Seismic reflection interpretation begins with the time it takes an elastic wave to travel down, reflect from a boundary, and return to a receiver. If you are trying to read a common-midpoint (CMP) gather, normal moveout (NMO), stacking velocity or a two-way-travel-time seismic section, start by distinguishing the recorded trace from the geological surface being inferred. A reflector is an interpretation of wave contrasts; the coloured or wiggly event on a display is not a photograph of an underground layer.
A reflection survey records many traces with source and receiver positions. Processing can sort those traces by midpoint, estimate the moveout of coherent reflections across offsets, correct the timings and stack compatible waveforms. These operations can clarify real reflections but can also create misleading continuity when velocities, statics, mute windows or trace selection are wrong. Reading the processing choices is as important as looking at the final section.
This guide explains near-surface and exploration-style seismic reflection images through the measurement chain: acquisition geometry, shot records, CMP fold, hyperbolic moveout, velocity analysis, normal-moveout correction, stacking, seismic bandwidth, migration, time-to-depth conversion and validation. Numerical examples use synthetic data and explicit units. The aim is to help a reader identify what a seismic image can support, what it cannot decide alone, and which checks prevent a plausible-looking reflector from becoming a false geological claim.
Quick answer: what is a CMP gather?
A CMP gather is a collection of seismic traces whose source–receiver pairs share approximately the same midpoint on a survey line. The pair locations need not coincide: one trace might use a source 100 metres to the left of the midpoint and a receiver 100 metres to its right, while another uses a shorter or longer separation. Their common location on the map allows an analyst to compare how a putative reflection arrival time changes with source–receiver offset.
In a simple horizontally layered model with a flat reflecting interface, reflection travel time grows with offset in a predictable hyperbolic pattern. A processing workflow estimates a velocity that best flattens that event, then combines compatible traces to improve visibility. That is the central CMP idea, not a guarantee that all rays reflected from exactly the same subsurface point in complex geology.
Why reflection differs from first-break refraction
Surface seismic refraction interpretation often starts with the earliest arriving energy and asks how its travel time changes with source–receiver distance. Reflection interpretation focuses instead on coherent later arrivals returning after interaction with interfaces that contrast in elastic impedance. One record can contain direct, refracted, reflected, converted and surface waves; they are not interchangeable simply because the plot uses time on the vertical axis.
A reader who confuses the two methods can mistake a refracted head wave for a reflection horizon or use a refraction velocity formula on a CMP curve. The preceding surface seismic refraction guide concentrates on first breaks and slope changes. This article addresses later reflected energy, midpoint sorting and the transformations that lead to a stacked image.
What is measured before a subsurface interpretation exists?
A seismic receiver senses a form of ground motion, often through a geophone or accelerometer, or pressure in marine applications. A source injects an elastic disturbance into the environment. The instrument records voltage or digital samples calibrated through its response, along with timing and source–receiver geometry. Raw data are sequences of measured waveforms, not direct measurements of bedrock depth.
A recorded trace may include signal, ground roll, ambient noise, source signature, coupling effects, instrument response and several propagating wave types. The presence of one large amplitude does not establish a geological interface. The analyst needs repeatability across neighbouring traces, physically plausible moveout and supporting velocity information before assigning a subsurface meaning.
Source position and receiver position are not the same as midpoint
Suppose a source is at horizontal coordinate 40 metres and a receiver at 160 metres. Their midpoint is 100 metres, while their full offset is 120 metres. The midpoint represents the arithmetic average of the endpoints; the offset is their separation. Some equations use half-offset, so the symbol h may equal 60 metres even though the measured source–receiver distance is 120 metres.
This distinction is fundamental because the moveout equation changes appearance depending on whether x denotes full offset or h denotes half-offset. Before copying an equation, check its definitions. A factor-of-two error in offset can produce a very different inferred velocity while leaving the algebra superficially tidy.
What does a shot gather show?
