The Role of Preexisting Basement Structures During Continental Collision: Insights from Analog “Sandbox” Modeling

Caroline Fernandez, Kathryn Boardman, Angelica Alvarez

see our research proposal posterhere

Initial Observations:

In the article, "Impact of erosion, sedimentation, and structural heritage on the structure and kinematics of orogenic wedges:Analog models and case studies" (Malavielle, 2010), recently discussed in Advanced Structure various compressional models were constructed and analyzed. These models, however, always considered the basement to be detached. From structural geology, we know that normal faults can be reactivated as reverse faults during compressional events. These observations have led us to question how normal faults in the basement rock would effect the geometry and evolution of an accretionary wedge in the processes of mountain building.


When a previously normal faulted basement layer is incorporated into a compressional analog model, we expect the faults to reactivate easily as reverse faults and change the shape and evolution of the accretional wedge. The faulted blocks should start to pop-up as shortening occurs, changing the geometry of the wedge and therefore effecting how the wedge accommodates the horizontal stress.

Experimental Design:

To properly study the role of pre-existing basement structures, we plan to run at least three experiments. The first experiment will be an end-member compressional model, intended to represent continental collision where the basement rock is detached and therefore does not affect the accretionary wedge. The other two models will incorporate a section of rigid board at the base of the sand which will function as the basement. The basement will consist of four separate blocks, constructed by making three cuts at 60 degree angles through the board to simulate normal faults. One model will have normal faults dipping in the same direction creating uniform fault blocks. While the other will have a fault that dips in the opposite direction, forming "Horst" and "Graben" structures. The compression in the analog models is achieved by the incremental movement of one of the walls of the box by a hand crank. The results are viewed from plexiglass or glass windows on either side of the sandbox. The first model is our only sandbox that has plexiglass windows instead of glass windows. To prevent the colored sand from smudging against the plexiglass, we sprayed several layers of "Static Guard" to the inside of the windows before filling the box with sand. This step is not necessary if using glass for the windows. The layering of the sand is maintained as uniform as possible with the use of a tape measure and a flat edged tool. Each sandbox will have two distinct colored sand layers so that the evolution of the wedge can be tracked. The shortening will be conducted in 2 centimeter intervals. At each interval the following data will be collected; top and side view pictures, fault characteristics (dip, length, sequence), and maximum wedge thickness.


The goal of these analog models is to help us better understand the role of inherited structures, specifically normal faults, of basement rock in the formation of orogenic wedges. We hope that our models will correlate with existing orogenic wedges so that we can perhaps contribute to the better understanding of their geologic history.

Data and Discussion:

Model 1: End-member Compressional Analog Model without Basement Structures

Purpose- This model is our control experiment for compassion, representing the upper crust during continental collision when the basement rocks are not involved in the deformation.
Design- This model is composed of layered homogeneous sand. The original length of the sand section is 70 centimeters and the initial thickness of the sand was 6.8 centimeters. Data was collected at each interval until the sandbox was shortened by 24 cm.

Percent Shortening described in terms of extension= e%= Δl/l(o) * 100

Initial Conditions (picture)

Results- The accretionary wedge grows through frontal accretion of new tectonic units involving forward propagation along low angle trust faults.

Discussion- The accretionary wedge of our end-member model exhibits a comparatively high angle of taper or critical angle. The critical angle of an accretionary wedge is responsible for the geometry and shape of the wedge. It is a characteristic dependent on the material (lithology-rock types) of the wedge. However, Model 1 does not display imbrication or back-thrusts which according to Malavielle's analog model experiments, should be expected in wedges with a high taper angle because deformation within the wedge is more easily achieved then deformation along the basal decollement (detached basement). It is our belief that Model 1 receives its high taper primarily from the heavy sediment load put upon it, resulting in a higher basal friction. This means that the ability of the wedge to deform horizontally along the decollement is made difficult by the increased load, therefore material builds up and forms a high taper angle. The lack of vertical deformation (imbrication, back-thrusts) and how the wedge maintains its critical angle suggests that deformation within the wedge is even more difficult then horizontally along the basement. The critical angle is maintained through frontal accretion which decreases the taper angle when a critical angle is reached, forming faults as rigid material breaks (fails) and is propagated forward along planes of weakness. Case studies of this model demonstrating an increased load would exist along continental collisions where a large proportion of dense (mafic) minerals are incorporated into the continental (upper plate) crust from migration of magmatic fluids or in environments with slow erosion rates. Understanding this concept is fundamental in exploration for mineral deposits.

