Reflection Tomography
Reflection tomography is a seismic velocity analysis technique that uses the travel times of reflected seismic waves recorded at multiple offsets (source-receiver distances) to iteratively update a subsurface velocity model, with the objective of building an accurate three-dimensional velocity field that enables correct imaging of complex geological structures where the simple layered-earth velocity assumptions of conventional analysis break down; the technique adapts the mathematical framework of medical computed tomography (CT scanning) — which uses multiple X-ray beams crossing the body from different angles to reconstruct internal density images — to the seismic reflection problem, where reflected energy arriving at a spread of receivers provides the equivalent of different angular ray paths through the subsurface velocity structure; reflection tomography is particularly valuable for imaging beneath salt bodies, thrust belts, and other structures with steep velocity contrasts, where incorrect velocity models produce migrated seismic images that are defocused, incorrectly positioned, or structurally distorted to the point of misleading interpretation; the input to reflection tomography is typically residual moveout measurements on common image point (CIP) gathers in the depth domain — the residual moveout (apparent curvature remaining on the reflection events after migration with the current velocity model) encodes the error in the velocity model, and the tomographic inversion updates the model to minimize this residual moveout across all gathers simultaneously.
Key Takeaways
- The fundamental challenge that reflection tomography addresses is the nonuniqueness of seismic velocity model building: many different velocity models can produce the same observed travel times from a set of surface receivers, so additional constraints are required to select the geologically reasonable model from the infinite space of mathematically valid solutions; reflection tomography addresses this by minimizing residual moveout across all offsets simultaneously (requiring the velocity model to make reflected events flat in CIP gathers at all offsets) and by incorporating structural constraints (the velocity model must be geologically plausible, with smooth spatial variations except across known boundaries such as salt flanks, fault planes, and unconformities where velocity can jump discontinuously); well velocity control (sonic logs, checkshot surveys, VSP data) provides additional hard constraints on the velocity at specific locations, anchoring the tomographic solution to ground truth and reducing the nonuniqueness in the vicinity of wells; despite these constraints, multiple geologically plausible models may still fit the data, making reflection tomography a tool for reducing rather than eliminating velocity uncertainty.
- Salt-related velocity model building is the most demanding application of reflection tomography and the primary driver of its development in the deepwater Gulf of Mexico, where thick Jurassic salt bodies with velocities of 4,480 meters per second create extreme contrasts with the surrounding sediments (1,800-2,500 meters per second); rays passing through or beneath salt are severely bent by the impedance contrast at the salt boundary (salt flanks and overhangs), and the ray paths predicted by an incorrect salt geometry model are wrong in direction and length, producing migrated images that are severely defocused and structurally incorrect in the subsalt region; the workflow for salt velocity model building alternates between reflection tomography (to optimize the velocity in the sediment above and below salt) and salt geometry interpretation (to redefine the salt body boundary based on the improved image), iterating until both the velocity model and the image are mutually consistent and geologically reasonable; this iterative workflow requires close collaboration between geophysicists and geologists, and can take months to years for complex salt provinces.
- Full waveform inversion (FWI) is a more powerful but computationally demanding alternative to reflection tomography for velocity model building that uses the full information content of the seismic wavefield (amplitudes, phases, and waveforms, not just travel times) to update the velocity model; reflection tomography uses only the kinematic information in the data (arrival times and moveout), while FWI uses both kinematics and dynamics (wavelet shape and amplitude), giving FWI more sensitivity to fine-scale velocity variations and more potential resolution in the updated model; however, FWI requires accurate knowledge of the seismic source wavelet, low-frequency data content (below 4-5 Hz) to converge to the correct solution rather than a local minimum, and extremely intensive computation (an FWI on a large 3D dataset requires millions of forward modeling runs on high-performance computing clusters); reflection tomography remains the practical workhorse velocity model building method for most production seismic processing because its computational cost is orders of magnitude lower than FWI, even though FWI produces higher-resolution velocity models when the data quality and acquisition design support it.
- Anisotropic reflection tomography extends the standard isotropic approach to account for the directional dependence of seismic velocity that is present in shales and other layered materials where waves travel faster horizontally than vertically (vertical transverse isotropy, VTI) or faster in one horizontal direction than another (horizontal transverse isotropy, HTI, caused by aligned fractures or stress); ignoring anisotropy in a region with thick shale overburden causes the tomographic velocity model to be systematically wrong in ways that distort the depth and geometry of underlying structures, because the upgoing reflections from deep targets have traveled through anisotropic shale at velocities that depend on their propagation direction and this directional dependence is not captured in an isotropic model; modern reflection tomography implementations update multiple anisotropy parameters simultaneously with the isotropic velocity, significantly improving the accuracy of depth imaging in areas with significant overburden anisotropy; the additional model parameters require additional data types (wide-azimuth or multi-azimuth seismic acquisition) and add to the nonuniqueness of the inversion, making anisotropic tomography more complex but necessary in tectonically compacted basins with significant shale volume above the reservoir.
