Geophysical Research Letters

Out-of-sequence thrusting in experimental Coulomb wedges: Implications for the structural development of mega-splay faults and forearc basins


  • Saad S. B. Haq

    Corresponding author
    1. Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana, USA
    • Corresponding author: S. S. B. Haq, Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, IN 47906, USA. (

    Search for more papers by this author


[1] We have investigated how an arc-ward increase in bulk mechanical strength in an experimental accretionary prism influences the development, timing, and duration of slip on out-of-sequence thrusts. We have monitored the structural development and kinematics, in side-view, during the development of a frontally accreting Coulomb wedge growing out in front of a critically tapered and mechanically stronger inner wedge. The inner-wedge initially behaved as classic backstop to deformation with the most actively slipping thrust occurring near the deformation front on the forward most thrust structures. With continued growth, however, significant out-of-sequence slip on reactivated fore-thrusts occurred in conjunction with slip on newly formed back-thrusts in the inner-wedge. This out-of-sequence deformation resulted in punctuated, rapid uplift of the model forearc basin and a noticeable break in topographic slope in the outer pro-wedge. Cyclical out-of-sequence fore- and back-thrusting, driven by ongoing frontal thrust imbrication, continued with periodic recovery of taper and was followed by additional out-of-sequence faulting and associated basin uplift.

1. Introduction

[2] The long-term increase in margin-normal width of an accretionary prism occurs primarily by successive imbrication of thrusts at the deformation front [Chapple, 1978; Davis et al., 1983] where this accretion motivates slip on older faults that bound the thrust slices that comprise the wedge. The long-term profile of the wedge is dictated by several basic physical parameters [Davis et al., 1983; Dahlen, 1990], where additional accretion requires the occasional displacement of older thrusts leading to internal deformation and uplift of the wedge to maintain a characteristic constant or critical taper. Modeling and theoretical studies of frictional wedges indicate that, over the short-term, stresses are only periodically at a critical state, as indicated by the episodic motion of the thrusts that comprise them and due to the transfer of mass from subduction erosion or frontal accretion [Gutscher et al., 1998; Wang and Hu, 2006; Del Castello and Cooke, 2007; Cruz et al., 2010]. The net addition of mass by frontal thrust imbrication will, however, eventually necessitate the occasional displacement on existing faults behind the deformation front or the creation of new structures, and is termed out-of-sequence (OOS) thrusting [Morley, 1988]. Used here, OOS thrusting refers specifically to major fault displacements localized well inboard from the deformation front.

[3] Significant OOS thrusting is frequently associated with a break in the surface slope of the wedge from a constant taper and has been observed in a number of accretionary prisms including, Nankai [Moore et al., 2007; Bangs et al., 2009], Sunda [Kopp and Kukowski, 2003], Barbados [Westbrook et al., 1988], along the Alaskan margin [Fruehn et al., 1999] and in Southern Taiwan [Lin et al., 2009]. The location and shape of the crystalline backstop, a mechanically stronger boundary, will strongly influence the style of deformation in the wedge [Silver and Reed, 1988; Byrne et al., 1993]. It has been recognized that increases in mechanical strength occur in the sedimentary wedge during ongoing accretion and that these gradients in strength are often associated with location of major OOS structures [Kopp and Kukowski, 2003]. In this study, we examine the timing and mechanics of OOS faulting in relation to the geometry and strength of an accreted sedimentary backstop. A strong but deformable, or dynamic, backstop arises when there is a stable inner wedge, possibly due to decreases in pore-pressures or the rotation of older faults out of their optimal orientation for slip, or from fault healing landward of an actively deforming outer wedge [Kopp and Kukowski, 2003; Wang and Hu, 2006]. The sudden break in an accretionary prism's taper has also been attributed to variations in strength in the décollement [Wang and Hu, 2006; Miyakawa et al., 2010]. Décollement strength does have an important influence on wedge morphology [Davis et al., 1983; Gutscher et al., 1998], however, in this study we examine the kinematics of wedge growth in response to a mechanical increase in the inner wedge, rather than on the décollement, in the propagation of the OOS faulting.