The in-sequence faulting is characteristic of ramp-flat geometry and exhibits a step-like or foreland propagation pattern. Flats are fault surfaces that form parallel to the strata and usually in weak rock units, such as evaporites and shales. Ramps cut across more resistant rock units, for example sandstone and limestone, forming a dip angle that is typically 30 to 45 degrees. With a ramp in place, new cracks can propagate forward from the base of the ramp and continue until they ramp and join a higher flat. In our experiments the fault (floor thrust) follows along the basal decollement (detached basement), and cuts up through the layers of sediment (sandstone) until it reaches the surface. The ramp produces an upthrust of older rocks, so there will be a sequence with older rocks on top of younger. These ramp-flat sequences are important to petroleum geologists because the hanging wall structures can act as hydrocarbon traps, for both oil and gas. Faults that previously acted as conduits for the migration of oil and gas up through the crust, could latter form traps as rock materials become cemented resulting in a decrease in pore space.


18 Centimeters (25.7% of Shortening)
Side 1


Ѳ(1) = angle with respect to y axisTop-18(2).JPG
Ѳ(2) = dip angle
S(y) = magnitude of stress component acting on y

S(x) = magnitude of stress component acting on x
F(xo) = force acting in direction of OX
F(yo) = force acting in direction of YO
L = length of the fault plane

α = taper angle
σ = stress vector

Fault trace and direction of transport

Dynamic Analysis
- The goal of dynamic analysis in structural geology is to understand the stress conditions that trigger faulting so that we can predict how materials will respond (strain) to an imposed stress. The forces and stresses involved in a our end-member model of an accretionary wedge formed during a compressional event are the confining pressure and the lithostatic stress. The confining pressure is the amount of force divided by the area needed to compress and shorten the wedge at increments of 2 cm using an arm crank. The lithostatic stress is the weight (m*g) of the load divided by the area. Therefore, the magnitude of principle stress (σ ) is not simply a function of the force (F) from which it was derived, but also related to the area on which the force acts. Stress will permanently deform a body of material only if the strength (cohesive, tensile, internal friction) of the body (rock) is exceeded. Whether a plane of a certain orientation will become a fault significantly depends on the magnitudes of normal and shear stresses on that plane. For example, normal stress that is high compared to shear stress tends to inhibit movement on the surface. Whereas, high shear stress tends to promote fault movement. In order to predict whether faulting will occur on a given surface, we must determine the values of normal and shear stress. To do this we use the derived stress equations:

normal stress ( σn ) = [(σ(1) + σ(3))/2] - [((σ(1) - σ(3))/2) * (cos2Ѳ)] shear stress (σs) = [(σ(1) - σ(2))/2] * sin2Ѳ

Sigma 1 and 2 are the principle stress directions, also known as the axes of the stress ellipse. Sigma 1 is the direction of greatest principle stress, while sigma 3 is the direction of least principle stress. For a compressional event, the maximum principle stress is oriented horizontally in the earths crust as explained in Anderson's theory of faulting.


The diagram shows that the state of stress for compressional settings is ideal for the formation of faults with dip angles near 30 degrees. This is demonstrated in our end- member compressional model by the formation of low angle thrust faults as illustrated in the graph below (21 to 38 degrees) The formation of new faults can be determined for a given state of stress through the use of mohr circle diagrams and the principles involved in Coulomb' Law of Failure for intact rock.


Model 2- Compressional Analog Model with Pre-existing Normal Faults (dipping same direction)

Purpose- To study the role of pre-existing, 60 degree, normal faults in the basement on the evolution of an accretionary wedge as compared to the end-member model.

Design- This model was our first experimental model with rigid board sections at the base of the sand to represent the faulted basement. The model had the same sand layers as model 1. In addition, the original length and thickness of the sand was the same as in model 1 (length = 70cm, thickness=6.8cm). Data was collected at each interval until the sandbox was shortened by 24 cm.

