Abstract
Analogue models are used to study the mechanical relationship between basement strike-slip faults and salt diapirs. Displacement along a strike-slip fault in 1 g models resulted in extension along pre-existing jogs and the formation of oblique extensional faults, where reactive diapirs were triggered in some models. In the centrifuge models, prescribed cuts, simulating pre-existing structures, were reactivated during simple shear deformation of the models, resulting in formation of pull-apart basins, which were intruded by diapirs. The models show that because of the low ratio of salt to overburden thickness in the Zagros (0.15–0.35), it is unlikely that diapirs have formed solely, if at all, as a result of movement along basement strike-slip faults. Two mechanisms are suggested. First, pre-kinematic thin overburden and continuous movement along a releasing bend in the cover units may have triggered some of the diapirs in the Zagros, which were later downbuilt to their current geometry by additional sedimentation. Second, movement along the strike-slip faults (e.g. the Kazerun and Mangarak faults) induced oblique movement along NW–SE Zagros structures (folds or thrusts) resulting in the formation of pull-apart basins where diapirs were eventually intruded. Fault plane solution of shallow earthquakes supports the second scenario, which is also in agreement with previous interpretations that some of the salt diapirs associated with basement faults are younger than Zagros shortening and young southwestwards.
It has been proposed that normal movement along faults in the overburden units is the most important feature in triggering salt diapirs (e.g. Vendeville & Jackson 1992; Koyi et al. 1993). Previous studies (Koyi 1991; Richard 1991; Koyi & Petersen 1993; McClay & Dooley 1996; Withjack & Callaway 2000; Schlische et al. 2002) have indicated that extension (thin- or thick-skinned) causes faulting and weakening of overburden layers and provides the space through which a buoyant layer could rise by differential loading across the fault zone. Normal movement along the faults in frictional overburden can be generated as a result of thick-skinned extension (Koyi 1991; Richard 1991; Koyi & Petersen 1993; Krzywiec 2006), thin-skinned extension (Vendeville & Jackson 1992) or in pull-apart basins along offsets in basement strike-slip faults (Talbot & Alavi 1996). Alignment of some of the salt diapirs in the Zagros fold–thrust belt prompted some workers to attribute the initiation of these diapirs to basement strike-slip faults (Kent 1979; Fürst 1990; Talbot & Alavi 1996). Talbot & Alavi (1996) attributed the formation of some salt diapirs in the Zagros fold–thrust belt to opening along the releasing jogs of some strike-slip faults (e.g. the Mangarak fault zone).
In this study, we use scaled analogue models (both sand-box and centrifuge models) that are intended to systematically investigate the kinematic and mechanical relation between releasing bends of strike-slip faults and salt diapirs. These models are designed to study the spatial and temporal relationship between salt diapirs and basement strike-slip faults and are not scaled to any specific area. Nevertheless, the results of these models are compared with field data from the Zagros fold–thrust belt to explain the spatial relationship between the Hormuz salt diapirs and the Pan-African basement strike-slip faults.
Geology of the Zagros fold–thrust belt
The Zagros Mountains extend within the Alpine–Himalayan orogenic belt for 2000 km between the central Iran micro-continents (Takin 1972) and the northwestern Iranian plate in the NE and the Arabian plate in the SW. Subduction of Neo-Tethyan ocean beneath Iran sutured Iran to Arabia (e.g. Takin 1972; Alavi 1980, 1994; Berberian & King 1981) and the subsequent continental convergence resulted in the formation of the Zagros fold–thrust belt.
Geophysical and geological data indicate that the Zagros fold–thrust belt is currently active and is being shortened at a rate of 20–30 mm a−1 (Vita Finzi 1987, 2001; DeMets et al. 1994). However, recent global positioning system (GPS) measurements (Nilforoushan et al. 2003; Hessami et al. 2006) show a lower rate (5–10 mm a−1) of shortening across the belt.
The Zagros sedimentary cover contains several evaporitic units with temporal and spatial variations (O'Brien 1957; Talbot et al. 1989). Among them, the Neoproterozoic Hormuz salt located at the base of the sedimentary column and the Gachsaran Formation (Miocene) higher in the Zagros stratigraphic column are the thickest and most widespread evaporitic units (O'Brien 1957; Kent 1979). These two units have acted as viscous décollements during the evolution of the Zagros fold–thrust belt. The estimated 1.3 km thick Hormuz salt is overlain by 6–7 km of lower Palaeozoic to upper Mesozoic sandy shale, carbonate and sandstone (O'Brien 1957; James & Wynd 1965). However, it has been suggested that the Phanerozoic sediments in the belt can be as thick as 14 km in some places (Motiei 1995). More than 200 salt structures of the Hormuz salt have been recognized in the northeastern Arabian platform (O'Brien 1957; Kent 1979; Edgell 1996; Talbot & Alavi 1996).