A shot gather contains traces recorded from one source activation at multiple receivers. On a shot record, arrivals can display several different slopes and waveforms, including direct waves, refracted arrivals, surface waves and deeper reflections. The horizontal axis may denote receiver number, distance or offset, while the vertical axis is often recording time after the shot.
A shot gather helps judge source repeatability, coherent noise, dead channels and the presence of visible reflection energy. It is not the same as a CMP gather because the source location is held approximately fixed rather than the midpoint. Confusing the two can lead to incorrect claims about reflection hyperbolas and fold.
Geometry assignment is part of the result
Field notes, coordinates, elevation, source numbering and receiver channels must be correctly associated with traces. A single swapped receiver coordinate can move a trace into the wrong CMP bin. Incorrect source timing or polarity can create visible anomalies that processing later smooths without necessarily resolving their cause.
When interpreting results from a contractor, ask whether survey geometry was independently checked against field records. If a conspicuous line appears in a stacked section at a source-station boundary, consider an acquisition or processing artefact before mapping it as a fault. A clean image does not prove the underlying trace geometry was correct.
What does fold mean?
CMP fold describes how many traces contribute to a midpoint gather or a stacked bin, subject to actual acquisition coverage and processing selections. A fold of twelve does not mean twelve geological layers; it means that, for the bin in question, twelve trace contributions are present before or after stated editing decisions.
Fold varies near survey edges, around obstructions and after muting. A high nominal fold may conceal a much lower effective fold for a particular reflection time if many far-offset traces were removed. Interpret the fold map alongside the section rather than assuming identical coverage everywhere. Discontinuities that track fold changes deserve extra scrutiny.
The first useful equation: zero-offset two-way time
In an idealised uniform medium with propagation velocity v, a horizontal reflector at depth z has zero-offset two-way travel time t0 equal to 2z divided by v. The factor two arises because the wave travels to the reflector and back. Rearranging gives z equals v multiplied by t0 divided by two.
For a synthetic reflector at depth 200 metres in a medium with seismic velocity 2,000 metres per second, t0 equals 0.200 seconds. In a layered environment, conversion requires an appropriate velocity model, and one constant value may be inadequate. The example explains units and geometry, not the true depth beneath any particular survey.
Why the moveout curve looks hyperbolic
For a simple flat reflector above a constant-velocity medium, the two-way time at full source–receiver offset x follows t(x) squared equals t0 squared plus x squared divided by v squared. Taking the square root gives a hyperbola in a time-versus-offset display. The formula assumes idealised geometry and propagation; it is not an exact universal equation for dipping reflectors, anisotropy or strongly laterally varying velocity.
The expression is powerful because timing changes across offset provide information about how fast waves propagate in the assumed model. A geophysicist searches for velocities that align a reflection event across traces. This is where a CMP gather contributes beyond one isolated seismic arrival.
A fully checked moveout calculation
Use the illustrative reflector with t0 equal to 0.200 seconds and v equal to 2,000 metres per second. At full offset x equal to 300 metres, x divided by v is 0.150 seconds. Therefore t(x) equals the square root of 0.200 squared plus 0.150 squared, or the square root of 0.0625 seconds squared, which equals 0.250 seconds.
The normal-moveout difference is 0.250 minus 0.200, or 0.050 seconds. This calculation demonstrates the use of full offset. If the midpoint-to-source half-offset were 150 metres, it would enter the equivalent formula as 4h squared divided by v squared. Substituting h as if it were full offset would make a different, incorrect time prediction.
What normal-moveout correction tries to do
NMO correction shifts traces so that energy from the same modelled reflection arrives at a common reference time despite differing source–receiver offsets. If the velocity function is appropriate, the event becomes flatter across the CMP gather and can be stacked coherently. If velocity is wrong, residual curvature remains.
NMO does not mean moving rock layers vertically or physically straightening an underground surface. It is a time-axis operation applied to waveforms under a moveout model. Read the original gather and corrected gather together when possible. A flattened event is evidence that a chosen timing model matches the data; it is not automatic proof of a unique geological velocity or horizon.
