Velocity analysis

Velocity analysis software

A highly interactive velocity workflow for 2D and 3D: tunable semblance, super-gathers and constant-velocity mini-stacks, vertical picking down the trace and horizontal picking along a horizon, and the model tools that turn a set of picks into a field NMO, migration and depth conversion can use.

Download free trial How picking works Watch Demos

Vertical + horizontal

Semblance down the trace and along a horizon

Manual + auto

Corridor-constrained picking

At topography

Picked on the recording surface

Velocity analysis with semblance panel and NMO-corrected CDP gather in g-Platform

Horizon-guided velocity analysis workspace in g-Platform with horizontal semblance

In view

Semblance, mini-stacks & gather

Velocity analysis is the search for the field that flattens the data. A range of trial velocities is applied to a CMP gather, each one NMO-corrects it, and a coherence measure — semblance — scores how well the reflection events line up. High semblance means the events are aligned, which means the velocity is close to right at that time and place. The picks are interpolated into a function, and that function then governs NMO, stacking, migration and depth conversion alike: it is the single input that most decides whether the final image is focused. In g-Platform the work is done interactively, on displays that update as the pick moves, and the velocities are picked at topography rather than on an assumed flat datum.

Picking

Every view you need to trust a pick.

A semblance panel on its own hides as much as it shows. The picking session pairs it with the gather, the mini-stacks, the offset panels and the section being built, so a decision can be checked against the data it came from.

Semblance, tuned to the data

Set the velocity range and step, the smoothing and normalization windows and the display resolution; choose plain, weighted or similarity-weighted semblance; and have NMO applied in real time during the scan so every trial velocity is judged on its own merits.

Super-gathers where fold is thin

Pool neighbouring bins in the inline and crossline directions for the spectrum and the common-offset view. A multiplication factor widens that pool further and stabilises the semblance laterally — more events survive, at some cost in lateral detail.

Constant-velocity mini-stacks

Pick on stacked panels as well as on the spectrum: mini-stacks built at a chosen velocity increment, or around the current interpolated velocity within a maximum deviation, at an inline or crossline orientation and a radius you set.

Gathers with and without NMO

The raw gather, the NMO-corrected gather and common-offset views are tracked to the picking window with the travel-time curves overlaid, and the stretch factor that governs the display is the same one that governs the stack.

A corridor, then the auto-picker

Pick a corridor skeleton at a few locations; it interpolates between them, and the auto-picker stays inside it. A semblance threshold keeps it from picking noise, and semi-automatic mode simply snaps your own picks to the nearest maximum.

Horizontal velocity analysis

Pick along a horizon rather than down a trace: with a reference stack and a picked or imported horizon, the horizontal semblance shows velocity against CDP along that reflector and the field is updated layer by layer.

Migration velocity scanning

Stacking velocity is not migration velocity. The migration side migrates the data across a range of velocity percentages into one storage file, then picks Vrms and delta-Vrms on those migrated panels — the automatic velocity-percentage scan.

Referenced to topography

Picking happens on the recording surface: the moveout equation takes source, receiver and bin elevations with the replacement velocity, and the elevation statics are applied automatically. Shifting to a final datum is a later, explicit step.

From picks to a model

A field of picks is not yet a model.

Picks have to be converted, cleaned and stored before the rest of the chain can use them. These are the modules that do it — and the ones that make the next picking session faster than the last.

Conversion

Create velocity model

Turns gathers, a SEG-Y handle, an internal picks collection, an ASCII picking file, a constant velocity or a third-party format — Geocluster, CGG 3D, ProMAX table — into the internal velocity model that NMO, Kirchhoff PreSTM and the depth chain consume.

Editing

Velocity editor

Manual correction of a 2D or 3D model in time or depth: smoothing, spike removal, removal of unrealistic inversions, interpolation across edited areas, replacement with a constant inside a region, and polygon-based surgical editing on vertical sections and 3D volumes.

Storage

Precompute VA

Computes the velocity analysis for the whole volume once and stores it on disk, so picking on a large survey reads from storage instead of recalculating semblance each time.

Application

NMO

Removes the offset-dependent travel-time delay so a gather can be stacked, with a form of the equation that accounts for source, receiver and bin elevations against the replacement velocity.

Beyond the hyperbola

When flat gathers need more than one velocity.

NMO assumes hyperbolic moveout, which holds for horizontally layered ground and moderate offsets. Long offsets, fractures and anisotropic sequences break that assumption, and residual curvature survives even a perfect isotropic pick. Three routes address it.