Kent (1979) described some dextral transcurrent fault zones, decorated by salt extrusion, trending obliquely across the Zagros; for example, the Kazerun and Mangarak zones (Fig. 1). These zones display a complex anomalous relationship between strike-slip faults that trend north–south to NNW–SSE and Zagros folds and thrusts that trend NW–SE. Some of the Zagros folds are displaced along the surface expression of these strike-slip faults (Hessami et al. 2001a). Kent (1979), Talbot & Alavi (1996) and Hessami et al. (2001a) considered the Kazerun and Mangarak fault zones as the surface expression of reactivated basement strike-slip faults trending north–south or NNE–SSW. Talbot & Alavi (1996) reported that some salt diapirs in SW Iran intruded through rhombic pull-apart basins formed along jogs along these basement strike-slip faults.
Simplified map of the Zagros fold–thrust belt and central Iran. Outlined area shows the location of Kazarun fault (K.f) and Mangarak fault (M.f). Inset shows a line drawing of the current kinematics of the Mangarak fault zone (circled on the map) stylized to an almost continuous pattern of faults (from Talbot & Alavi 1996). 1, Ophiolite rocks; 2, Urumieh–Dokhtar Magmatic Assamblage; 3, thrust fault; 4, strike-slip fault; 5, known distribution of Hormuz salt. M.Z.T., Main Zagros Thrust Zone; ⋅, location of known extruded and blind salt structures.
Both sand-box models and centrifuge models were designed to investigate the relation between strike-slip faults with releasing bends and salt diapirs in the Zagros fold–thrust belt. The sand models were used to study different parameters (thickness, displacement along the strike-slip fault, location of the strike-slip fault) systematically, whereas the centrifuge models were designed to study the effect of a pre-existing strike-slip fault in the overburden.
Modelling materials, experimental procedure and scaling
In this study, results of seven 1 g models and a centrifuge model are presented. In each of the 1 g models, two different rheologies were simulated: the frictional behaviour of overburden sediments and the ductile behaviour of rock salt. Because loose sand exhibits a nearly perfect Mohr–Coulomb behaviour (Krantz 1991; Weijermars et al. 1993; Cobbold & Castro 1999; Cotton & Koyi 2000; Schellart 2000), dry quartz sand sieved to an average grain diameter of 350 μm is used as a suitable analogue for the sedimentary overburden in the upper crust. To simulate the ductile behaviour of rock salt, transparent silicon (SGM-36) was used. SGM-36 is a Newtonian viscous fluid (ρ=0.987gcm−3) with no yield strength (Weijermars 1986) and an effective viscosity of 5 × 104 Pa s at room temperature (c. 20 °C).
The releasing bend fault systems were introduced in the models using a thin plastic sheet (0.2 mm thick) that was cut in such a way that it produced a releasing bend at an angle of 45° to the strike of the PDZ (Fig. 2). Two versions of side-steps or releasing bends were introduced in the 1 g models. In the first set, a releasing bend initiated at the base of the model, simulating a releasing bend in the basement (i.e. beneath the ductile layer) (Fig. 2b). These models, with the plastic sheet at the bottom of the model, were placed on a metal plate simulating a rigid basement and movement along the plastic sheet simulated thick-skinned deformation and deformed both the overburden and the buoyant layer. In the second set of models, the plastic sheet had a similar configuration to that in the first set of models (i.e. a cut parallel to the strike of PDZ and a side-step at an angle of 45° to the strike of PDZ). However, the other side of the sheet was straight, and intended to see the effect of a pure strike-slip movement (without a releasing bend). The plastic sheet in this set of models was placed in the basal part of the overburden layer, between the overburden and buoyant layers (Fig. 2c). This configuration simulated thin-skinned deformation.
(a) Schematic illustration of the 1 g model series and their respective variables. Thickness ratio (Tr) = Ts/To (thickness of salt/thickness of overburden). Thickness/displacement ratio (Dr) = Ts/D (thickness of salt/amount of strike-slip movement). (b, c) block diagrams of initial set-up of the models with (b) a strike-slip fault separating the overburden from the salt layer and (c) a strike-slip fault in the basement. The strike-slip had a side-step at an angle of 45° to the strike of PDZ. (d) Plan view of model configuration showing the initial arrangement of the ductile (dark grey) and the frictional (sand) substrates.
All 1 g models were built in a pure-shear deformation rig (Fig. 2). The models had a fixed initial length of 25.5 cm and width of 30 cm. Various thickness estimations for the Hormuz salt (between 0.9 and 4 km) and sedimentary cover (between 5 and 20 km) have been suggested in different places along the Zagros fold–thrust belt (e.g. O'Brien 1957; James & Wynd 1965; Stöcklin 1968; National Iranian Oil Company 1975; Berberian 1976; Kent 1979; Beydoun 1991; Motiei 1993; Talbot & Alavi 1996; Sepehr 2001). Such variations in thickness of the sedimentary cover and Hormuz salt were not simulated in our models. Instead, to show the influence of strike-slip faults on diapirs in the models, different thickness/displacement ratio (Dr) and source/overburden thickness ratio (Tr) were considered in our models (Table 1). Dr is the ratio between the thickness of the source layer (salt layer) and the amount of strike-slip displacement along the fault. This ratio defines the degree of decoupling between the basement fault and cover units; small Dr results in decoupling.