2. Analog Modeling Methodology

[4] The experiments in this study were conducted in a sand box and imaged through the box's clear glass sidewall. The setup consists of a thin polyester sheet (μb = 0.34) that is pulled through a slot below a vertical backstop to simulate convergence at a subduction zone (Figure 1). In these experiments, the entire apparatus was tilted (β= 6°) toward the backstop to simulate the dip of the down-going plate. Sidewall frictional drag was minimized by the application of a lubricating compound (e.g., Rainx) to the glass sidewalls, which is important in order for the imaged part of the experiment to accurately reflect the overall deformation in the box.

Figure 1.

Schematic of sand box experiment setup (31 cm wide by 150 cm long). The rigid backstop is a wall that is 90° to the base of the box, and the deformable backstop is a wedge of sand with an experimentally determined tapered profile that is governed initially by the same peak strength as the incoming undeformed sand and has no preexisting shear zones. In these models the basal friction (μb) is 0.34, for sand on polyester film (red dashed line), and the peak internal friction angle for the sand is ∼34° (μpi = 0.68), with a stable sliding friction ∼31° (μsi = 0.61).

[5] The basis for analog Coulomb wedges is in the linear scaling of pressure dependent, rate independent, frictional failure [Byerlee, 1978] and a robust theoretical justification [e.g., Davis et al., 1983, Dahlen, 1990] that describes the framework for their long-term growth. In these experiments, a well sorted, sub-rounded, cohesionless sand is used to simulate the deforming brittle upper crust [e.g.,Davis et al., 1983; Lallemand et al., 1994] and is appropriate for simulating deformation in the upper lithosphere [Lohrmann et al., 2003] where frictional mechanisms dominate. The self-similarity of deforming convergent systems leads to linear scaling in length of frictional experiments. These experiments are time-independent for the experimental rates used (typically 0.17–0.4 mm/sec). The peak strength of the frictional material is governed by both the internal friction angle (Φpi), where the tan(Φpi) = μpi (or 0.68), and the manner in which the material is placed in the box (e.g., sifted vs. poured) [Lohrmann et al., 2003]. The sand deforms by the development of dilatant shear zones (faults), which exhibit net strain weakening and have the potential to be reactivated throughout the duration of the experiment governed by its lower stable sliding internal friction (Φsi), where the tan(Φsi) = μsi (or 0.61). The pervasive thrusting at a variety of angles, therefore, lowers the bulk internal strength of the model wedge (to that of Φsi) compared to a wedge with the same profile and no initial internal shear zones, which would initially be governed by Φpi. To determine the shape for our mechanically stronger backstop, we ran an initial experiment to obtain a topographic profile compatible with our décollement and sand friction coefficients (Figure S1 of the auxiliary material). In the subsequent experiment, we prepared a critically tapered backstop with this profile by sifting sand to produce the inner-wedge, which produced a small gradient in strength, so the inner wedge was initially governed by peak friction instead of stable sliding friction of preexisting shear zones.

3. Kinematic Analysis

[6] The temporal evolution of the experiment is tracked by the analysis of a series of images taken through the clear glass sidewall at small intervals of convergence. The coordinates of passive colored markers layered in the sand, with the same material properties, are determined using a series of algorithms that are optimized here for side-view and result in a precise color-weighted centroid location for each marker [e.g.,Haq and Davis, 2009]. By correlating marker positions for the span of the experiment we can determine a robust velocity field for any interval in which comparable points are present. With these velocities we also calculate an interpolated shear rate field using the technique of Haines and Holt [1993]. The cameras used in these experiments (5184 × 3456 pixels) allow for the determination of coordinates to a precision of about 60 microns over the imaged region. This spatial and temporal resolution allows us to ascertain the exact timing of thrust initiation and to assess the distribution of slip in the entire wedge during frontal accretion.

