Smile: Seismic Migration Artifacts, Noise Migration, and Velocity Model Errors
A smile is a concave-upward, roughly semicircular event seen on migrated seismic data that, as the name suggests, has the upturned shape of a smile and is an artifact rather than a real geologic reflector. It is one of the most recognizable signatures of imperfect seismic migration, the processing step that repositions reflection energy from where it was recorded to where it actually originated in the subsurface. Migration works by collapsing diffractions and moving dipping events to their true positions using a velocity model; when that process is fed bad input, isolated noise, or an incorrect velocity, the migration operator smears the offending energy along its impulse response, which for most algorithms is a downward-opening arc. The result is a string of concave-up smiles, also called migration smiles or simply smiles, that can clutter the image and be mistaken for structure by an inexperienced interpreter. Two distinct causes produce them. The first is overmigration, using a migration velocity that is too high, which overcorrects events and bends them upward into smiles, the mirror-image problem of undermigration (too low a velocity), which leaves residual downward-curving frowns and incompletely collapsed diffractions. The second is the migration of noise: a spike of incoherent energy, a bad trace, a spatial-aliasing artifact, or an isolated burst of ground roll that has no true reflector behind it, gets spread by the operator into a perfect smile because the algorithm assumes every input sample is real reflection energy to be repositioned. Smiles are especially common at the edges of a survey, at the bottom of the section where the velocity model is poorly constrained, and in steeply dipping or structurally complex settings such as the deformed Foothills belt along the western Western Canadian Sedimentary Basin. Recognizing a smile for what it is, an imaging artifact tied to velocity error or unmigrated noise, is a core skill in quality-controlling a seismic migration, refining the velocity model, and avoiding the costly mistake of mapping a processing artifact as a drillable structural trap.
Key Takeaways
- An artifact, not a reflector: A smile is a concave-upward, semicircular event created during migration, not a real geologic boundary. It appears when the migration operator spreads energy, either noise with no true source or correctly-recorded energy moved with the wrong velocity, along its arc-shaped impulse response, leaving a false event that mimics structure.
- Overmigration produces smiles: Using a migration velocity that is too high overcorrects events and curves them upward into smiles. The opposite error, velocity too low (undermigration), leaves downward-curving frowns and uncollapsed diffractions. Interpreters use the appearance of smiles versus frowns as a direct diagnostic of whether the velocity model is too fast or too slow.
- Noise migration is the other cause: Isolated noise, a dead or spiking trace, spatial aliasing, or a burst of ground roll, has no reflector behind it, so migration spreads it into a textbook smile. This is why pre-migration noise attenuation and trace editing are critical: garbage energy in becomes arc-shaped artifacts out.
- Worst at edges and depth: Smiles cluster at survey edges (migration aperture limits), at the bottom of the section where velocity control is weakest, and in steep-dip settings like the WCSB Foothills thrust belt. These are exactly the zones where the velocity model is least constrained and where an interpreter must be most skeptical of upturned events.
- Real interpretation risk: A smile can be misread as an anticline crest or a fault-bounded closure, the kind of structure that gets a well proposed. Mistaking a migration artifact for a drillable trap can sink millions of CAD into a dry hole, so smile recognition is a frontline part of seismic interpretation quality control before any prospect is matured.
Smiles Versus Frowns as a Velocity Diagnostic
The direction an artifact curves is a practical readout of velocity error. When the migration velocity is too high, events are overmigrated and curl upward into smiles; when it is too low, they are undermigrated and sag into downward-opening frowns with residual, uncollapsed diffraction tails. Processors exploit this during migration velocity analysis: they migrate the data with a range of velocities and watch whether diffractions collapse cleanly (correct velocity), arch into smiles (too fast), or remain as frowns (too slow). Iterating the velocity field until diffractions just collapse, with neither smiles nor frowns, is how a balanced migration velocity is tuned across a WCSB survey before the final image is delivered to interpreters.
Suppressing Smiles in Processing
Two main strategies reduce smiles. The first is aggressive pre-migration conditioning: trace editing to kill dead and spiking traces, ground-roll and noise attenuation, and anti-alias filtering so that little incoherent energy survives to be smeared into arcs. The second is a better-constrained velocity model and a properly chosen migration aperture; limiting the operator aperture and tapering survey edges curbs the edge smiles that arise where the migration has too little data to image correctly. In complex Foothills imaging, prestack depth migration with an iteratively refined interval-velocity model, often guided by well ties and tomography, is the standard route to an image with diffractions collapsed and smiles suppressed.
Fast Facts
The smile is, in effect, the visible impulse response of the migration operator drawn on the data. Every migration algorithm answers the question of where a single input sample could have come from, and for a constant-velocity medium the answer is a semicircle (in time) or an ellipse (in depth) of equally probable source positions. When real, coherent reflections are present, thousands of these arcs interfere constructively along the true reflector and destructively elsewhere, building a clean image; but feed the operator a lone spike of noise and there is nothing to cancel its arc, so the full semicircular smile is laid bare, a direct picture of the math the migration is doing on every sample.
Related Terms
A smile is a failure mode of seismic migration, the processing step that repositions reflection energy to its true subsurface location. Its most common root cause is an inaccurate velocity model, since over- or under-stated velocities curl events into smiles or frowns. Because smiles can mimic closure, they pose a direct risk to structural trap interpretation, and they often arise from poorly attenuated seismic noise that migration spreads into arcs.
Real-World WCSB Scenario: A Smile Mistaken for Foothills Closure
A junior explorer reprocessing a 3D survey over a structurally complex Foothills play west of the WCSB sees a bright, upturned event near the base of the section that looks like a fault-bounded anticlinal closure at the Mississippian carbonate level, a tempting drilling target. A processing review flags the event's perfectly semicircular geometry and its location at the bottom of the velocity-control window, both classic smile signatures, and a migration velocity test shows the feature flattens and disappears when the velocity is lowered, confirming it as an overmigration artifact, not structure.
Catching the smile before maturing the prospect avoids proposing a well on a processing artifact. A deep Foothills exploration well can cost well over 10 million CAD, so the few weeks of reprocessing and velocity-model refinement, a small fraction of that, prevent a potentially dry hole drilled on a smile.