Scaling parameters between models and nature
A passive 1.5 cm × 1.5 cm sand grid was sieved onto the upper surface of the model so that the surface strain could be recorded during deformation. One motor-driven worm screw, moving at a constant rate of 4.7 × 10−4 cm s−1, produced dextral strike-slip movement in the model. After each 1 cm increment of deformation, the surface of the model was photographed. After completion, each model was covered with loose sand to stop further deformation and impregnated with water. The saturated sand pack had enough cohesion to allow vertical sections to be cut in any orientation. Then the side panels of the rig were removed and sections were cut sequentially and documented photographically.
Scaling of 1 g models
The 1 g models described here are scaled with a geometric similarity (length ratio) of c. 10−5, which means that 1 cm in the models simulates 1 km in nature. In model series 7, the thickness of the ductile substrata and overburden were changed relative to the amount of strike-slip movement to study the effect of change in the thickness/displacement ratio (Dr=T/d, where T is thickness of ductile material and d is amount of strike-slip movement) on diapirism (Table 1).
For dynamic similarity between a model and its natural prototype, a set of dimensionless ratios that relate to physical properties of the model materials and rocks should be similar for the two cases. To achieve dynamic similarity in model and nature, the intrinsic material properties, such as the coefficient of cohesion (τ0) and coefficient of internal friction (μ), need to be equal (Koyi & Petersen 1993; Weijermars et al. 1993). The internal friction angle of the upper crust rocks (<10 km) is averaged to 40° (Brace & Kohlstedt 1980), which gives a coefficient of internal friction of 0.85. The internal friction angle of uncompacted loose sand used in the models is 36°, giving a coefficient of internal friction of 0.73. Cohesion, however, is scaled by equality between the non-dimensional shear strength both in the model and in nature: $$mathtex$$\[({\rho}l\mathbf{g}{/}{\tau}_{0})_{m}=({\rho}l\mathbf{g}{/}{\tau}_{0})_{n}\]$$mathtex$$(1) where ρ is density, l is length, g is acceleration due to gravity, and the subscripts m and n denote the model and nature, respectively. This non-dimensionalized ratio was calculated for the 1 g models and nature using shear strength of sedimentary rocks ranging between 1 and 10 MPa (Hoshino et al. 1972). For clastic sediments, the shear strength and density have been taken to be 5 MPa and 2550 kg m−3, respectively. Our loose sand acquired cohesion during scraping. Its cohesion is c. 140 Pa and its density is 1550 kg m−3. These figures give the non-dimensional shear strength in equation (1) a value of 1.6 × 10−1 and 7.5 × 10−1 for model and nature, respectively. These two ratios, which are within the same order of magnitude, suggest that our 1 g models fulfil the criterion for dynamic similarity.
For the ductile layer, dynamic similarity was achieved by simulating the ductile behaviour of rock salt with the Newtonian viscous material SGM36, giving a viscosity scaling ratio of 2.9 × 10−14 to 10−15 (Table 2). The models were extended at a constant rate of 3.2 × 10−4 cm s−1, which corresponds to c. 10 mm a−1 (3.17 × 10−8 cm s−1) in nature.
The thickness ratio and thickness/displacement ratio in the models
These models are designed to study the spatial and temporal relationship between salt diapirs and basement strike-slip faults and are not scaled to any specific area. Nevertheless, the results of these models are compared with field data from the Zagros fold–thrust belt to explain the spatial relationship between some of the Hormuz salt diapirs and the Pan-African basement strike-slip faults.
Kinematics and results of 1 g models
On the basis of the location of the strike-slip fault, the 1 g models can be divided into two groups. In the first group, strike-slip movement was generated in the basement (Fig. 2b); in the second group, strike-slip movement deformed only the overburden units (Fig. 2c). Below, we describe the results of these two types of 1 g models.
Side-step dextral basement strike-slip fault
In one of the models (model 3), there was no ductile layer between the loose sand overburden and the basement. The initial deformation on the surface of this model appeared as two oblique-slip extensional faults above and parallel to the basement releasing bend. As a result of normal-slip movements along these faults a graben (pull-apart basin) formed exactly above the basement side-step (Fig. 3).
Line drawings of (a) plan view and (b) a vertical section (see (a) for location of AB) of model 3 (Dr = 0) after 3 cm displacement (see inset for initial configuration); this model contains a side-step basement strike-slip fault. The rhomboidal-shaped pull-apart basin is visible on the surface of the model. Grey shading shows deformed zones. The normal faults were generated by boundary effects of the model.