- Reflection tomography for time-lapse (4D) seismic involves comparing velocity models derived from repeat surveys to detect changes in the subsurface velocity field caused by fluid movement during production — water displacing oil reduces pore pressure and increases bulk modulus, changing the seismic velocity by amounts that are detectable in time-lapse velocity analysis; 4D tomography must separate the genuine velocity changes caused by fluid movement from changes in the measured velocity caused by differences in the acquisition geometry, source characteristics, and noise between the repeat surveys; achieving this requires consistent processing of both surveys (using the same tomographic parameters), careful normalization of the velocity models (to remove acquisition-related differences), and statistical testing to determine whether the observed velocity changes are larger than the noise floor of the difference; successful 4D velocity tomography has been demonstrated at several North Sea fields where the velocity changes are large enough (North Sea chalk fields with large velocity sensitivity to water saturation changes) to be reliably detected with available technology, providing direct subsurface information about sweep efficiency and bypassed oil that guides infill drilling decisions.
Fast Facts
The mathematical connection between seismic tomography and medical CT scanning is more than analogy: both techniques solve variants of the Radon transform problem, which was first analyzed mathematically by Johann Radon in 1917, more than 50 years before CT scanning was invented (Hounsfield and Cormack received the 1979 Nobel Prize in Physiology or Medicine for CT scanning). The seismic tomography inversion problem is actually harder than the medical CT problem because seismic rays bend through the subsurface (their paths depend on the velocity model being solved for), while X-rays travel in straight lines through tissue, making medical CT a linear problem and seismic tomography a nonlinear one requiring iterative solution.
What Is Reflection Tomography?
Reflection tomography is how seismic processors build the velocity model that makes depth imaging work beneath complex structures. The seismic image of a reservoir is only as accurate as the velocity model used to migrate the data, and in areas with salt, thrust faults, or steep velocity contrasts, the simple velocity models that work in sedimentary basins produce images that are blurry, structurally distorted, and potentially misleading about where the reservoir actually is. Reflection tomography uses the fact that seismic waves recorded at different offsets have traveled along different ray paths through the subsurface, and any errors in the velocity model cause those different paths to produce inconsistent images of the same reflector. By systematically measuring those inconsistencies and updating the velocity model to eliminate them, tomography converges on a velocity field that produces a focused, geometrically accurate depth image — and that is what the exploration team uses to locate the drill target.
Synonyms and Related Terminology
Reflection tomography is also called seismic tomography or tomographic velocity updating. Related terms include velocity model building (the general workflow of constructing the three-dimensional velocity field needed for depth migration of seismic data, of which reflection tomography is the primary iterative tool), residual moveout (the apparent curvature remaining on reflection events in common image point gathers after migration with an imperfect velocity model, which is the primary input to reflection tomography), full waveform inversion (FWI, the more computationally intensive velocity model building method that uses amplitude and phase information rather than just travel times, providing higher resolution than reflection tomography in favorable conditions), depth migration (the seismic processing algorithm that repositions reflected events to their correct subsurface locations using the velocity model, requiring accurate velocity to produce focused structural images), and salt imaging (the specific seismic imaging challenge in areas with halite bodies, where the extreme velocity contrast and irregular geometry require iterative reflection tomography to build an accurate subsalt velocity model).
Why Velocity Model Quality Determines Whether a Drillable Target Exists at All
In simple geology, a rough velocity model is adequate: the structural highs on a reflector look like highs regardless of small velocity errors, and the well goes to the right place. In complex geology — below salt, in thrust belts, beneath carbonate platforms — velocity errors of 5-10% in the overburden can move a migrated reflector by 300-500 meters laterally or vertically. At those scales, a reflection tomography iteration that eliminates the velocity error can reveal a structural closure that was previously smeared into the background, or can remove a false closure that was an artifact of velocity-related misimaging. The drilling decision changes based on what the tomography produces. This is why deepwater exploration companies invest millions of dollars in specialized tomographic processing before committing to a deepwater well that may cost $100-200 million to drill: the cost of getting the image right before drilling is trivial compared to the cost of drilling into a structure that the imaging said was there but wasn't.