Stacking velocity is not automatically geological interval velocity
A velocity selected to flatten CMP events is often called NMO or stacking velocity. In the simple constant-velocity, isotropic flat-layer case, it can coincide with propagation velocity. In layered or anisotropic settings, stacking velocity may differ from local interval velocity and from the velocity needed for accurate depth imaging.
Consequently, multiplying an NMO velocity by half a reflection time may not yield a trustworthy depth to every horizon. Depth conversion can require interval-velocity analysis, well ties, checkshots, tomography or other constraints. A measured or fitted moveout parameter and an interpreted physical velocity should remain separate columns in a technical summary.
Velocity scans and semblance panels
Velocity analysis often tests a range of candidate NMO velocities against CMP gathers and calculates a measure of alignment or coherence, such as semblance, for particular times. A semblance panel displays the relative coherence of corrected events across trial velocities. Bright concentrations can help analysts pick a velocity trend, but they can also reflect coherent noise or multiple arrivals.
The horizontal axis may be velocity, while the vertical axis is zero-offset time. Check the panel’s units and whether its colour scale represents semblance, energy or another score. A well-defined maximum near a plausible event is useful, but a single attractive pixel does not settle model validity. Continuity, uncertainty and expected geology still matter.
Why a low velocity can overcorrect a gather
For the hyperbolic approximation, a smaller trial velocity generally predicts a larger moveout at a given offset. If the analyst applies too low a velocity, the correction may pull far-offset arrivals too strongly; residual moveout changes sign relative to an undercorrected event. Too high a velocity can leave the event insufficiently flattened.
One should inspect corrected-gather residuals rather than relying exclusively on a picked coloured semblance path. When event character is weak, several trial velocities may appear similarly plausible. State that ambiguity, or use external constraints, instead of selecting an arbitrary value with more decimal places than the data can support.
NMO stretch can distort the waveform
NMO correction changes the time sampling of an event, especially at short zero-offset times and large offsets. This can stretch a wavelet and distort its frequency content. Processing often mutes heavily stretched regions so those traces do not damage the stacked reflection. The trade-off is that the mute reduces effective fold.
A shallow event may therefore be represented by fewer useful offsets than the acquisition map suggests. A reader who compares its apparent continuity with a deeper event should check stretch mute and residual coverage. A smoother line after processing can be partly an artefact of selective trace removal rather than a newly discovered continuous geological horizon.
Static corrections and near-surface conditions
Variable topography, weathered layers and source–receiver elevations can introduce timing differences that are not caused by deeper reflectors. Static corrections attempt to account for some of these changes. If the correction is wrong or insufficient, traces from the same reflector can fail to align, creating false structural relief or destroying stack coherence.
The specialist guide to seismic statics and false structure covers that narrower problem. For CMP reading, the practical lesson is to ask whether processing used an appropriate elevation datum, weathering model and residual statics strategy. A visually attractive stack cannot substitute for a defensible near-surface correction.
Why a CMP bin is not a point in the ground
Sorting by midpoint simplifies data geometry, but real reflection points can depart from that midpoint in dipping or heterogeneous structures. The common-depth-point shorthand can be misleading when taken literally. Different offsets may also sample different paths and reflection angles, especially beneath steeply dipping features.
The Society of Exploration Geophysicists CMP sorting explanation makes clear that source–receiver geometry is transformed into midpoint–offset coordinates for processing. That operation does not turn complex three-dimensional rays into identical straight down-and-back paths. Distinguish the bookkeeping midpoint from the actual wave propagation path.
Coherent reflection versus coherent noise
Not every aligned wave train represents a desired subsurface reflector. Ground roll, guided waves, multiples and acquisition artefacts can show organised moveout across traces. Filtering may help remove these events but can also damage reflection amplitudes or create a cleaner-looking result whose geology is uncertain.
