Multivariate Supervised Classification, Application to a New Zealand Offshore Field
Hami-Eddine Kamal, Pascal Klein, Loic Richard, Dorra Elabed, Eric Chatila, Andrew Furniss, 2011. "Multivariate Supervised Classification, Application to a New Zealand Offshore Field", Attributes: New Views on Seismic Imaging–Their Use in Exploration and Production, Kurt J. Marfurt, Dengliang Gao, Art Barnes, Satinder Chopra, Antonio Corrao, Bruce Hart, Huw James, Jory Pacht, Norman C. Rosen
Download citation file:
This paper presents a new lithofacies prediction method using seismic prestack data and well lithofacies. The Democratic Neural Networks Association (DNNA) is used to perform the prediction. The association of neural networks with a Bayes approach helps to increase the robustness of the prediction and to quantify the uncertainty. The primary aim of this process is not only to qualify reservoir heterogeneity and lateral extension but also predict lithology where no well has yet been drilled.
Lithofacies prediction from surface seismic data has been applied to the technically challenging Cretaceous rift of the Taranaki Basin where exploration objectives are focused primarily on the Miocene and the Pliocene turbidite sandstones. The prediction using DNNA is compared with two methods commonly used in the oil and gas industry.