20b). Moreover, the boundary normal fault f₆ in Model 1 was similar to fault F₄ as the boundary fault of the Jixi Subsag (Fig. 20a, b). However, identifying the trace of fault F₅ on the footwall of fault F₄ was difficult because of the quality of seismic reflection. Thrust fault f₄ was cut by normal fault f₁₂ at the tip of the upper segment and by fault f₁₄ at the backend. The configurations of f₄ and f₁₂ in the modelling sections were similar to those of F₇ and F₇' in the seismic profiles (Fig. 20a, b). Normal fault f₉ in the modelling sections was comparable to fault F₉ in the seismic profile B–C.

The deformation patterns between the horsts and grabens in the modelling sections (Model 2) showed the same features as those developed in the seismic profiles (Fig. 20c, d). The graben formed by f₄ and f₅ in the modelling sections was similar to the Aolun Subsag, which was formed by F₂ and F₂' in the seismic profiles. The horst formed by f₉ and f₈ in the modelling sections was comparable to the Zhage fault zone, which was formed by F₂₀ and F₂₂ in the seismic profiles. The other grabens and horsts developed between the Aolun Subsag and Zhage fault zone also presented the same patterns as those developed in the modelling results. The Huanghe Subsag was also found to be similar to the graben formed by f₈ and f₁₃. The evidence that f₂ was cut off by f₄ in Model 2 can be used to interpret the origin of thrust faults f₅ and f₆ in seismic profiles, which were cut at the later extension(Fig. 20c, d).

However, there still exist differences between the modelling results and nature, such as the general half-graben shape of the Jilantai Sag and the thickness of the sediments on the horst to the left of the Hanghou Sag in nature. The simulation results of Models 1 and 2 did not generate as many secondary faults as interpreted in the seismic section (Fig. 20). The trending of the major faults and the distance between them in the simulation results are not completely compatible with that observed in nature (Fig. 19).

8.a. Cretaceous differential compression model

The differential compression model representing the kinematic evolution of the Linhe Depression at the shortening stages during Early Cretaceous time (Fig. 21) was constructed on the basis of field observations, seismic interpretation and simulation results. The model is separated into two parts by the Dengkou transfer zone. The southwestern segment of the model was almost simultaneously subjected to compression from the NW and SE directions. The NW contractional stress originated from the Bayanwulashan fold–thrust belt, and the SE contractional stress originated from the Helanshan fold–thrust belt (Yang & Dong, Reference Yang and Dong2018). Piggyback basins (defined by Ori & Friend, Reference Ori and Friend1984) or wedge-top depozones (defined by DeCelles & Giles, Reference DeCelles and Giles1996) developed in front of the fold–thrust belt at each side, while the foredeep depozone developed at the basin centre. Boulder conglomerates and coarse-grained sandstone were deposited at the bottom of the piggyback basin, while siltstone and mudstone were deposited at the centre of the foredeep depozone (Fig. 18). The NE segment of the model was only subjected to the contraction from the NW of the Langshan fold–thrust belt to control the deposition of the Lisangou Formation (k₁ l). Wedge-top, foredeep, forebulge and back bulge depozones developed successively from the NW to the SE (Fig. 18). The forebulge shape was not regular along its trending axis. The thickness of the Lisangou Formation on the forebulge was less than that in the other subsidiary units, and the lithology comprised sandstone interbedded with mudstone according to the borehole data of well S5 and the seismic interpretation. Coarse-grained facies of alluvial fan were deposited in the wedge-top units, while fine-grained facies dominated the front part of the foredeep depozone close to the forebulge.

8.b. Cenozoic negative inversion model

We constructed a negative inversion model to demonstrate how the rifted basin in the Cenozoic period superimposed the Lower Cretaceous foreland basin after erosion, following the deposition of the Guyang Formation (Fig. 22). The extensional stress from the SE direction might be linked to the clockwise rotation of the rigid Ordos block (Fu et al. Reference Fu, Fu, Yu, Yao, Zhang, Ma, Yang and Zhang2018) or the gravitational collapse of a thickened crust after at least 40 km of shortening (Darby & Ritts, Reference Darby and Ritts2007).

In the southwestern segment, the boundary normal fault F₄ cut off the upper part of the Jixi piggyback basin and thrust fault F₅ to generate the Jixi Subsag. Moreover, the new steep normal fault F₅ partially reactivated thrust fault F₅ to form a partial inversion. Normal faults F₇ and F₈ were partially reactivated on the top branch point of the thrust faults to generate the Jizhong and Jidong subsags, respectively. In the northeastern segment, the Langshan Fault (F₂) cut off the foredeep depozone and the northeastern part of the wedge-top depozone to generate the Hanghou Sag. Several conjugate normal faults developed along the forebulge boundary to form the boundary faults of the Central Fault Zone, including the three segments of Song 5, Zhage and Xinglong fault zones. The Huanghe Subsag developed on the back bulge depozone and was controlled by the right boundary fault of the Central Fault Zone.