OBJECTIVE LINKING OF DIKE SEGMENTS TO EVALUATE MESOSCALE STRUCTURE OF DIKE SWARMS

Dike swarms represent a direct window into the frozen structure of magmatic systems. Some of the largest dike swarms on Earth are associated with Large Igneous Provinces, but disentangling the dynamics of emplacement to infer crustal storage zones and time-evolving crustal stress states is challenging. Exposed dikes are often segmented by erosion, super-imposed and segmented, making assessment of primary structures challenging. We develop a new framework to deconvolve signatures of magmatic and tectonic signals in dike orientation and distribution. Using mapped dike segments, we apply a coordinate transform commonly used in computer vision called the Hough Transform, which transforms individual lines into a 2D angle, perpendicular distance space. Combining the Hough Transform with hierarchical clustering, we are able to link dike segments across large Cartesian domains to reconstruct primary dike structures. We apply this method first to a small continental dike swarm, the 24.53-23.36 Ma Spanish Peaks Swarm in southern Colorado, USA. This provides a proof of concept that permits application to the much larger Chief Joseph Dike Swarm associated with the 16.7-15.9 Ma Columbia River flood basalts and the Nasik-Pune/Central Deccan Swarm associated with the 67-65 Ma Deccan Traps. These dike swarms are linked to some of the largest eruptive formations in their respective flood basalt stratigraphies but their structure is extremely complex with multiple overlapping trends and thousands of individual segments. By evaluating the structure in Hough space we are able to pull out multiple radial and linear dike swarms from the overlapping structures. We find that the average length of linked dikes reaches ~25 km in the CJDS and ~35 km in the Coastal Deccan swarm with the longest dikes reaching over 250 km. For the majority of dikes, less than 25% of the total length is exposed at the surface and mapped. Within the CJDS there are two identifiable radial trends however we did not identify any radial structure within the Central Deccan swarm. Our method links the structure of the dikes and identifies subswarms allowing for further interpretation of the magmatic storage systems.