In models 1, 2 and 5, which were also deformed above a basement strike-slip fault with a releasing bend similar to model 3, a ductile layer separated the ‘basement’ from cover units (Fig. 2b). In these models, initial deformation was characterized by development of a set of oblique-slip faults parallel or subparallel to the basement releasing bend (Fig. 4). With progressive deformation, the number of faults and the amount of normal slip along the faults increased (Fig. 4a–c). These faults were initiated on the flank of a subsiding area that was located approximately above the basement releasing bend. However, these faults in the overburden covered a wider area than the subsiding area above the basement releasing bend (Fig. 4d).
Plan view (a–c) and line drawings of (d) plan view and (e) a vertical section of model 1 (Tr = 1, Dr = 0.25), that contains a side-step basement strike-slip fault (d, e). (a) After 2 cm displacement. (b) After 4 cm displacement. (c) After 6 cm displacement. (d) After 15 cm displacement (see inset for initial configuration; d is side-step length). Grey shading shows deformed zones. (e) Vertical section (see (d) for location of section AB).
In model 1 (Table 2), the width of the basement releasing bend (d in Fig. 4d) was 10 cm and the thickness ratio between the viscous layer and its overburden was 0.5. The initial deformation on the surface of the model was localized along a set of dextral normal-slip faults striking subparallel to the basement releasing bend (Fig. 4b). With increased displacement along the basement fault, the number of parallel normal faults increased and a symmetrical graben formed in the overburden (Fig. 4b–d). Reactive diapirs formed along the grabens and half-grabens. However, only the diapir that rose along the central graben was about to extrude (Fig. 4e).
In model 2 (Table 2), the width of the basement releasing bend was decreased to 5 cm and the thickness ratio between the viscous layer and its overburden was 0.5. During deformation of model 2 (Fig. 5a–c), oblique slip faults parallel to the basement releasing bend formed initially (grey area in Fig. 5c). With increased displacement along the basement fault, similar to model 1, a set of normal faults formed in the overburden and reactive diapirs formed along a restricted graben structure localized along the oblique-slip fault zone (Fig. 5c and d).
Plan view (a, b) and drawings of plan view and a vertical section of model 2 (Tr = 1, Dr = 0.25), that contains a side-step basement strike-slip fault (c, d). (a) After 2 cm displacement. (b) After 4 cm displacement. (c) Line drawing of model surface after 6 cm displacement (see inset for initial configuration). Grey shading shows deformed zones. (d) Vertical section (see (c) for location of section AB). Normal faults were generated by boundary effects of the model.
In model 5, which had a similar configuration to model 2, the thickness ratio between the viscous layer and its overburden was unity (Table 2). Initial deformation was characterized by development of a set of oblique-slip faults parallel or subparallel to the basement releasing bend (Fig. 6a–c). In contrast to model 2, these faults were limited to the width of basement releasing bend and did not continue beyond it. With progressive deformation, above the basement side-step, the thickness of the viscous layer was decreased and a depression formed at the surface. Two symmetrical grabens were generated in the vicinity of this depression. In both models 2 and 5, where the final amount of strike-slip movement was 6 cm, small passive diapirs formed along the grabens bounded by oblique-slip faults (Figs. 5d and 6d⇓).
Plan view (a, b) and line drawings of plan view and a vertical section of model 5 (Tr = 1, Dr = 0.22), that contains a side-step basement strike-slip fault (c, d). (a) After 2 cm displacement. (b) After 4 cm displacement. (c) After 6 cm displacement (see inset for initial configuration). Grey shading shows deformed zones. (d) Vertical section (see (c) for location of section AB).
Overburden strike-slip fault
The plastic sheet that was used to initiate an overburden strike-slip fault consisted of a straight edge simulating a straight strike-slip fault and an edge that made an angle of 45° to PDZ. In model 6 with this configuration, movement of the plastic sheet that was placed beneath the overburden layers and had a releasing bend initiated two strike-slip faults in the overburden layer. The initial deformation on the surface of the model was localized along a sinistral strike-slip fault above the straight edge whereas oblique-slip extensional faults formed above the releasing bend of the plastic sheet (Fig. 7a). After 1 cm of displacement, oblique faults, which were parallel to the releasing bend and did not extend beyond it, formed a symmetrical graben (pull-apart basin) exactly above the initial location of the in-built releasing bend (Fig. 7b and d). With further displacement along the strike-slip fault, these overburden oblique-slip faults linked across the offset and formed a side-wall fault system. These bounding faults were steep, with dips of c. 70°. With increasing strike-slip movement, a cross-basin fault developed and cut the pull-apart basin (Fig. 7c). A large reactive diapir formed along the pull-apart basin (Fig. 7d). Along the straight segment of the sinistral strike-slip fault a small 2D perturbation in the ductile layer formed that never became diapiric.
