11:00 AM - 1:00 PM
[SCG52-P13] A statistical approach to quantifying dynamic recrystallisation in naturally deformed quartz based on EBSD analysis
Keywords:dynamic recrystallization, recrystallised microstructure, electron backscattering diffraction (EBSD), cluster analysis
Microstructures of minerals affected by plastic strain evolve by dynamic recrystallisation. The evolution of these microstructures would greatly alter the mechanical properties of the minerals. As plastic deformations are often reported to occurs heterogeneously, quantifying the dynamic recrystallisation fraction becomes important for the application of palaeo-piezometry and for identifying strain. However, current methods to discriminate dynamic recrystallisation fraction are based on an assignment of an arbitrary value for a certain criterion.
The goal of the presentation is to establish a statistical approach in partitioning recrystallised and relict grains that had undergone dynamic recrystallisation. We present an unsupervised clustering method on grain reference orientation deviation (GROD) parametres using gaussian mixture model. The parametres applied for the clustering include grain orientation spread (GOS) and the normalised GOS (GOS/Diametre) calculated from the ratio of GOS and grain diameter while the number of clusters were determined by means of Bayesian Information Criteria (BIC). We found that the relict grains could be confidently recognised based on GROD values distributed larger than that of the three standard deviation value from the cluster with the largest mean for both GOS and normalised GOS.
In this presentation, we will apply the above-mentioned method on quartz domains found within the Ryoke mylonites along the Median Tectonic Line, SW Japan. (Katori et al, 2021) Our samples include protomylonites to ultramylonites that have experienced a different degree of strain. We will demonstrate how the results of the clustering could be reflected in the interpretation of grain evolution in terms of grain size, misorientation angle, recrystallised fraction, crystallographic preferred orientation, etc during dynamic recrystallisation.
The goal of the presentation is to establish a statistical approach in partitioning recrystallised and relict grains that had undergone dynamic recrystallisation. We present an unsupervised clustering method on grain reference orientation deviation (GROD) parametres using gaussian mixture model. The parametres applied for the clustering include grain orientation spread (GOS) and the normalised GOS (GOS/Diametre) calculated from the ratio of GOS and grain diameter while the number of clusters were determined by means of Bayesian Information Criteria (BIC). We found that the relict grains could be confidently recognised based on GROD values distributed larger than that of the three standard deviation value from the cluster with the largest mean for both GOS and normalised GOS.
In this presentation, we will apply the above-mentioned method on quartz domains found within the Ryoke mylonites along the Median Tectonic Line, SW Japan. (Katori et al, 2021) Our samples include protomylonites to ultramylonites that have experienced a different degree of strain. We will demonstrate how the results of the clustering could be reflected in the interpretation of grain evolution in terms of grain size, misorientation angle, recrystallised fraction, crystallographic preferred orientation, etc during dynamic recrystallisation.