Exploring factors affecting the performance of deep learning in seismic fault attribute computation

Abstract

Recently, the computation of seismic fault attribute that may be significant in seismic interpretation is that seismic fault detection is treated as an image segmentation problem using different deep-learning (DL) architectures. To do this, researchers have concentrated on applying cutting-edge DL architectures in computing seismic fault attributes. To explore the factors that may affect the accuracy of seismic fault attribute, we compare the computed fault probability using DL architectures under different scenarios. The designed scenarios aim to highlight the leading factors that may affect the accuracy and resolution of seismic image segmentation. The discussed factors include the dimension and size of training data, training data preparation, ensemble learning, and batch size in DL. The proposed comparisons are applied to one marine seismic survey from New Zealand and one land seismic survey from China. The results demonstrate that properly preparing training data is far more important than choosing a cutting-edge DL architecture in computing seismic fault attribute. We also propose a practical workflow that can include real seismic data and corresponding interpreted fault sticks in training data for a specific seismic survey.