Plan view (a, b) and line drawings of plan view and a vertical section of model 6 (Tr = 1, Dr = 0.25), which contains a side-step dextral fault and a straight sinistral strike-slip fault only in overburden layer (c, d). (a) After 2 cm displacement. (b) After 4 cm displacement. (c) After 6 cm displacement (see inset for initial configuration). Grey shading shows deformed zones. (d) Vertical section (see (c) for location of section AB).
Centrifuge models
The set-up of the experiments
It was not practically possible to produce a pre-existing thrusts in the overburden package of loose sand in the sand-box models. Therefore, centrifuge models, with Plastilina layers in which cuts could be prescribed, were used to study the effect of pre-existing discontinuities on the spatial relationship between salt diapirs and basement faults. The centrifuge model consisting of a layered sequence was deformed in a shear-box during centrifuging at 700 g. The simple shear box consisted of two cuboidal half shells providing enough space for a model with dimensions 100 mm × 80 mm × 100 mm. Simple shear deformation was achieved via two brass screws that initiated relative motion between the two halves. Four deformation steps were carried out: 12 mm (γ=−0.33), 8 mm (γ=−0.38), and two 10 mm steps (shear strain γ=−1.07). After each step of shearing, the models were centrifuged.
The centrifuge models consisted of a 25 mm thick overburden of semi-brittle Plastilina (μ=5×107Pas at a strain rate of 10−3 s−1, ρ=1.705gcm−3; corresponding to 5 km thick upper crust rocks with μ ≈ 1024 Pa s and ρ ≈ 2.55 g cm−3) and a 6 mm thick buoyant PDMS layer (μ=4×104Pas, ρ=0.964gcm−3; corresponding to a 1200 m thick layer of rock salt with μ=4×1017 to 1018 Pa s and ρ ≈ 2.2 g cm−3 (Table 3). Both overburden and buoyant layers were stratified. The Plastilina overburden consisted of a sequence of white and grey strata giving a final thickness of 25 mm. A passive grid was printed on the surface of the model. The buoyant layer, as in experiment 1, consisted of one yellow layer atop a blue layer (each 3 mm thick). No perturbation was initiated in the buoyant layer. However, two 20 mm long cuts penetrating the entire overburden sequence were prescribed. These prescribed cuts, which simulated pre-existing faults in the cover, were made to see whether the buoyant material could use them as pathways to intrude the semi-brittle overburden units during shearing. The cuts were arranged at an angle of 45° to the long axes of the model and perpendicular to the maximum principal stress σ1 (Fig. 8) The sides and bottom of the model were confined to prevent leakage during centrifuging. Ramberg Number (RN) is defined for the centrifuge model. The relatively small difference in RN for overburden between the model (7.2 × 10−12) and nature (1.5 × 10−12) shows that there is a very good approximation for scaling the model to its natural prototype (Table 3).
Scaling parameters between models and nature for the centrifuge model
Photographs of the centrifuge models. (a) Top view of the initial stage showing the orientation of the pre-existing cuts relative to the shear direction. (b) Top view of the model after 20 mm shearing in total (γ = −0.38) and 8 min of centrifuging at c. 700 g. (Note the closure of the pre-existing cuts and strike-slip movement along them, which has resulted in formation of pull-apart basins in the cover.) (c) Top view after 25 mm shearing in total (γ = −0.47) and 20 min of additional centrifuging. Diapirs have surfaced through the pull-apart basins and form overhangs. (d) Section of the model at stage (b), showing the geometry of the pull-apart basin with depth. (e) Section of the model at stage (c) showing one of the diapirs intruding a pull-apart basin.
Kinematics and results of the centrifuge model
During shearing (total angular shear γ=−0.33), the two pre-existing cuts in the semi-brittle overburden, which were perpendicular to σ1, were closed and thrusts formed along the surface expression of the cuts. Centrifuging for 8 min at 680–720 g plus 8 mm further shearing (total angular shear γ=−0.38) led to right-lateral slip along the pre-existing cuts, which in turn triggered the formation of three pull-apart basins (up to 20 mm long, 7 mm wide and 15 mm deep), with their long axes perpendicular to the initial cuts (Fig. 8b and c). After an additional 20 min centrifuging at 720 g and 5 mm shearing, the buoyant layer intruded through the pull-apart basins and formed overhangs on the surface of the model (Fig. 8c). For its intrusion, the diapir has not taken a straight path upwards. Instead, it has followed the irregular pathway created by the pull-apart basin (Fig. 8e).
Discussion
Basement strike-slip fault
The influence of basement strike-slip faults on diapirism depends on the types of faults they create in overburden units above the salt layer. The magnitude and location of overburden faults is related to how the basement transmits movement to the overburden units; that is, coupling between the basement and its cover (Koyi et al. 1993). Surface expression of all sand-box models shows that initial deformation in overburden layers was represented by the formation of oblique-slip extensional faults oriented at 30–40° to the strike of PDZ and parallel to the prescribed releasing bend in the basement fault. Normal slip movements along these faults, although secondary, affected the entire overburden layer and triggered reactive diapirs. However, none of these diapirs extruded (Figs. 4e, 5d and 6d⇑⇑). Additional extension might have allowed these reactive diapirs to extrude.