4. Results

[7] The first experiment (Figure S1 of the auxiliary material), shows the development of a single sided wedge that is the basis for the backstop in the main experiment (Figure 2 and Figure S2 of the auxiliary material). This baseline wedge exhibited self-similar and predictable deformation, where thrust activity was largely restricted to the interface of the first two frontal thrusts during its development. At any given time in the cycle of frontal accretion the youngest (most frontal) thrust accommodated the bulk of the convergence on its bounding surfaces. Late in the experiment (after 9 thrusts) there was some minor OOS fore-thrusting, which accounted for a small amount of the total contraction in that thrust interval. Similar amounts of OOS displacements have been noted in other analog studies [e.g.,Lohrmann et al., 2003]. The development by largely in-sequence thrusting, in this case, was reflected in the long-term, generally constant trajectories of material in the thrust slices and the constant taper of the wedge.

Figure 2.

Image series showing stages of wedge development at the formation of each new thrust, with 8 total. The velocity field is normalized by convergence, and plotted with the associated shear strain rates (red is CCW and blue is CW) over the experiment. Intervals (Figures 2g and 2h) show significant OOS fore- and back-thrusting. The velocities and trajectories define three distinct regions: the inner, and outer wedges, and the undeforming portion of the backstop. Figure 2j has cumulative particle motion for select points in the wedge; larger dots indicate the end of a thrust interval. Movies S1 and S2 of theauxiliary material show the evolution of these models overlain with velocity and particle motion, respectively.

[8] The critically tapered profile obtained from this first experiment (Figure S1 of the auxiliary material), which was then used in the main experiment, is based on a wedge with strain softened shear zones or a bulk strength that is lower (governed by μsi = 0.61) than an equivalently tapered sifted wedge with the same profile but starts with no shear zones. Using the profile of the wedge governed by stable sliding friction (μsi) as a starting point for the second experiment (Figure 2) made the inner-wedge, before additional deformation, stronger or super-critical. This small increase in bulk strength of the wedge inFigure 2 is used as a proxy for the aggregate effect of fault healing, lithification of accreted sediments, and the rotation of older fault structures out of their optimal orientation for slip as occurs in nature over time. All these commonly occurring factors in nature have been suggested as potential mechanisms for the development of the observed gradient in strength in the inner part of accretionary wedges [Byrne et al., 1993; Kopp and Kukowski, 2003].

[9] Deformation in this experiment initially localized at the inflection of topography at the base of the inner-wedge (Figure 2a). The kinematics of the first five thrust cycles were similar to the experiment that generated the inner-wedge (e.g., Figure S1 of theauxiliary material), with fault slip largely restricted to the frontal thrusts. During this initial growth several frontal structures were observed to be simultaneously active near the front, however, deformation was overwhelmingly focused on the frontal thrusts as indicated by the high shear strain rates (Figures 2a–2e) and the gradients in the associated velocity fields. One initial difference from the baseline experiment at this stage was the small amount of slip distributed on several back-thrusts in the older inner-wedge. This back-thrusting occurred in direct response to accretion at the front and was responsible for thickening the inner-wedge just behind the first new thrust (T1). Individually these back-thrusts (B1 to B3), as indicated by their orientation and slip (Figure 2i), are responsible only for minor uplift and accommodate a small fraction of the total contraction in the convergence interval that included the development of the first five new frontal thrusts (T1 to T5).