Initial Conditions


Results- The accretionary wedge grew through frontal accretion of new tectonic units involving forward propagation along low angle thrust faults, a single high angle back fault, and imbrication of long tectonic unit The normal faults in the basement did not reactivate as expected. However, the basement board did act as a ramp and prevented the development of thrust faults over the rigid board. The continuation of shortening should result in imbrication and a large incline (ramp) at the surface. Inactivation of the basement structures could be due to the heavy lithostatic load, the surface of the boards (representing basement rocks) resistances to slip, or because the field of stress (see Model 1 Dynamic Analysis) was not ideal for the reactivation of normal faults as reverse faults. Surface roughness inhibiting reactivation could be caused by insufficient stress, resulting in the inability to overcome the strength of the rock, or cementation of pre-existing faults. For example, the increased load or depth to basement could increase the confining pressure which in turn increases the strength (yield,ultimate,rupture) and ductility of the rock. One way the strength of the rock can be decreased is by the presence of an elevated hydrostatic fluid pressure by rising magmatic fluid.

24 Centimeters (34.3 % shortening)

Side 1


12 Centimeters (17,1% of Shortening) Side 2

The back fault is not a back thrust which is what we would have expected to form in the wedge, but a reverse fault with a dip angle near 60 degrees. It could be that the presence of the wall forces a high angle fault. The fault forms in reaction to the transport of sediment in Fault 1, making room for the build up of sediment by "poping up" the wedge. The sediment layers follow the movement on Fault 1 at the intersection. We can not determine whether or not the pre-existing structures play a role in the formation of the back fault. This may be why a back fault does not develop in model 1.

Comparison with End-Member model

Graph on Left- At 14 cm of shortening fault propagation in model 2 aligns with pre-existing normal faults in the basement, resulting in the impediment of new fault development. Fault 4 develops, but by imbrication of the previous tectonic unit (fault 3). While model 1 exhibits a continuous forward propagation of faults. However, the amount of shortening needed to produce new faults increases as the faults move away from the area of maximum thickness in the accretionary wedge.

Graph on Right- Model 1 and 2 show about the same change in maximum thickness of wedge with change in shortening. This suggests they have the about the same critical angle that the wedge is trying to maintain. Therefore, we can conclude that the presence of basement structure (ramp) does not change the geometry or shape of the wedge. It does change the evolution, or how the wedge maintains its critical angle. Model 1 by forward propagation of new tectonic units, and Model 2 by cessation of new fault development and then imbrication of long tectonic units. The processes in Model 2 forms better trap structures due to the extensive folding of the long tectonic unit.


Graph Above- The dip angles of Model 2 range from 16 to 40 degrees, a slightly larger range then Model 1. Most of the variation occurs at 20 to 24 cm of shortening, suggesting that this is caused by the development of the ramp which pushes up the fault planes and therefore increases their dip angle. A pattern was also noticed in the progress of dip angles. One where the dip of each previous fault successively drops before the formation of a new fault, and then they increase again. This is the processes of the wedge maintaining its critical angle by forward propagation of new tectonic units through faulting.

Proposed Modifications to the Experimental design: Based on the results from model 2, we feel that certain changes to the experimental design will improve the chances of fault reactivation in the basement board. Additional models will incorporate one or more of the following proposed modifications to the sandbox model design;

1) Lubricate the faults (achieved by painting the cut edges of the board to achieve a smooth surface and adding graphite to make the surfaces slippery) representing faults acting as conduits for fluid flow, as opposed to faults that are cemented together.
2) Add an additional board section to move the pre-existing faults significantly closer to the moving wall of the sandbox, therefore increasing the amount of stress acting on it.
3) Decrease the thickness of the sand so that there is less crustal loading resisting the reactivation of the basement.
4) Use low-angle faults, like 30 degree thrust faults, as the pre-existing faults in the basement. Low-angle faults should reactivate more easily under compression than high-angle faults.

Model 3 :

Model 3: sandbox before shortening. Modified to have a longer board section and lower-angle faults (approx. 30 degrees) in the basement.

Model 3: Model after 14 centimeters of shortening. First faulted block (nearest the moving wall) has been displaced upward along the reactivated fault.

Model 3: Model after 22 centimeters of shortening. Both faulted blocks have been displaced along the leading fault. Both basement faults were reactivated and the blocks ended up one on top of the other. (please note that the lower right corner of each block is still visible while the rest of the block has been obscured by falling sand between the blocks and the glass window.)