Processing (Seismic)

Key Takeaways

• The velocity model is the most critical input to seismic processing and the most difficult to determine accurately — seismic migration algorithms reposition reflectors to their true subsurface locations using the subsurface velocity field; if the velocity model is wrong, reflectors end up in the wrong positions in the migrated image, leading to drilling targets that miss the actual reservoir; velocity model building typically involves iterative tomography (using the curvature of reflections across different offsets to estimate velocities), full waveform inversion (using the full waveform content of the data rather than just travel times), and well calibration (using checkshot and VSP data from existing wells to anchor the velocity model to known depth-velocity relationships).
• Multiple reflections are a major processing challenge in marine seismic data — multiples are seismic energy that has reflected more than once before reaching the receivers, arriving at the same time as primary reflections from deeper targets and obscuring them; the most problematic are water bottom multiples (energy that bounces between the sea floor and the water surface repeatedly) and interbed multiples (reflections between pairs of subsurface interfaces); multiple attenuation methods include surface-related multiple elimination (SRME, a data-driven method that uses the data itself to predict and subtract multiples), parabolic Radon demultiple (which separates primaries from multiples using their different velocity-offset moveout), and high-resolution Radon methods for interbed multiples.
• Prestack depth migration (PSDM) has become standard for structurally complex areas — conventional post-stack time migration handles moderate structural complexity adequately, but subsalt plays (Gulf of Mexico, North Sea), thrust belts (Rocky Mountain, Zagros), and areas with strong lateral velocity variation require prestack depth migration to accurately image subsurface structure; PSDM uses the full velocity model in depth coordinates and migrates individual seismic traces before stacking, allowing the algorithm to account for complex ray paths that would be distorted in time-domain processing; the transition of the Gulf of Mexico deepwater industry from time migration to depth migration in the 1990s and 2000s was a primary driver of improved subsalt imaging that enabled the major deepwater discoveries of that era.
• Full waveform inversion (FWI) represents the current frontier of seismic processing for velocity model building — FWI uses the entire seismic waveform (not just travel times and amplitudes of first arrivals) to iteratively update a velocity model by minimizing the difference between synthetic seismograms computed from the model and the actual recorded field data; FWI provides dramatically higher resolution velocity models than conventional tomography, resolving features at scales of tens of meters rather than hundreds of meters; FWI is computationally intensive (typically requiring thousands of CPU-core-hours for a production survey) and requires high-quality data with rich low-frequency content, which is why modern acquisition systems and broad-band seismic equipment have been developed specifically to provide the low-frequency content that makes FWI work effectively.
• Processing quality control (QC) at each stage is as important as the processing algorithms themselves — every step in the processing sequence can introduce artifacts if parameters are chosen poorly; experienced processors apply systematic QC at each step, checking that noise has been attenuated without also attenuating signal, that velocities are consistent with geological knowledge, that migration has correctly positioned dipping reflectors, and that the final stack volumes are free from acquisition footprint and processing artifacts; the iterative relationship between processor and interpreter — where interpretation insights feed back into processing parameter adjustments — is recognized as a best practice for producing images that support confident geological interpretation.