3:30 PM - 5:00 PM
[SCG56-P06] Automatic determination of number of stress conditions from heterogeneous fault-slip data
Keywords:fault-slip analysis, stress tensor inversion, information criterion, friction coefficient
The crustal stress states have been investigated by applying stress tensor inversion techniques to fault-slip orientation data. The data set is called heterogeneous when it includes faults ascribed to multiple stress states. Fault-slip data from geological outcrops are likely to be heterogeneous reflecting spatiotemporal variation of stress conditions. However, the stress tensor inversion techniques have not fully succeeded in automatic determination of numbers of stress tensors recorded in heterogeneous fault-slip data.
This study attempted to determine the number of stress conditions by utilizing the Bayesian information criterion, which has been already used in stress tensor inversion techniques from dikes, veins and calcite twins. This approach requires a probability distribution model for fault-slip data. Since the stress tensor inversion is based on the Wallace-Bott hypothesis which constrains only slip directions to be parallel to shear stress vectors, the orientations of fault planes are usually free. Then the direct modelling of probability distribution of fault-slip data is not suitable in this case. Therefore, this study proposes to model the distribution of objective function for inversion. The function is composed of the misfit angles between slip directions and shear stress vectors. When the orientations of fault planes have some concentrations, we can join fault instabilities to the objective function to estimate friction coefficients.
The new method was tested by analyzing an artificial heterogeneous fault-slip data set including 50 faults caused by a N-S compressional stress and another 50 faults caused by a NE-SW compressional stress. As the result, the number of stress conditions was correctly chosen to be two by the Bayesian information criterion and the stress conditions were also correctly estimated.
The method was applied to a natural fault-slip data set from the Pleistocene Sekinan Group, southwest Japan which filled the Beppu-Shimabara graben. Two normal-faulting stress conditions were detected with N-S trending and ENE-WSW trending tensional axes. Since the overlaying Oita Group includes faults with good agreement with N-S tension, the result suggests a stress transition from ENE-WSW tension to N-S tension around 1 Ma.
This study attempted to determine the number of stress conditions by utilizing the Bayesian information criterion, which has been already used in stress tensor inversion techniques from dikes, veins and calcite twins. This approach requires a probability distribution model for fault-slip data. Since the stress tensor inversion is based on the Wallace-Bott hypothesis which constrains only slip directions to be parallel to shear stress vectors, the orientations of fault planes are usually free. Then the direct modelling of probability distribution of fault-slip data is not suitable in this case. Therefore, this study proposes to model the distribution of objective function for inversion. The function is composed of the misfit angles between slip directions and shear stress vectors. When the orientations of fault planes have some concentrations, we can join fault instabilities to the objective function to estimate friction coefficients.
The new method was tested by analyzing an artificial heterogeneous fault-slip data set including 50 faults caused by a N-S compressional stress and another 50 faults caused by a NE-SW compressional stress. As the result, the number of stress conditions was correctly chosen to be two by the Bayesian information criterion and the stress conditions were also correctly estimated.
The method was applied to a natural fault-slip data set from the Pleistocene Sekinan Group, southwest Japan which filled the Beppu-Shimabara graben. Two normal-faulting stress conditions were detected with N-S trending and ENE-WSW trending tensional axes. Since the overlaying Oita Group includes faults with good agreement with N-S tension, the result suggests a stress transition from ENE-WSW tension to N-S tension around 1 Ma.