Implementing Platerecipy: an open access tool utilizing a graph theory method for detecting tectonic plate boundaries in geophysical data sets and numerical model output

The emergence of plate-like surface motion in self-consistent mantle convection models is a key behaviour requiring detection in numerical experiment results featuring terrestrial characteristics. However, the identification and verification of candidate plates is a challenging task, in practice. On Earth, narrow divergent, convergent, and strike-slip plate boundaries as well as regions exhibiting widespread diffuse deformation, comprise roughly 10 to 20% of the lithosphere that does not adhere to rigid body motion. Accordingly, the detection of candidate plates must be performed in light of the existence of diffuse deformation occurring regularly as a tectonic characteristic. To address this challenge, we have recently developed a new plate detection tool, `platerecipy`, that utilizes the Random Walker (RW) segmentation algorithm to identify candidate plates in both mantle convection model output as well as global geophysical data sets and terrestrial measurements. We describe how the discrete probability solution arising from RW can be used to both assess confidence in the association of each location with a distinct rigid plate, and to identify diffuse surface regions. Furthermore, we show how utilizing the RW probabilities can significantly improve Euler vector inversion for fitting the plate motion as a probability field allows for a systematic means of incorporating uncertainties inherent to the plate detection process. We demonstrate the effectiveness of our method by applying it to the surface of a mantle convection model and a terrestrial strain-rate dataset. We show how our findings can be used for an Euler vector inversion that allows plate rigidity analysis.