Results of models 1, 2, and 5 show that when Dr was higher than 0.14 (i.e. thick viscous layer), the movement along the straight segment and releasing bend of the strike-slip faults formed oblique-slip extensional fault zones in the overburden oriented at 30–40° to the strike of PDZ. These fault zones were extended beyond the original width of the prescribed releasing bend. Only in model 4, where the thickness ratio between the viscous layer and overburden units is 0.5 and an initial releasing bend existed, with increase in the displacement ratio (Dr=0.125) no pull-apart basin or graben formed. Instead, a strike-slip fault formed in the overburden above the straight part of the basement fault. This suggests that when the thickness ratio is less than unity, the movement along the side-step in the basement does not generate a pull-apart basin in the cover.
In model 5, where the thickness ratio is unity and an initial releasing bend existed in the basement, the oblique-slip extensional fault zone did not extend beyond the width of the basement side-step fault and a depression formed above the releasing bend. Thinning of ductile material as a result of extension above this releasing bend did not allow formation of diapirs in this subsided area.
The thickness ratio of the viscous layer to its overburden (Tr) and the ratio of the thickness of the viscous layer to the total amount of displacement along the basement fault (Dr) (Table 2) play a significant role in weakening and faulting of overburden units during basement strike-slip faulting, and in turn trigger diapirism. Larger Tr and Dr promote decoupling between basement and cover. For the same Dr, a lower Tr results in a narrower deformation zone in the overburden layer, whereas higher Tr decouples deformation between the basement and overburden layers. Flow of ductile material, which is induced by movement along the basement strike-slip fault, deforms the overburden layer exactly above the area of flow. Consequently, when the viscous layer is thick, movement along the basement fault induces flow within a wider zone in the ductile layer, which leads to a wider deformation zone in the overburden. For the same Tr, a lower Dr, which couples basement deformation with that in the cover, results in producing a strike-slip fault in the overburden layer exactly above the basement fault. A lower Dr refers to larger displacement along the fault or less thickness of viscous layer. Then the result of reducing thickness of the ductile layer to half is roughly equivalent to doubling the displacement along the basement strike-slip fault. In general, lower Dr could represent the later stages of deformation when the buoyant layer thins and displacement increases.
Overburden strike-slip fault
During extension of model 6, where the in-built releasing bend along the dextral strike-slip fault was at the contact between the viscous and overburden layers, the brittle overburden was cut by oblique-slip faults at 45° to the PDZ and was limited to the zone of the releasing bend. Model results indicate that the finite structure of the resulting pull-apart basin is controlled by the geometry of the releasing bend built into the model. An elongated rhomboidal graben formed above the 45° releasing bend of the dextral strike-slip fault.
The resulting normal movement along the side-wall faults of these pull-apart basins weakened the overburden layer and allowed the buoyant material to intrude through this thinned and hence faulted zone, forming a reactive diapir. The diapir never extruded as the model was stopped. These model results suggest that strike-slip movement along the releasing bend of a fault can trigger diapirs only when the fault thins overburden units to such a degree that differential loading can drive reactive diapirs upwards, as has been suggested in extension regimes by others (Koyi 1991; Vendeville & Jackson 1992).
Diapir–strike-slip fault link in the Zagros fold–thrust belt; comparison between models and nature
Some salt diapirs of the SE Zagros are aligned along pre-existing basement faults (Kent 1979; Fürst 1990; Talbot & Alavi 1996; Hessami et al. 2001b). Talbot & Alavi (1996) proposed a model explaining the generic relationship between some salt diapirs and basement strike-slip faults in the Zagros. They reported that some salt diapirs in the Zagros Mountains in SW Iran intruded pull-apart basins that formed by movement along the basement strike-slip faults (e.g. Kazerun and Mangarak faults).
In general, movement along basement strike-slip faults initiates fault zones in cover units. The intensity, width and characteristics of the fault zone in the cover units depend on the thickness and lithology of the cover units, whether or not these units are decoupled from the basement by a viscous layer, the presence of releasing bends along the basement fault, and the amount of displacement along the basement fault. However, strike-slip basement faults that do not have a releasing bend segment probably do not form significant extensional faulting, if any, in the cover units. Also, because diapirs are triggered and promoted by faulting and weakening of overburden units during extension, which also provides the space for diapirs (Koyi 1991; Vendeville & Jackson 1992; Koyi & Petersen 1993), as has been demonstrated by the 1 g models, it is unlikely that pure strike-slip basement faults alone could trigger diapirs (e.g. in the Zagros). However, based on the results of the models presented here we suggest the following two plausible, but not conclusive alternatives to explain the spatial link between the basement strike-slip faults (Kazerun and Mangarak faults) and some of the salt diapirs in the Zagros.
