Unsupervised machine learning and multi-seismic attributes for fault and fracture network interpretation in the Kerry Field, Taranaki Basin, New Zealand

Abstract Unsupervised machine learning using an unsupervised vector quantization neural network (UVQ-NN) integrated with meta-geometrical attributes as a novel computation process opposed to traditional methodologies is currently used effectively in the 3D seismic structural interpretation for high-resolution detection of fault patterns, fracture zones, and small-scale faults (SSFs). This technology has crucial role locating prospective well sites building model while saving time cost. The innovation current workflow involves combining geostatistical filtering, optimal geometrical attributes, UVQ-NN automatic major faults, SSFs volumes extraction due unavailability logs cores. To sharpen edges discontinuities, steered volume was first extracted. Structural filters were then applied volume, dip-steered median filter (DSMF), followed by diffusion (DSDF), finally, both DSMF DSDF combined generate enhancement (FEF). After that, computed extracted, such similarity, FEF on maximum curvature, polar dip, density, thinned likelihood (TFL) attributes. Finally, selected inserted input layer segmentation matching volumes. On other hand, TFL voxel connectivity (VCF) patches extraction. results from VCF identified locations, orientations, extensions main SSFs, networks. implemented approach innovative can be employed future identification, extraction, classification geological networks any region world.