Cluster analysis of velocity models around the Hudson Bay region, Eastern Canada

SUMMARY Understanding deep crustal structure can provide us with insights into tectonic processes and how they affect the geological record. The be studied using a suite of seismological techniques such as receiver function analysis, body surface wave tomography. Using models derived from these methods, it is possible to delineate boundaries regions that may have been affected by similar processes. However, often velocity are grouped in somewhat subjective manner, potentially meaning some insight missed. Cluster based on unsupervised machine learning, used more objectively group profiles and, thus, put additional constraints structure. In this study, we apply hierarchical agglomerative clustering shear obtained previous studies focused region joint inversion functions dispersion data at 59 sites surrounding Hudson Bay. This location provides an ideal natural laboratory study Precambrian processes, including 1.8Ga Trans-Hudson Orogen. We use Ward linkage define distance between clusters, gives most physically realistic results, after testing number clusters 2 10, find there 5 main stable models. then compare our results different parameters, schemes (K-means GMM), well for azimuths that, overall, consistent. form correlate geology, thickness, regional tectonics geophysical concentrated specific regions. Archean domains (Rae, Hearne Superior) clearly distinguished influenced Proterozoic orogenic events (Southern Baffin Island Ungava Peninsula). Further, crust Melville Peninsula found same cluster western coast Peninsula, suggesting Our shows promising learning interpreting structures gain new insights.