[10] During the interval that accommodates the sixth frontal thrust (T6) OOS thrusting began in earnest, with the first formed outer-wedge fore-thrust (in interval T1) being reactivated in conjunction with back-thrusting in the inner-wedge. At this point, when both the OOS fore- and back-thrusts are simultaneously active, the steep back-thrusts are at a characteristic angle to the shallower OOS fore-thrust suggesting that they are forming as conjugates as would be predicted for a moderately strong décollement [e.g.,Davis and Engelder, 1985; Bonini, 2007]. Their relative orientations and friction on the décollement suggests a maximum principal stress (σ1) axis with a shallow plunge toward the deformation front. During the same thrust interval a new back-thrust forms further inboard of the older wedge. This last OOS back-thrust is periodically active for the remainder of the experiment in conjunction with several different OOS fore-thrusts. All of these OOS fore-thrusts have a dip between 23°–24° and the most active back-thrust (B4) starts with a dip of 25° and ends with a dip of about 23° (Figure 2i). The dips and slip on these structures cause a large part of the inner wedge to be rapidly uplifted in the latter OOS thrusting cycles.

[11] During the last thrust interval (T7 and T8) the locus of highest shear strain periodically moves from the frontal thrust inboard to the OOS thrusts. A large amount of slip is transferred, by bulk translation of the pro-wedge, onto the reactivated fore- and back-thrusts. This cyclical nature of OOS shear localization can be seen very clearly over the smaller intervals of deformation in Figure S2 of theauxiliary material(for two full thrust cycles T6 to T8 at ∼10 mm intervals). The cycle begins with under-thrusting and imbrication at the wedge front, followed by OOS fore- and back-thrusting. While the location of the most active and final back-thrust (B4) appears stable relative to the position of the original rigid backstop the OOS fore-thrust migrates toward the deformation front, occurring on several distinct structures. The final and most active OOS fore-thrust (Fo3) and back-thrust (B4) have relatively shallow dips, in the opposite direction, of about 23°–25°. Their slip is coeval and driven by accretion at the front and leads directly to periodic rapid uplift and arc-ward migration of the model forearc basin, suggesting a nearly horizontal maximum principal stress (σ1) axis in the rear of the wedge. The mechanical justification for this interpretation is based on the disparity between peak and stable strength of the inner and outer wedges and is explained by Figure 3. The difference in strength allows for a range of potentially simultaneously active fault planes relative to the décollement.

Figure 3.

(a) Mohr circle illustrating the range of potential active failure planes in accreting analog experiments. The range, darker gray area, of active planes is constrained by the peak (μpi = 0.68) and the stable sliding (μsi = 0.61) internal friction criterions of the sand. New structures initiate at orientations compatible with the peak frictional strength of the sand and are subsequently active governed by lower stable sliding friction allowing for a range orientations over which the faults can continue to actively slip. When measured with respect to the décollement [after Davis and Engelder, 1985] (b) the wide orientation of potentially active thrusts (bounding dashed lines), include newly initiated frontal, reactivated, and new optimally oriented OOS faults that can accommodate thickening in the wedge. The basal friction (μb = 0.34) and dip (6°) are compatible with a nearly vertical σ3 direction. δb and δf are dips w.r.t the décollement and ψb is the orientation of σ1 to the décollement.

5. Discussion

[12] The absence of exploitable structure, in the critically tapered backstop, that can be used during continued growth to maintain force balance in the wedge is key in the development of OOS fore-thrusting in these models. The experimentally determined backstop (and taper) with no internal shear zones resulted first in the development of OOS conjugate thrust faults, where their orientation was determined by the peak internal and basal friction and basal dip [e.g.,Davis and Engelder, 1985]. The steep back-thrusts that initially formed in response to wedge growth were active only briefly to accommodate uplift near the front of the wedge. This was unlike the baseline model (Figure S1 of theauxiliary material), which could be thickened by incremental slip on any of a number of fore-thrusts that were present. In order to efficiently thicken the stronger inner wedge inFigure 2, the location of OOS back-thrusting quickly moved inboard localizing the majority of back-slip motion on a single shallower back-thrust that was more closely compatible with failure at the peak frictional strength of the sand. In each case successive OOS fore-thrusting initially reactivated an older in-sequence thrust but eventually cut across existing structures, particularly towards the surface where the most fault rotation had occurred after initial in-sequence formation. With continued outer-wedge growth additional inner-wedge thickening was required in order to maintain a functional taper or force balance in the pro-wedge [e.g.,Davis et al., 1983] which was obtained by the combination of slip on the OSS fore- and back-thrusts. Rapid uplift occurred primarily on a shallower back-thrust, which also caused the migration of the model forearc basin landward. In Southern Taiwan OOS thrusting has been identified and is coincident with back-thrusting below the Southern Longitudinal Trough and forearc basin [Lin et al., 2009]. Rapid uplift of the forearc basin, in relation to mega-splay development, has also been documented in Kumano Basin along the Nankai trough [Gulick et al., 2010]. These modeling results validate a plausible mechanism, inner-wedge thickening, to explain these observations of major OSS fore-thrusting.

