In recent years, marine controlled-source electromagnetic (CSEM) surveying has become an effective supplemental interpretation tool to the seismic reflection method to help mitigate risk in an offshore exploration setting. Interpretation of marine CSEM data is commonly achieved via finite-difference inversions on rectilinear meshes, which has its merits, but the results are typically of very low resolution. The alternative is forward modeling, which requires a model to be known a priori, but the detail of the model can be created to reflect realistic geologic conditions. What is typically seen in the literature are applications of EM forward modeling codes to synthetic, and sometimes complex synthetic, models. However, what the literature is missing is an application that overcomes the challenges of applying a 3D forward modeling method to real models constructed from real information. We have developed an application of a 3D marine CSEM finite-element forward modeling method to the Bay du Nord prospect in the Flemish Pass Basin offshore Newfoundland. The 3D resistivity model, composed of four topographical layers and the Bay du Nord reservoir body, was built using 2D seismic data, one well log, and a marine CSEM inversion. Although other mesh representations have their merits, we chose to discretize our 3D model into an unstructured tetrahedral mesh because its flexibility enabled the accurate representation of complex structures while minimizing the number of unknowns. The availability of measured marine CSEM data allowed for the resistivities of each layer in the 3D model to be refined, and it also allowed for the simulated data to be assessed in the context of the real noise levels. A subsequent sensitivity analysis of the forward modeling results provided insights regarding the detectability of the Bay du Nord reservoir.

No abstract

No abstract

No abstract

Machine-learning techniques allow geoscientists to extract meaningful information from data in an automated fashion, and they are also an efficient alternative to traditional manual interpretation methods. Many geophysical problems have an abundance of unlabeled data and a paucity of labeled data, and the lithology classification of wireline data reflects this situation. Training supervised algorithms on small labeled data sets can lead to overtraining, and subsequent predictions for the numerous unlabeled data may be unstable. However, semisupervised algorithms are designed for classification problems with limited amounts of labeled data, and they are theoretically able to achieve better accuracies than supervised algorithms in these situations. We explore this hypothesis by applying two semisupervised techniques, label propagation (LP) and self-training, to a well-log data set and compare their performance to three popular supervised algorithms. LP is an established method, but our self-training method is a unique adaptation of existing implementations. The well-log data were made public through an SEG competition held in 2016. We simulate a semisupervised scenario with these data by assuming that only one of the 10 wells has labels (i.e., core samples), and our objective is to predict the labels for the remaining nine wells. We generate results from these data in two stages. The first stage is applying all the algorithms in question to the data as is (i.e., the global data), and the results from this motivate the second stage, which is applying all algorithms to the data when they are decomposed into two separate data sets. Overall, our findings suggest that LP does not outperform the supervised methods, but our self-training method coupled with LP can outperform the supervised methods by a notable margin if the assumptions of LP are met.