AUTOMATICALLY IDENTIFYING AND EXAMINING GEODYNAMIC FEATURES IN PLATE TECTONIC MODELS

Plate tectonic reconstructions facilitate visual links between geokinematics and geodynamics of mappable features and the symptoms of geological processes reflecting the evolution of the Earth. For example, geodynamic process episodically fertilise crust or mantle and strongly influence the location of economic mineralisation. Geological processes, plate boundaries and their locations through time from the geosphere impact the behaviour and the interaction among the atmosphere, biosphere and hydrosphere. Accurate plate tectonic boundaries for any reconstruction model provide researchers a mechanism to validate and improve plate models but also to constrain the location of important geological processes. In most detailed reconstruction models, crustal blocks are represented by complex shaped geodynamic unit (GDU) polygons, and the motion of preserved crustal blocks older than the Mesozoic is complicated. There is also very little preserved evidence of oceanic blocks prior to the Mesozoic. Thus, there are no detailed reconstruction models with continually closing plate boundary topologies prior to the Mesozoic and late Paleozoic. The current manual methodology of geodynamic boundary identification is very time consuming and easily introduces errors when applied to detailed models. We are developing methods to automate the procedure of identifying and delineating divergent zones (rifts and mid-ocean ridges), convergent zones (subduction zones) and transform faults using a variety of open source software code. Different approaches have been evaluated for optimization in run-time and memory for general laptop or desktop computer systems. Our approach also promotes the integration of multiple geo-information layers for machine learning to derive additional insights and to further constrain the interaction among tectonic plates in the development of reconstruction models.