Paper Summary
Marine seismic data sets contain highly redundant information. Data analytics and machine learning-based solutions should provide opportunities to reduce turnaround and improve confidence levels in output data volumes. A proof-of-concept (POC) thrust regime example from Indonesia illustrates that parameter testing can almost be eliminated if existing project parameter data can be mined from a database. Where quality control (QC) is required for complex challenges such as noise removal, supervised classifiers are a platform that can enable rapid global quantitative decisions based on relevant data attributes, moving behind the subjective art of observational QC. Finally, many early processing steps depend on reasonable knowledge of the velocity model in addition to the explicit dependence of imaging steps. A POC Monte Carlo-based model building exercise in West Africa used an efficient tomographic platform to demonstrate that turnaround can be reduced from 90 days to only a few days, even when the starting model was significantly wrong. These examples illustrate that a lot is already within our reach, and the development of embedded feedback loops will improve the level of automation further, particularly if humans can learn to let the data speak for itself.