Post-Migration Data Conditioning
Kurt J. Marfurt
The University of Oklahoma
Proprietary company surveys are usually acquired and processed to optimize the data quality for one or more specific objectives with fixed budget constraints. Later in the life of the survey, new play concepts may arise, or there may be a need to better image shallow drilling hazards or deeper targets for waste-water disposal. In contrast to a proprietary survey, service company “speculative surveys” are usually acquired and processed to provide the best over-all image. In general, the service company does know the detailed play concepts of a given customer; indeed, their goal is to sell the same speculative survey to as many customers as possible. Finally, many new exploration plays are covered by older seismic data that were acquired to image what are now highly mature fields. These data may have been obtained through asset acquisition, data trade with a partner, or through the use of 3rd party data brokers. Given all these factors, there are usually a few processes the interpreter can evaluate in an effort to improve resolution and reduce ambiguity.
Like medical doctors following the Hippocratic Oath, seismic processors are trained to “do no harm” to the seismic data. They use all of their insight, skills, and the technology of the day to produce as good an image as possible without introducing features or artifacts that may somehow bias the interpretation of the data. In contrast to the seismic processor, the interpreter’s job is to extract the most information possible from the data. The interpreter often has access to well data, knows the depositional environment, and either knows, or wishes to evaluate a suite of potential play concepts. Although the interpreter does not have access to all of the high-tech tools of the seismic processor such as tomographic velocity analysis, prestack migration, and 5D interpolation, by having access to geologic models and additional data, the interpreter can be more aggressive with the relatively limited tools at his or her disposal. Whereas overly aggressive filtering by the seismic processor might be considered as “biasing”, “massaging”, or “hammering” the data, the same filters applied by an interpreter can be quality controlled through their understanding of geologic processes, modern and ancient analogues, and their access to well control.
Post-migration data conditioning includes phase rotation, amplitude balancing, band-pass filtering, spectral balancing, structure-oriented filtering, and footprint suppression. In this presentation I will review current best practices in spectral balancing, summarizing the assumptions and limitations of different techniques and their impact on seismic resolution, attribute response, and impedance inversion.
Biography
Kurt Marfurt has divided his career nearly equally between industry and academia working at Amoco in Tulsa, and teaching at Columbia University, The University of Houston, and The University of Oklahoma, where he is currently an emeritus professor of geophysics. For the past 28 years he has focused on seismic attributes and machine learning to aid the seismic interpreter. Marfurt has served as editor-in-chief for the SEG/AAPG journal Interpretation, has delivered two SEG distinguished instructor short courses, was the 2021-22 AAPG/SEG distinguished lecturer, and was honored with AAPG’s Robert R Berg award for research in 2019 and SEG’s Maurice Ewing medal in 2023. Satinder Chopra and Marfurt are coauthors of the SEG October 2024 book Essentials of Seismic Attributes and Impedance Inversion.