A strong interpretation checks several displays: shot gathers, CMP gathers, velocity panels and the final stack. Ask whether the claimed horizon persists under reasonable processing alternatives and whether it follows a reflection-like moveout pattern rather than a known noise mode. Signal coherence is a necessary clue, not a complete geological identity.
Why the stack improves visibility
After timing correction, traces associated with an event can be combined so coherent reflection energy reinforces while some uncorrelated noise cancels. Under highly simplified conditions of equal-amplitude signal and independent equal-variance noise, the signal-to-noise ratio of an average of N traces improves by roughly the square root of N. Real data rarely meet every assumption.
If nine truly independent traces contributed to an event, the idealised improvement factor would be about three. That number is not a guaranteed field performance claim. Coherent noise, varying coupling, velocity errors and missing traces can substantially change the outcome. A fold count alone is not a reliable measurement of the quality of the stacked event.
What a stack means and what it leaves out
A seismic stack is a calculated trace made by combining selected records after applying time corrections. Processing does not simply average everything collected. The analyst may exclude noisy channels, mute far offsets, apply scaling, filter waveforms and reject inconsistent arrivals. These decisions determine which information remains visible in the final image. High nominal fold does not guarantee high effective fold at every time.
A claim of unusually good signal-to-noise ratio should be supported by representative gathers or quality measures rather than by a coloured stacked section alone. Coherent noise can also reinforce during stacking, especially if it has a similar moveout to the desired event. The correct reader’s question is not merely whether the stack looks smooth, but whether multiple observations consistently support the interpreted arrival.
Amplitude balancing changes what brightness means
Before stacking, processors may apply geometrical spreading compensation, amplitude recovery, filtering or automatic gain control. Such treatments can make weak features easier to see, but they also change relative amplitudes. A bright line after strong gain does not necessarily represent a physically stronger underground impedance contrast than a dimmer line.
For structural interpretation, visibility can be useful. For quantitative amplitude-versus-offset analysis, inappropriate gain or normalisation can be harmful. A report should identify whether its product preserves relative amplitudes. When it does not, describe apparent reflectors without assigning a specific material property or fluid saturation on the basis of brightness alone.
Two-way time is not depth
The vertical axis of a stacked section is often two-way reflection time, measured in milliseconds or seconds. It denotes a wave’s total return journey under the processing reference, not metres beneath the surface. A horizon at 0.400 seconds two-way time cannot be reported as 400 metres deep simply by reading the axis label.
Under a simplified constant-velocity medium with speed 2,000 metres per second, 0.400 seconds corresponds to about 400 metres of depth because the wave travels down and back. But if the representative speed were 1,800 or 2,200 metres per second, the same travel time would imply 360 or 440 metres respectively. A seismic-to-depth conversion therefore needs a velocity model and honest uncertainty.
Different seismic velocity names mean different things
An NMO or stacking velocity is fitted to offset-dependent reflection timing. A root-mean-square velocity is a model construct used in layered reflection analysis. An interval velocity describes a particular layer under a set of physical assumptions. Although related in some idealised settings, none should be casually labelled the direct wave speed of every layer in a complex earth model.
Converting picked velocities to interval velocities can magnify errors, particularly when nearby time picks are uncertain. A spiky interval-velocity curve may therefore reflect unstable calculation rather than abrupt lithological change. When an article supplies velocities, record how each was obtained, its units, reference time and model. The dictionary word velocity alone is too broad to carry that information.
Why depth calculations need independent ties
A seismic reflection horizon can appear perfectly continuous while its physical depth remains uncertain. A borehole intersecting a recognisable boundary can constrain travel time and velocity at one location, but lateral changes in saturation, lithology or weathering may invalidate a constant-speed extrapolation along the whole line.
A careful interpretation distinguishes a modelled depth surface from an observed borehole intersection. If there is no independent tie, say that depth conversion is approximate under an assumed velocity. Precision in a spreadsheet is not geological certainty. A depth reported to the nearest centimetre can still be unreliable when the velocity model is only approximate.
