Skip to the content

Menu

Sparse Adaptive Waveform Inversion for Rapid Sub-Basalt Prospect Identification

Identifying hydrocarbon prospects beneath basaltic layers has historically posed significant challenges due to the complex seismic imaging required. Traditional full-waveform inversion (FWI) techniques often struggle with issues such as cycle skipping, especially when starting models are less accurate or when higher starting frequencies are used. Adaptive waveform inversion (AWI), a reformulation of the misfit function used in FWI, offers a promising alternative. This case study examines the application of sparse AWI for rapid sub-basalt prospect identification, demonstrating its efficacy in creating accurate velocity models from minimal data.

The dataset used in this study originates from a 2500 km² narrow-azimuth towed-streamer (NATS) survey conducted in 2011 over the Skoll High area of the Vøring Marginal High in the Central Norwegian Sea. The survey area features rugged extrusive basaltic flows and deeper layered intrusive sills, with no existing wells. The 3D seismic data were acquired using two 4980 cu.in. flip-flop sources at a 10-meter depth and ten 8000-meter Broadseis slanted cables, spaced 100 meters apart, towed at varying depths (8 meters at the front and 50 meters at the back). This configuration allowed for the acquisition of both low and high-frequency data, crucial for effective sub-basalt imaging.

The sparse AWI approach employed in this study was designed to minimize computational effort while maximizing the utility of the available data. Key methodological steps included:

  • Data Preprocessing: Raw field data were used, retaining only a third of the original sources and low-pass filtering the data at 4 Hz.
  • Sparse Sampling: Only ten iterations of waveform inversion were performed, using just a fifth of the retained sources at each iteration, effectively utilizing only a fifteenth of the originally acquired sources.
  • Model Initialization: AWI began from a smooth velocity model with an accurate one-dimensional velocity structure in the water column and a well-defined seabed.
  • Inversion Process: The inversion allowed variations in the water column, seabed position, and seabed density contrast. Below the seabed, Gardner's law was used to determine density, and a simple smooth regional model of VTI anisotropy was applied.

The results demonstrated significant improvements in the velocity model and reflectivity images:

  • Velocity Model: The AWI-generated velocity model revealed a high-velocity basalt layer crossing the entire section, which was consistent with the uppermost bright rugged reflector in the pre-stack depth migration (PSDM) volume. This layer, representing basaltic flows, was generated directly from the data, unlike the interpretively inserted layer in the legacy model.
  • Reflectivity: The AWI reflectivity model showed more continuous and less noisy reflectors compared to the legacy PSDM, indicating improved accuracy in the depth section. Notably, the AWI model identified high-porosity clastic layers between and below the basaltic layers, which are potential hydrocarbon targets.
  • Data Match: The comparison of field data with synthetic data generated by the AWI model showed a significantly improved match, even at the low frequency of 4 Hz. Further iterations at higher frequencies are expected to enhance the data match and model resolution.

The sparse AWI approach demonstrated in this study offers several advantages:

  • Cost-Effectiveness: By using a minimal subset of the available data and iterations, the computational cost is significantly reduced, making it suitable for rapid re-evaluation of large legacy datasets.
  • Robustness: AWI proved to be robust in the presence of cycle skipping, successfully converging to accurate models even with less accurate starting models and higher starting frequencies.
  • Direct Interpretation: The velocity model generated by AWI provided direct insights into sub-basalt structures, aiding in the identification of high-porosity clastic targets without relying heavily on interpretive inputs.

Sparse adaptive waveform inversion (AWI) has shown to be a powerful tool for rapid sub-basalt prospect identification. Its ability to generate accurate velocity models from sparse data and minimal iterations makes it an attractive option for re-evaluating large legacy seismic datasets in basaltic provinces. The approach not only reduces computational costs but also enhances the accuracy and reliability of sub-basalt imaging, thereby facilitating more effective hydrocarbon exploration.

 

Acknowledgements:

The authors thank Wintershall Dea AG and S-Cube for permission to publish this work.

 

References:

Warner, M., & Guasch, L. (2014, 2016). Adaptive waveform inversion.

Guasch, L., Warner, M., & Ravaut, C. (2019).

Adaptive waveform inversion: Practice. Geophysics, 84, R447-R461.

 

Keywords: 

Full Waveform Inversion, FWI, Adaptive Waveform Inversion, AWI, Cycle Skipping, Sub-basalt, basalt, seismic imaging, FWI case study, narrow-azimuth towed-streamer, Norwegian Sea