Convolutional Neural Network for Prediction of Igneous Seismic Facies in the Santos Basin Pre-Salt

Abstract

Summary Automatic seismic interpretation is one of the main applications of machine learning for exploration geophysics. In recent years, an increase in the popularity of convolutional neural networks to perform interpretation related tasks has been observed. Here we develop and train a 3D Convolutional Neural Network for predicting igneous seismic facies in the Santos Basin Pre-Salt. Often one of the main challenges regarding automatic seismic interpretation using neural networks relates to the scarcity or lack of training data. To overcome this problem, we used a sampling strategy to train the network with just a few partially interpreted sections. After the training process, the network was applied in the full seismic amplitude volume, outputting the igneous facies probability for the whole area. The results show good conformity with the input manual interpretation and with well logs, which was not part of the training.