First Break - Volume 43, Issue 9, 2025

The authors analysed the regional distribution of Rogaland Group sands, which were predicted using AI-based Rune Inversion and the Gyllenhammar equation for clay volume estimation applied to the ultra-large Elephant 2.0 database (constructed in 2022). The results were compared with the stratigraphic framework established by Brunstad et al. (2013). Drawing upon publicly available information and insights from personal comments made by Harald Brunstad, the study demonstrates general compliance with the work of Brunstad et al. and reveals potential new areas and thicknesses for sandstone accumulations. Considering the challenges in estimating volume of sand from post-stack inversion, the results show a high degree of correlation, revealing more details between the wells.

While crucial to understand the subsurface, interpreting 3D depositional elements from seismic data is often challenged by issues such as effectively visualising subtle stratigraphic features and accurately transforming them into 3D objects that reflect geological reality. This paper introduces an interactive framework that directly addresses both through an innovative three-step workflow. The process begins with chronostratigraphic colour blending on stratal slices, enabling rapid visualisation of depositional patterns and sequence relationships. Next, to enhance feature separation, interpreters interactively place seed points within target elements and background facies, allowing for the automatic determination of optimal spectral decomposition frequencies. Finally, the interpreter guides an image segmentation foundation model with simple point prompts, facilitating the precise extraction of depositional elements without the bottleneck of geological-specific training data. Application of the method to fields, namely channel systems from the Maui Field, New Zealand, and carbonate build-ups and reefs from the Poseidon Field, Australia demonstrates a reduction in interpretation time from weeks to just days, all while maintaining high accuracy. This chronostratigraphic-centric approach guarantees that extracted objects represent true depositional architecture rather than arbitrary seismic anomalies.

Marine seismic data often lacks low-frequency content, limiting its usefulness for advanced modelling and inversion techniques such as Full Waveform Inversion (FWI). In this study, we applied a self-supervised deep learning (SSL) framework for low-frequency extrapolation, tested on field data from the Asri Basin, Indonesia. Our approach uses a modified U-Net architecture and a two-stage learning scheme, namely: synthetic warm-up followed by iterative data refinement (IDR), to train directly on unlabelled band-limited seismic data. When used as input for a source-independent FWI (SI-FWI) method, the low-frequency extrapolated data significantly improves inversion results, demonstrating better recovery of deep velocity structures and reduced cycle skipping compared to inversions using band-limited data as inputs. The integrated SSL and SI-FWI workflow provides practical value for reprocessing legacy datasets lacking both low-frequency energy and source wavelet information. This method offers a data-driven, label-free approach for expanding the spectral bandwidth of field seismic data and further improving inversion convergence.