Bandwidth limits vertical resolution
Seismic wavelets occupy finite intervals of time. Reflections from two closely spaced interfaces can overlap, interfere or form a single apparent event. A thin layer’s reflectivity may influence the waveform without producing a separately resolvable top and bottom. It is therefore wrong to infer one neat geological boundary from every peak and trough in a wiggle trace.
A commonly used rough scale for vertical resolution is one-quarter wavelength under specific assumptions. At a wave speed of 2,000 metres per second and a dominant frequency of 50 hertz, the wavelength is 40 metres, giving an illustrative quarter-wavelength scale of 10 metres. This heuristic is not a guaranteed ability to distinguish every pair of layers 10 metres apart.
Temporal sampling is not the same as field bandwidth
With a sample interval of two milliseconds, the digital sampling frequency is 500 samples per second and the Nyquist frequency is 250 hertz. This merely sets an upper mathematical frequency limit in the sampled representation. It does not mean the field source, earth and receivers produced useful seismic energy across that range.
A signal might contain most of its reliable energy within a much narrower band because of attenuation, coupling and filtering. When someone claims fine resolution because the plot has many pixels, ask for actual spectrum, wavelet duration and noise. A digital trace can be closely sampled yet still contain little high-frequency information about thin layers.
Migration is an imaging operation with assumptions
A nonmigrated seismic display can place energy at apparent coordinates that differ from the location of the reflecting or diffracting source, particularly for dipping reflectors and compact scatterers. Migration attempts to reposition this energy using an assumed propagation model. It may improve imaging when acquisition geometry and velocity are adequate.
Migration can also create spurious focusing, false edges or misplaced features when those assumptions fail. The specialist article on reflection velocity pull-up and migration artefacts treats that narrower interpretive risk. This guide focuses on explaining the basic CMP-to-stack chain before imaging claims are added.
Why an unmigrated diffraction resembles a hyperbola
Small discontinuities, edges or isolated scatterers can generate curved arrival patterns across neighbouring traces. In a time section, these may resemble arches or elongated tails even though the underlying target is compact. The moveout shape depends on acquisition and propagation geometry, so simply tracing the top of a hyperbola does not establish the object’s shape.
A migration step can focus compatible scattered energy toward a candidate location, but it depends on velocity and sampling. If a bright arch vanishes or moves after migration, consider whether it was diffraction energy rather than a continuous reflector. Conversely, a newly sharp spot may still be a processing-dependent image. Strong interpretations compare multiple stages.
Multiples imitate extra reflecting layers
A multiple is a wave that undergoes additional reflections along its path before returning to the receiver. A multiple can be coherent and strong enough to resemble a primary event in a stack. Its later arrival may tempt an interpreter to invent a deeper boundary that was not independently sampled by the assumed primary path.
Check known strong interfaces, expected timing relationships, moveout and processing aimed at attenuating multiples. Do not assume every later coherent event is primary merely because it can be drawn as a continuous line. A good geological interpretation names uncertain events as candidate primaries or multiples until the distinction is supported.
Wavelet phase and polarity must be specified
Seismic data can be displayed with different colour conventions, amplitude gains and processing phase choices. A positive or negative lobe has physical meaning only with a calibrated polarity convention and a sufficiently understood wavelet. Some processing may change phase or frequency content while preserving useful timing information.
If a claim depends on a polarity reversal, inspect the sign convention and any synthetic seismogram or well tie used to relate the waveform to an impedance boundary. A blue-to-red colour change is not a universal code for clay-to-sand transition, groundwater or hydrocarbons. The same colours can be assigned to opposite signs in different plotting systems.
Why elastic impedance contrasts create reflections
In a simplified normal-incidence acoustic model, impedance equals density multiplied by wave propagation speed. When a wave encounters a boundary separating materials with different impedance, a portion of the energy can be reflected. The idealised reflection coefficient is the lower impedance minus the upper impedance divided by their sum.