The first alternative is based on the results of the 1 g models. Here, the diapirs that are today aligned along the Kazerun and Mangarak faults might have formed when the overburden units were thin (Tr=1 or higher), and were faulted, weakened and thinned further during movement along the basement strike-slip faults to allow the formation of reactive diapirs. Movement along the basement strike-slip faults might have faulted the overburden units and caused differential movement that eventually drove the Hormuz salt to rise as reactive diapirs along the releasing bends of the overburden faults similar to our 1 g models. Sedimentation of the rest of the overburden units, which in some places reach c. 12 km thickness, could have downbuilt the diapirs to their current geometry. Our models show that basement strike-slip faults may initiate salt diapirs only if the former create a releasing bend (pull-apart basin) in the cover units. In addition, continuous movement along the basement strike-slip faults during the Zagros orogeny would result in the formation of very large pull-apart basins, which are missing in the Zagros fold–thrust belt.
There are several arguments that contradict triggering of diapirs by releasing bends in basement strike-slip faults. In models with a releasing bend along the basement faults (models 1, 2 and 5), movement along the basement strike-slip faults initiates faulting in and thins the overburden layers. In addition, it thins the ‘salt’ layer itself as well. Thinning of the salt layer undermines the formation of diapirs: there is less salt supply to feed any potential diapirs. Furthermore, a releasing bend in the basement will result in local basement depressions that are likely to ‘suck’ the mobile salt downwards rather than promoting it to rise diapirically. The models described here also show that a basement strike-slip fault (without a releasing bend) decoupled from its cover by a thick viscous layer does not initiate the space in the overburden that can be used by the diapir to intrude. Therefore, based on our model results, it is argued to be unlikely that movement along basement strike-slip faults alone could have triggered salt diapirs in the Zagros, where the thickness ratio of the Hormuz salt to its overburden is very small (Tr ranges between 0.1 and 0.3).
In the second alternative, the spatial distribution of salt diapirs and basement faults in the Zagros is explained on the basis of the results of the centrifuge models. Here, the spatial relationship between diapirs and basement strike-slip faults (e.g. Kazerun and Mangarak faults) is attributed to the interaction between the basement strike-slip faults and the cover structures (e.g. Zagros trend thrust-related folds). During movement along the basement faults, the pre-existing NW–SE-trending Zagros structures act as riddles, along which strike-slip movement occurs resulting in the formation of pull-apart basins, where diapirs may intrude (Fig. 9). This is supported by field observations that salt diapirs, which are located along the Kazerun and Mangarak strike-slip faults, are associated with anticlines in the Zagros fold–thrust belt (Fig. 10). The diapirs are associated with Zagros salt-cored anticlines, which can be fault-bend folds, detachment folds or fault propagation folds.
Line drawing of the surface of the model, initial (top), after (γ = −0.2) shearing (middle) and (γ = −0.38) shearing (bottom). (Note that the general strike-slip motion (large arrows) has induced oblique slip along the pre-existing cuts (thrusts), which in turn has resulted in formation of pull-apart basins.
(a) Satellite image showing part of the Mangarak fault (see Fig. 1 for location). After Callot et al. (2005). (Note the presence of salt diapirs (Kuh-e-Gach and Jahani) at the intersection between the segments of the Mangarak strike-slip fault and the Zagros-trending structures. (b) Line drawing of the surface of the centrifuge model showing diapirs intruding pull-apart basins at the intersection of pre-existing structures and the main strike-slip fault (north–south).
This second scenario must suit the timing of formation of the diapirs. Timing of diapir formation in the Zagros is not entirely clear. There are only two field observations on the timing of salt flow and extrusion (Kent 1979). Kent (1979) suggested that thinning of Jurassic strata across a salt pillow east of Bushire indicates an early stage of salt flow, and the presence of insoluble residue of Hormuz salt in Cretaceous sediments around the salt diapirs indicates salt extrusion. Based on these findings, Jackson & Vendeville (1994) suggested that the most vigorous period of diapirism in the Zagros was between 145 and 100 Ma, and related their triggering to a regional extensional phase associated with the rift or drift phase of the Neo-Tethys ocean. This extensional phase would then have faulted and weakened the 5 km thick overburden barrier (Jackson & Vendeville 1994). However, as illustrated below, this interpretation may not be valid for all the diapirs in the Zagros fold–thrust belt.
The diapirs located along the southern segments of the Kazerun and Mangarak strike-slip faults are active structures and possess a dynamic bulge (a salt fountain) whose crest rises 400–800 m above the surrounding plains, whereas those located along the northern segments of the faults are inactive and do not possess a dynamic bulge. The fact that these structures are active indicates that their salt supply exceeds salt dissolution and erosion (Koyi 1998). The diapirs in SW Iran could thus be of different ages (Koyi 1988) and some of them (e.g. those along the southernmost segment of Kazerun fault; Namak Dashti and Darang diapirs) could be younger, post-dating the Zagros shortening, and could have formed during the Miocene or even later (see below).
