Refraction statics — g-Platform
Refraction statics

Refraction statics software

g-Platform comes with interactive manual and robust semi-automatic first-break picking, followed by a model-free or layer-based refraction statics solution — fast turnaround on 2D and 3D projects of any size. The solver works from the picks themselves: slopes inside an aperture, a global or local solution, azimuthal directions, and a residual term for the short wavelength.

Download free trial How the solution works Watch Demos
Manual or automatic
Interactive and semi-automatic picking
Model-independent
No near-surface model required
2D & 3D
Projects of any size
Refraction statics workspace in g-Platform before the correction, with the source and receiver location map, the time-offset QC crossplot and a source gather carrying its first-break picks
The same source gather in g-Platform after the refraction statics solution is applied, with the first arrivals aligned along the picked trend
g-Platform statics analysis workspace with a shot gather, a first-break trend panel, a colour velocity display and the module parameter panes
In view
Before refraction statics
Onshore, statics decide how good the data can ever be. Field statics from uphole times and elevation statics come first, but the correction that does the real work is refraction statics — and it is not automatically the best answer on every survey, which is why g-Platform ships three routes to it rather than one. All three begin in the same place: the first arrivals, picked interactively or corrected from an automatic pass. From there you can take a model-independent regression on the picks, a cross-correlation of neighbouring gathers over the first-break window, or a tomographic inversion that builds the near-surface velocity model explicitly.
The solution

Everything the solver needs is already in the picks.

The regression solution reads the first-arrival times, computes slopes inside an aperture that gathers every source and receiver within reach, and solves for a static correction at each surface location — without a near-surface model standing between the data and the answer.

Where it sits in the sequence
Field statics from uphole time and depth, then elevation statics, then refraction — the common and efficient solution onshore, though not always the best one on a given survey, which is why more than one route exists.
Pick first, solve second
Pick the first breaks by hand or correct what the automatic pass produced, then hand them to the solver. The same module does both, so the picks and the solution they feed are never out of step.
Slopes inside an aperture
The solution is computed by calculating slopes within a user-defined aperture, with every source and receiver inside that radius taking part, and the slopes optionally recalculated at each iteration.
Model-independent
The refraction statics solution is completely model-independent: it does not require a near-surface velocity model. One can be produced alongside it, but it plays no role in the answer — which is what keeps the turnaround short.
Global and local
Solve globally for the long-wavelength component, or locally within a solver aperture, each with its own iteration count — and a fast-solver option when the run needs to be quicker rather than exhaustive.
Azimuthal directions
The solver can work into several azimuthal directions rather than one, which is what a 3D survey with a real azimuth distribution actually needs.
A residual, short-wave term
Switching on the residual adds a further correction at each source and each receiver — the short-wavelength part of the refraction solution — on top of the smoothed long-wavelength answer.
Supergather picking
Pick across an aperture of locations either side of the current position, scanning a velocity range in steps, with the picks set more accurately on the phase and the local solver's result fed back to improve them.
Routes to the correction

Three ways to the statics, and the tools that judge them.

Regression on the picks is the default. Cross-correlation is the route when picking every break is impractical. Tomography is what you use when the regular solution under-performs or the weathering layer is too variable for it.

Picks & regression
Refraction FB picking - azimuthal solver

The combined picker and solver: interactive or corrected automatic first breaks, a regression solution built on slopes inside an aperture, global or local solving into a chosen number of azimuths, and an optional residual correction at every station.

No pick per trace
Refraction statics - cross correlation

Pick or import an LMO trend instead of every break, then cross-correlate adjacent source and receiver gathers over the window bracketing the first arrivals. Forward and reverse estimates are combined for stability and a moving-average filter removes the long-wavelength trend.

Near-surface model
Tomo statics 2D/3D

Refraction tomographic inversion of the first-break times into a depth-domain near-surface velocity model, with statics obtained by integrating through it from the acquisition surface to the datum. Velocity regularization constrains how smooth the model may be.

Fresnel + SIRT
2D Tomo refraction statics

The 2D tomography engine: Fresnel wave paths rather than rays, updated by the SIRT algorithm until modelled travel times match the picks. Runs on CPU or GPU and distributes across nodes; import the picks, then solve.

Drift
Weathering/drift statics

Computes the correction from an interval velocity model between the recording datum and a pseudo-datum — constant, floating with topography, or taken from bin elevations — then grids it with Kriging or ABOS interpolation across the survey.

Judgement
Statics azimuthal QC

Source and receiver statics correction maps with the matching gathers, and the statics along a chosen source line and receiver line — the QC companion to the azimuthal solver, with the general statics QC module covering the tomographic solutions.

Workflow

Six steps from first arrivals to a corrected gather.