Azimuthal NMO analysis

On wide-azimuth data, NMO velocity varies with azimuth. An ellipse is fitted to the azimuth-versus-velocity cloud at each bin and time sample — robustly, with outlier rejection — giving the fast and slow velocities, the azimuth of the symmetry axis and the ellipticity: the indicators of fracture orientation and density.

Eta on migrated gathers

In the time-imaging chain the anisotropy parameter eta is scanned on NMO-corrected common image gathers, picked inside a corridor, and applied — producing eta-corrected gathers and stacks where an isotropic velocity alone leaves residual moveout.

Higher-order moveout

Picking is not limited to the second-order hyperbolic term: fourth-order and anisotropic moveout picking keep long-offset events flat where the hyperbolic approximation has run out, which matters as much for the mute as it does for the stack.

Workflow

Six steps to a velocity field you can hand on.

Decide the grid

Choose the picking interval in inlines and CDPs — a 2 km spacing on a 12.5 m CMP interval is every 160th CDP — or work bin by bin where the section changes character.

Set up the scan

Velocity range and step, semblance smoothing and resolution, stretch factor, super-gather radius, and mini-stacks if you want constant-velocity panels alongside the spectrum.

Pick a first pass by hand

Establish the trend at a coarse grid, watching the gather flatten and the interval velocity curve behave. Bad picks are removed with a polygon rather than one at a time.

Corridor, then auto-pick

Pick the corridor skeleton, check it against the semblance at several locations, then let the auto-picker fill the survey in inside those limits.

Refine and edit

Run a residual pass on the current bin or the whole dataset, guide the field along horizons where structure demands it, and clean the model in the velocity editor.

Build and hand on

Save the picks to file or to the project database, build the internal velocity model, and connect it to NMO and stacking, to the migration velocity scan, or to the depth model builder.

Modules in this group

The velocity family in the g-Platform module tree

Velocity analysis Stack imaging HVA stack imaging Precompute VA Create velocity model Velocity editor NMO Azimuthal NMO analysis RMO by picking PSTM imaging (eta / VTI)

More g-Platform capabilities

Velocity is the input everything else depends on.

The same project, the same batch processing and the same cluster carry a survey from field tape through to depth. These are the other topics in the g-Platform workflow.

g-Platform overview Stack imaging Time imaging Depth imaging FWI & RTM VSP processing Multiple attenuation Deghosting, debubble & designature Signal processing & denoise Deconvolution Refraction statics Residual statics First-break picking Geometry assignment & QC Data & geometry formats

FAQ

Questions, answered.

Where does velocity analysis happen in g-Platform?

Stacking velocity analysis lives inside the stack imaging application, so picking, mute, conditioning and the resulting stack are one session rather than four — it is the module to pick in. A lighter velocity analysis module displays the semblance panel on its own, which makes it the natural QC tool for judging a processing step by what it does to the spectrum, before and after a demultiple pass for instance. A precompute module stores the analysis for large surveys, and migration velocity analysis is a separate step on migrated panels, handled by the velocity selector and the imaging selectors.

Is picking manual or automatic?

Both, and usually in that order. A first pass is picked by hand at a coarse grid to establish the trend. Then a velocity corridor is picked as a skeleton at a few locations and interpolated between them, and the auto-picker works inside it, ignoring anything under a semblance threshold you set. Semi-automatic mode snaps your own picks to the local semblance maximum, and a residual pass re-picks either the current bin or the whole dataset once a first field exists.

What is horizontal velocity analysis?

Picking velocity along a reflector instead of down a trace. You pick or import a horizon on a reference stack, and the horizontal semblance panel shows velocity against CDP along that horizon, so the velocity is updated layer by layer where the structure actually is. It needs a reference stack and a horizon, and on a large survey it reads a precomputed velocity analysis rather than recomputing the semblance.

How do picks become a velocity model?

A dedicated module converts them. It accepts gathers, a SEG-Y handle, an internal picks collection, an ASCII picking file or third-party velocity formats, and writes the internal model that NMO, Kirchhoff PreSTM and the depth chain consume. The velocity editor is the companion for cleaning that model up: smoothing, spike and inversion removal, interpolation, polygon editing on 2D sections and 3D volumes, in time or in depth.

Get started

Pick a velocity field on your own gathers.

From a first manual pass through corridor-guided auto-picking to an edited model ready for migration — take g-Platform for a trial run, or talk to Geomage about your survey.

Download free trial Request pricing