Interpretation of magma transport through saucer sills in shallow sedimentary strata using an automated machine learning approach

Abstract

In sedimentary basins, emplacement of sheet intrusions such as sill complexes significantly contributes to the transport and storage of hot magma at shallow level. Such emplacement at shallow depth leads in doming of the overburden, which acts as plausible structural traps useful for hydrocarbon exploration. The petroliferous Canterbury Basin, SE offshore New Zealand, is a classic example of such phenomena, where saucer-shaped magmatic sills are emplaced within the Cretaceous to Eocene succession. This has resulted into forced folds and hydrothermal vents above the sill terminations within the Eocene sequences. The present study attempts to capture this scenario through a neural network by designing meta-attributes, called as the Sill Cube (SC) and Fluid Cube (FC). The meta-attributes are computed by unifying different seismic attributes that are trained over interpreter's knowledge on the geologic targets following a supervised scheme of neural learning. The approach prominently arrests the structural geometry of sill complexes and fluxed-out magmatic fluids within the Cretaceous to Eocene strata from 3D seismic reflection data of the Waka prospect, offshore Canterbury Basin. Based on the meta-attribute interpretation, the individual sills within the Waka prospect cover areas of ~1.5 km2 to 17 km2, where the principal sills namely the Sill W1 and Sill W2 spread over an area of ~12 km2 and 17 km2, respectively. Moreover, the fluxed out magmatic fluids vertically rise to a height of ~800 m through hydrothermal vents from the tip of the principal sills, and uplift the overburden. Such approach is automated and incorporates interpreter's acquaintances to effectively capture subsurface magmatic activities for better interpretation of 3D seismic data.