1
Field and elevation statics
Uphole times and depths give the field statics; elevation statics follow, and the two are compared before the refraction solution is asked to explain what is left.
2
Pick the first breaks
Flatten the arrivals with a linear moveout shift, then pick manually, correct the automatic picks, or run the supergather mode across an aperture with a velocity scan.
3
QC the picks, not just the statics
Bad picks make a confident wrong answer. The picks are also what the geometry checks use, so a mis-positioned source or receiver often shows up here first.
4
Choose the route
Regression on the picks, cross-correlation from an LMO trend, or tomography into a near-surface model — the last when the regular solution is not delivering.
5
Solve and inspect
Display the picked and the computed first-break times as two overlays on the current gather, then toggle the corrected gather against the raw one and watch the arrivals line up.
6
Apply, then go to datum
Apply the refraction correction to the data; moving from topography to the final datum is a separate step, so the two never get tangled together.
Parameters & QC

The controls that decide how the answer is shaped.

Two groups matter: what the solver is allowed to do with the picks, and how the smoothed reference map of static shifts is built from them — because that map is what separates the long-wavelength solution from the residual.

Solver & reference model
How the statics are computed and smoothed
Solver mode
Global or local, each with its own iteration count and the local one with its own aperture, plus a fast-solver switch when speed matters more than the last refinement.
Regression aperture
The radius within which picked points around a CDP take part in the low-frequency statics model — the single control with the most influence on how smooth the answer is.
Reference model step
The point step used to build the smoothed map of static shifts. A bigger step smooths more; a smaller one costs time, and it feeds both the low-frequency and the residual models.
Interpolation
Kriging or ABOS, with the Kriging covariance type, range and number of points exposed when Kriging is chosen.
Offset range
Minimum, maximum and step, so far offsets whose picks are unreliable at low signal-to-noise can be kept out of the solution.
Residual & azimuths
Find-residual for the short-wavelength correction at each station, and the number of azimuthal directions the solver works into.
Picking & inspection
What you see while the picks are made
Picking windows
Separate search windows for manual and for automatic supergather picking, with a threshold, a magnet phase to snap to and a sliding step from one gather to the next.
Supergather
An aperture of neighbouring locations picked either side of the current pick, scanning between a minimum and a maximum velocity in steps, with the local solver's result used to improve the result.
Gather views
Load receiver gathers alongside source gathers, build offset classes with a chosen offset step, and re-sort the display into another domain without leaving the module.
Pick overlays
Manual picks and the times the solver computed from them are drawn on the same gather in different colours, so a divergence is visible rather than inferred.
Before and after
Toggle the corrected gather against the uncorrected one — with the space bar or the mouse — and the improvement animates instead of having to be measured.
Dedicated QC
Azimuthal statics QC for the picking-and-solver route, and the general statics QC module for residual and tomographic statics.
Modules in this group
The refraction-statics modules named on this page — not the full list
Refraction FB picking - azimuthal solver Refraction statics - cross correlation Tomo statics 2D/3D 2D Tomo refraction statics Weathering/drift statics LMO Statics azimuthal QC Statics QC Apply azimuthal statics shifts
More g-Platform capabilities

The near surface is where an onshore survey is won or lost.

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

Questions, answered.

Do I need a near-surface velocity model?
Not for the regression route. g-Platform's refraction statics solution is model-independent: it works from the first-break times and the slopes computed inside an aperture, and the manual states outright that it does not require a near-surface velocity model — one can be produced, but it plays no part in the solution. The tomographic route is the opposite by design: it exists precisely to build a near-surface depth velocity model, and you reach for it when the regular refraction solution is not performing or needs improving.
Which of the three routes should I take?
Start with the regression solver on picked first breaks — it is the fastest turnaround and needs no model. Cross-correlation is the alternative when picking every break is impractical: pick or import an LMO trend instead, and the module cross-correlates neighbouring source and receiver gathers over the first-break window, combining forward and reverse estimates and removing the long-wavelength trend. Tomography is the third route, used when the regular refraction statics under-perform or where the weathering layer varies laterally or runs deep; it is supplied as a plug-in that integrates with g-Platform.
Does the refraction correction include the move to datum?
No, and that separation is deliberate. The module calculates the refraction correction only — replacing the low-velocity layers between the surface and the base of the calculated near-surface — while the elevation static that moves the data from topography to the final datum is applied by a separate module. Because most processes in g-Platform, normal moveout included, can be run from topography, keeping the two apart is genuinely useful rather than merely tidy.
How is a refraction statics solution quality-controlled?
Inside the module first: the current source or receiver gather displays the manual picks and the times the solver computed from them as two overlays, so a disagreement is visible immediately, and toggling the corrected gather against the uncorrected one animates the improvement. Beyond that, a dedicated azimuthal statics QC module shows the source and receiver statics correction maps with their gathers and the statics along a chosen source or receiver line — and the general statics QC module covers the tomographic solutions.
Get started

Solve the near surface on your own line.

Pick the first breaks, take the regression, the cross-correlation or the tomography, and QC the answer on the maps. Talk to Geomage about a demo, or take g-Platform for a trial run.

Download free trial Request pricing