[13] The consolidation of the inner-wedge doesn't allow incremental slip on preexisting faults, which would occur in a wedge with numerous faults and would produce only very distributed OOS thrusting (e.g., Figure S1 of theauxiliary material) with limited slip on any single structure. In nature, the rapid uplift of the inner-wedge and associated back-thrusting could be obscured or mitigated by the capture and increase in thickness of sediment in the forearc basin during wedge development. In the context ofFigure 3, sediment loading over the inner wedge would primarily increase σ3, which is close to vertical. The increase in σ3 would move the inner wedge away from failure, however, convergent motion and thickening would still need to be accommodated by OOS fore-thrusting in the outer wedge. As inFigure 2, the initial deformation forward of a stronger inner wedge leads to a sudden break in slope. This transient initial slope break would be instrumental in the capture of sediment and localized loading of an inner wedge basin. Additionally, it is also possible that because at many margins there is limited seismic resolution on the deeper structure of the forearc basin, where these back-thrusts would reside, these structures may not be as easily observed as frontal OOS structures. It is also notable in these models that OOS thrusting occurs even with no mechanical gradient in décollement strength. Such a variation would likely increase the contrast in taper angle between the inner and outer wedge in these models, and primarily influence the shape of the wedge front [Zhao et al., 1986]. A more bimodal distribution of strength on the décollement should also reduce the amount of uplift in the forearc basin in favor of slip on frontal structures. The development of strong back-thrusting in an initially single-sided wedge suggests an inevitable transition from forward vergence to a bi-vergent system with time, which is commonly associated with retro-vergence and migrating back-thrusting [Wang and Davis, 1996].

6. Conclusions

[14] Relatively small variations in landward strength in the wedge can lead to significant reorganization of deformation over the long-term evolution of an accretionary prism. Large variations in topographic slope and rates of uplift are evident in a wedge's short-term response to accretion but can be obscured by its long-term evolution towards a uniform taper. A landward gradient in strength gives rise to periodic, major OOS thrusting to satisfy the need for thickening in the model inner wedge and forearc basin to maintain force balance, if not explicitly a wedge taper, and results in its rapid punctuated uplift. The basal dip of the subduction zone is important in determining the early orientation of OOS faults but the long-term orientation of back-thrusting, which begins as conjugates to fore-thrusting, is more sensitive to the cumulative accretion. Inner-wedge thickening will be distributed ideally on optimally or sub-optimally oriented preexisting structures, fore-thrusts primarily. However, if those structures are not favorable for slip new structures must develop to accommodate deformation. In these experiments, driven largely by frontal accretion the inner-wedge first thickens by multiple short-lived back-thrusts and, later with additional accretion, by localized back-thrusting coincident with major out-of-sequence fore-thrusts. OOS fore-thrusting will migrate towards the deformation front of the pro-wedge in response to wedge growth and has a transient surface expression as a break in topographic slope or taper of the wedge.


[15] I would like to acknowledge the helpful reviews by Sean Gulick and an anonymous reviewer as well as useful conversations with Dan M. Davis. This work was supported by Purdue Research Foundation grant.

[16] The Editor thanks Sean Gulick and an anonymous reviewer for their assistance in evaluating this paper.