For an invented upper material with density 2,000 kilograms per cubic metre and speed 1,800 metres per second, impedance is 3.60 million in corresponding SI units. For a lower material with density 2,200 and speed 2,200, it is 4.84 million. The ratio (4.84 minus 3.60) divided by (4.84 plus 3.60) is approximately positive 0.147. That dimensionless model coefficient is not a directly calibrated field amplitude after gains and attenuation.
A large reflector amplitude does not identify one material
Impedance contrasts may arise from multiple combinations of density and wave speed. Soil moisture, weathering, stratification and rock fabric can affect acoustic properties. Two very different geological interfaces may have similar reflection coefficients, while two nearby locations of the same named material can differ. A reflection is evidence for a wave-property contrast under the interpretation.
Mapping a reflection directly to a rock type without boreholes, geological context or other measurements is a nonunique inversion. The phrase ‘reflector interpreted as a boundary between units’ is more accurate than ‘the seismic signal measured the clay surface’. This distinction matters when drilling, environmental remediation or construction decisions depend on a proposed depth.
Velocity pull-up and apparent time relief
Suppose waves travel relatively quickly through a lateral zone above an otherwise flat deeper reflector. Their two-way travel time may be shorter within that zone, causing the reflector to appear higher on a time image. A slower near-surface zone can cause apparent pull-down. Neither apparent bend must represent a physical fold or fault in the deeper interface.
Before interpreting an unusual structure, test whether local velocity changes explain it. Review near-surface statics, surface-wave measurements, borehole velocity information and alternative depth-conversion models. A time section illustrates timing variation first; a geological depth structure is a conclusion requiring a velocity model.
Fold gaps can make a continuous interface look broken
A seismic line may cross a road, vegetation or inaccessible ground where source or receiver positions are unavailable. Missing traces can reduce fold or offset diversity in a segment. A continuous reflector may then weaken or disappear from the stack simply because acquisition coverage deteriorated.
A fault interpretation is stronger when the discontinuity occurs despite comparable data quality and is supported by consistent offsets, reflector terminations and independent evidence. If the break aligns with a fold or mute boundary, the first hypothesis should include a data-coverage problem. Documenting uncertainty prevents a nongeological acquisition gap from becoming a false structural map.
What a two-dimensional line cannot see
A 2D survey is usually displayed as a section beneath one line, yet seismic waves propagate through a three-dimensional subsurface. Reflectors outside the survey plane can contribute energy. Dipping or curved structures may be projected incorrectly when processing assumes geology varies little across the line direction.
In a complex area, 3D acquisition or intersecting 2D lines may help test whether a feature belongs beneath a specific position. No seismic method turns complex wave propagation into a literal photograph. Even with more coverage, velocity errors and ambiguous contrasts remain. State the sampling geometry whenever describing lateral continuity.
Reading a finished stack systematically
Start with the axes: horizontal midpoint distance or station, vertical two-way time or depth, and the correct units. Identify whether the section has been stacked, migrated, depth converted, filtered, balanced or displayed after significant gain. Then examine where fold, velocity and signal quality change along the line.
Trace robust, repeated events only after locating representative source and CMP gathers. Assign confidence based on coherence, bandwidth, independent observations and reasonable processing alternatives. If an event is visible only after one strong filter or in a low-fold area, the interpretation should be more tentative. The strongest map separates confirmed measurements from geological hypotheses.
A two-reflector example
Imagine coherent stacked events at zero-offset times 0.20 and 0.40 seconds. Under a constant velocity of 2,000 metres per second, their illustrative depths would be 200 and 400 metres. The implied separation is 200 metres in that model. This arithmetic is correct within the uniform-speed assumption.
A layered velocity structure would change the conversion, and a multiple could make one of the events a misleading candidate horizon. Reporting the depths without assumptions would be an error of meaning rather than multiplication. A clear explanation includes the time observations first and then labels the depth estimates as model-derived.
A practical checklist for an unfamiliar section
If a contractor supplies only the final coloured seismic image, ask for source and receiver geometry, sample interval, fold map, representative gathers, velocity analysis and processing sequence. These items help identify what information was discarded or transformed. A geological map should also show where calibration data exist.
