Interpretation of gas chimney from seismic data using artificial neural network: A study from Maari 3D prospect in the Taranaki basin, New Zealand

Abstract

The seismic interpretation in Maari 3D prospect of the Taranaki basin in New Zealand based on artificial neural network has brought out gas migration pathways from the source rock through the faulted reservoirs to the seabed. The findings correlate reasonably with the results of Moki-1 well drilled within the study region. The gas chimneys are analyzed using modern 3D visualization tools by displaying the chimney probability cube over different vertical seismic sections, horizon slices and time slices respectively. The training of multi-seismic attributes resulted into 0.4–0.6 normalized RMS error giving rise to 5.08–10.26% misclassification during the training and testing phases. Several fault intersection zones (weak zones) within the reservoirs exhibit high probability of gas chimneys. This study acts as an add-on-tool for understanding the petroleum system and provides preventive clues for mitigating hazards in future exploitation program.