Evolution of fault patterns within a zone of pre-existing pervasive anisotropy during two successive phases of extensions: an experimental study

Abstract

The aim of the present work is to study the influence of preexisting pervasive strength anisotropy on the development of faults during two phases of extensions. Two different series of experiments are performed by deforming rectangular three-layered models either by orthogonal extension followed by oblique extension (series 1) or by oblique extension followed by orthogonal extension (series 2). The model represents a rectangular zone of rifting. The final fault architecture after two successive phases of extension is primarily controlled by the orientation of the pervasive strength anisotropy. The mode of far-field stress (orthogonal or oblique) plays a role in fault initiation during both the phases of extension. The growth of the faults which are orthogonal or oriented more obliquely (β = 45°/60°) with respect to the rift normal is controlled by the direction of extension. However, the less oblique faults (β = 15°) develop as strike-slip faults irrespective of the direction of extension. The phase 1 faults reactivate during the phase 2 extension only when they are parallel to the preexisting pervasive anisotropy. New faults parallel to the rift axis form only if the phase 2 extension is orthogonal (series 2). It is found to happen irrespective of the orientation of the strength anisotropy and of the 1st phase faults. Those faults act as linking faults for the highly oblique (β = 45°/60°) phase 1 faults. New faults are formed following the anisotropy during both orthogonal and oblique phase 2 extension only if the anisotropy is oriented at low angle (β = 15°) with the rift normal. The different fault patterns developed in the experiments can be matched well with natural examples reported from Karonga basin, Malawi rift, Kenya.

Appendix

Scaling of viscous material (pitch)

A model is well scaled to their natural analogues, when the model and its natural prototype remain geometrically, kinematically and dynamically similar (Hubbert 1937; Ramberg 1975). To achieve scaling condition, the model material has to be very weak as experiments are run on real time which is much faster than geological deformations. Geometrical similarity is represented by the model ratio of length (length of the model/length of equivalent natural prototype). For our setup, model ratio of length (λ) = l_m/l_n = 0.2 × 10⁻⁵ (calculation shown in the Table 4). Kinematic similarity is represented by the model ratio of time (τ). Time ratio is calculated as the ratio of time needed for achieving a certain amount of deformation in the experiment and time needed to achieve the same amount of strain in the corresponding natural situation (Hubbert 1937, p1467). For our setup, model ratio of time (τ) = t_m/t_n = 3.8 × 10⁻¹¹ (calculation is shown in the Table 4). Hubbert (1937) suggested that in case of very slowly moving viscous models, the forces due to inertia can be neglected without causing significant error. Consequently, dynamic similarity can be achieved if model ratios of length (l), mass (m), and time (t) are chosen arbitrarily and forces, viz. stress, pressure, shear strength, and viscosity are made to conform to the ratio μ = δλ³, where μ, δ, and λ are model ratios of mass, density and length, respectively (Hubbert 1937, p. 1489). In the present case, we have used viscous models (pitch) under slow strain rate (≈ 10⁻⁴/s). Therefore, we have assumed negligible inertial forces following Hubbert’s argument. When we compare the model viscosity ratio calculated from the fundamental model ratios (ζ = δλτ = 0.34 × 10⁻¹⁶) with the actual model ratio of viscosity found in model and its natural analogue (e.g., ξ = 0.51 × 10⁻¹⁶), they are found to be within the same order of magnitude (Table 4). Thus, our models achieve dynamic scaling at least approximately, and the results of our experiments can be reasonably extrapolated to natural situations.

Table 4 Model parameters and scaling ratio