Talbot & Alavi (1996) sketched and classified the diapirs along the Kazerun fault (Fig. 1). It can be deduced from their data that the diapirs young southward in the same direction as the propagation of the deformation front. The salt structures (Namak Dashti and Darang diapirs) associated with the southernmost segment of the Kazerun fault (50 km NW of Kangan) are young structures. Namak Dashti diapir is a young dome with a dynamic bulge, whereas, according to Talbot & Alavi (1996), Darang diapir is a pre-eruptive structure.
We interpret the data of Talbot & Alavi (1996) as follows. These diapirs have formed as the result of interaction between the Zagros folds or thrusts with the Kazerun and Mangarak basement strike-slip faults. As the NW–SE-trending thrusts formed, movement along the north–south-trending Kazerun fault created pull-apart basins at the junction between the thrusts (fault propagating folds, fault bend folds or detachment folds) and the basement fault and triggered diapirs through the pull-apart basins. As the Zagros deformation front propagated southwestward (Hessami et al. 2001b), new NW–SE-trending thrusts formed and interacted with the basement faults to trigger younger diapirs southwestwards. As younger diapirs formed along the basement strike-slip faults during the southwestward propagation of the deformation front, older diapirs in the NE became inactive as their salt source decreased in thickness or their stems were pinched by shortening. This scenario is simulated in our centrifuge models, where shear of pre-existing structures (Zagros thrusts) above a basement fault results in formation of pull-apart basins and thins the overburden unit where diapirs intrude.
The scenario described above is in accordance with both dextral and sinistral strike-slip basement faults, which have been reported from this part of the Zagros (Hessami et al. 2001a; Fig. 1). As such, this mechanism can explain the spatial association of salt diapirs in the Zagros fold–thrust belt along the Kazerun and Mangarak basement faults. This scenario requires oblique slip along the Zagros thrusts with which the salt diapirs are associated. Fault plane solution of ‘shallow’ earthquakes in the vicinity of the Kazerun and Mangarak strike-slip faults (Talebian & Jackson 2004; Tatar et al. 2004) shows a major component of reversal dip-slip and a minor component of strike-slip. We argue that some of these earthquakes (at least the shallow ones) occur along thrust ramps within the cover sequences (Koyi et al. 2000; McQuarrie 2002). The strike-slip component of displacement along these cover thrusts is induced by the strike-slip movement along the basement faults (Talebian & Jackson 2004; Tatar et al. 2004). These strike-slip components are small in present-day fault plane solutions (i.e. single earthquakes) (Talebian & Jackson 2004; Tatar et al. 2004). However, accumulated throughout the shortening history of the Zagros, a significant amount of strike-slip movement must have occurred along the Zagros-trend thrust surfaces (both blind thrusts beneath the Zagros salt-cored anticlines and thrusts that reach the surface) and resulted in the formation of pull-apart basins (some kilometres in diameter) along their intersection with the basement strike-slip faults. Any salt that has accumulated in the cores of the anticlines must have then used these faulted weak zones (i.e. pull-apart basins) to intrude to the surface.
Conclusion
Results of 1 g and centrifuge models and geological arguments are used to suggest two mechanisms for the spatial association of salt diapirs with basement strike-slip faults. The mechanism supported by 1 g models is that salt diapirs, suggested to be triggered by and intruded through releasing bends of basement strike faults in the Zagros fold–thrust belt, must have initiated when the overburden was thin. In addition, movement along the basement strike-slip faults must have continued during the deposition of overburden units after diapir initiation in the Zagros, which in some places reach c. 12 km thickness. In essence, the diapirs must have downbuilt after they were triggered by faulting of a much thinner overburden than is present in some parts of the Zagros at present. In this scenario, the diapirs associated with the basement faults are assumed to be older than the Zagros shortening. The age of Zagros diapirs does not support this alternative.
On the other hand, centrifuge models and field data suggest that some of the salt diapirs associated with the basement faults are younger than and/or formed simultaneously with the formation and propagation of the Zagros folds or thrusts. These diapirs intrude through pull-apart basins and damage zones that formed during oblique slip along the Zagros folds or thrusts induced by movement along the basement strike-slip faults. This study is in favour of the second alternative.
Acknowledgements
We would like to thank the reviewers, M. Bonini and P. Krzywiec, and the Editor, I. Alsop, for their reviews, contructive comments and suggestions. This work was initiated during A.G.’s visit to the Hans Ramberg Tectonic Laboratory, which was funded by the Swedish Institute. H.A.K. is funded by the Swedish Research Council.
- © 2008 The Geological Society of London