Avoid treating a missing document as proof the result is wrong. Instead state what remains unverified and how that affects confidence. For a preliminary teaching example, the basic event geometry may suffice. For a construction or environmental decision, the same uncertainty might require an additional survey or boring.
A worked transfer problem for a new CMP
A synthetic reflection has zero-offset time 0.300 seconds and an assumed NMO velocity of 1,500 metres per second. Predict the arrival at full source–receiver offset 300 metres. The offset divided by velocity is 0.200 seconds, so the hyperbolic travel time is the square root of 0.300 squared plus 0.200 squared. That is approximately 0.361 seconds.
The predicted moveout is about 0.061 seconds. This demonstrates the timing model, not a uniquely measured depth or material boundary. If the actual event shows different curvature, possible explanations include velocity heterogeneity, dip, anisotropy, geometry error, converted arrivals or misidentified signal. Do not force every curved event to match a chosen hyperbola.
Independent transfer problem about fold
Two CMP bins have nominal folds sixteen and four. Under an idealised model of independent equal-variance noise with identically aligned signal, square-root fold predicts relative signal-to-noise improvement factors four and two. Is it therefore certain that the first stacked bin is twice as reliable? No.
The real traces may contain coherent noise, different coupling, stretch mutes or velocity errors. A high-fold bin can be poor if the traces reinforce an unwanted arrival. A lower-fold bin can preserve a clear reflector in favourable acquisition. The square-root relationship teaches one limiting principle, not a universal quality ranking.
Frequent interpretation errors and targeted repairs
One mistake is treating two-way seconds as metres; repair it by retaining the time unit until a velocity model is supplied. Another is confusing shot gather with CMP gather; repair it by checking which coordinate is shared. A third is calling NMO velocity a direct depth velocity; repair it by identifying the physical model and calibration.
Other mistakes include taking amplitude colour as lithology, mistaking coherent multiples for primary reflectors, and assuming that migration always sharpens truth rather than also focusing errors. Repairs include well ties, processing comparisons, fold checks, representative gathers, wavelet inspection and alternative models. Every repair traces the final assertion back to a measurement that can be checked.
Answers to common questions
Does a CMP gather show one geological point in every situation? Not exactly; midpoint sorting is a geometrical approximation. Is a two-way-time axis already a depth scale? No. Can NMO velocity differ from an interval velocity? Yes. Can a high-fold stack contain coherent noise? Yes. Does a bright event necessarily indicate a particular rock? No.
Can wrong velocity create apparent folds or pull-up? Yes. Is migration always necessary for every use? It depends on the imaging goal and geometry. Can an unmigrated diffraction resemble a curved geological surface? Yes. Can a multiple appear deeper than a primary? Yes. Can a borehole help? Yes, by constraining depth and material interpretation locally.
Method sources and connected reading
The Society of Exploration Geophysicists explanation of CMP sorting identifies the relation between field source–receiver geometry and midpoint–offset coordinates. The SEG reference for NMO correction discusses moveout and stacking. An educational seismic CMP hyperbola calculation illustrates the timing formula. They are method sources; the numerical examples here are hypothetical.
For adjacent questions, read surface seismic refraction, downhole seismic interval velocity, seismic statics and apparent structure and cross-linguistic vocabulary transfer. These are related but do not substitute for the CMP measurement chain.
Closing: what the image actually knows
The most reliable seismic reflection account begins with waveforms and geometry, continues through moveout and stacking, and reaches geological structure only after examining velocity and processing assumptions. A polished image is valuable evidence, but it is not a direct picture of rock layers. Terms such as time, offset, amplitude, fold and depth carry different measurements.
When a reader can follow each transformation, a complicated CMP section becomes understandable without false certainty. That is the standard for translating a specialist geophysical result: give the main conclusion, show how it was derived and preserve the boundary between observation and interpretation.