Residual statics — g-Platform
Residual statics

Residual statics software

A 2D/3D/4D solution for the near-surface structure that refraction statics did not compensate — both high and mid frequency. Geomage's strategy is surface-consistent and coordinate-based, with model traces built at each CDP location, a global L1-norm inversion carrying a 4D term, and interactive updates after every iteration; a parallel implementation keeps project turnaround short while stack-power maximization does the work.

Download free trial How the shifts are solved Watch Demos
2D / 3D / 4D
Surface-consistent solution
Shot + receiver
One term each, one shift per trace
Every iteration
Stacks and velocities updated
Inline and crossline stack sections before residual statics in g-Platform, beside the source and receiver statics correction maps and the survey location map
The same inline and crossline stack sections after residual statics in g-Platform, with the reflectors continuous across the section
In view
Before residual statics
A stack is an act of faith in alignment. Sum traces that agree and the reflector appears; sum traces a few milliseconds out of step and it smears into the noise. Field, elevation and refraction statics remove the bulk of the near-surface delay, but the weathering zone varies faster than any of them can follow, and what is left is a short-wavelength shift under each source and each receiver. g-Platform measures it the surface-consistent way — one solution per shot gather, a separate one per receiver, combined into a single shift per trace — by correlating every trace against a model trace built from the stack itself, then decomposing the measured shifts back onto the stations that caused them.
How it is solved

Correlate against a model trace, then give the shift back to the station.

The whole method rests on one idea: a trace's misalignment belongs to its source and its receiver, not to the trace, so measure it many times and solve for the stations rather than for the individual measurements.

Short-wavelength, station by station
Residual statics are uniform time shifts that compensate delays in the highly variable near-surface weathering zone, near sources and near receivers — the short-wavelength anomalies left over once datum statics have done their part.
One shot term, one receiver term
A single static solution is computed for every trace from the same shot gather and a separate one for every trace from the same receiver. Combining the two gives the single time shift a given trace actually receives.
The model trace
Built from the CMP stack over a model aperture: at an aperture of one the centre trace is cross-correlated with the trace on each side, the result is stored as the model trace, and the window steps on to the next trace and repeats.
Measure, then decompose
The model trace is cross-correlated with each individual trace of the pre-stack CMP gather to measure its shift, and the resulting system is solved by Gauss-Seidel iteration for the shift belonging to each shot and each receiver.
Super-gather and dip composition
A super-bin aperture sums nearby traces to lift the reference fold on low-fold data, and a dip term — or a Multi-Focusing model scanning a range of dips and several directions — keeps a dipping reflector sharp in the sum instead of smeared.
Terms that stabilise, then step aside
A CMP term and a residual moveout term go into the equations to stabilise the solution and are excluded from the effective correction afterwards. A correlation threshold discards weak measurements, and trend removal takes the long wavelength out.
Watch it converge
Stacks can be created every N iterations and compared in the inline view, and the average stack energy improvement curve says when to stop — flat at iteration five means the next run does not need ten.
A flexible processing sequence
A sub-sequence inside the module conditions the data before the calculation — applying refraction statics, AGC and a band-pass filter, for instance — so the correlation sees prepared traces without a separate flow before it.
Solvers & QC

More than one route to the same alignment.

Different data want different references: a broad correlation window, a supergather built by radius, a specific interpreted reflector, or a per-trace trim where nothing surface-consistent is left to find.

The anchor
Statics - Surface Consistent

Source and receiver residual statics computed surface-consistently from a pilot trace built out of the CMP stack, with velocity updating available on each iteration and the stack response written out as the run proceeds.

Supergather
Statics - reflection based

Assembles a supergather from all CMPs within a radius, NMO-corrects it with the connected velocity model, and cross-correlates each trace against the supergather pilot stack. Successive passes widen the radius to refine the solution, and picks along an interpreted horizon can constrain it.

Horizon-guided
Statics - horizon based

Restricts the alignment to specific reflectors rather than a broad time window: the reflector time is estimated per bin, the correlation window is centred on it, and a least-squares solver decomposes the shifts. Built for areas of strong lateral velocity variation.

Multi-Focusing
MF statics - surface consistent

An iterative cross-correlation engine that decomposes each trace's shift into source, receiver and CMP terms against a reference stack that improves every pass, optionally built with the Multi-Focusing model for wider dip and azimuth coverage.

Per trace
Trim statics - Calculate

For the residual that is not surface-consistent: a short-period shift measured for each trace against a reference and stored in the headers or a trim file. A simple single pass, or an advanced algorithm with iterations, a refined reference and sub-sample interpolation.

Judgement
Statics QC

Source and receiver location maps coloured by correction value, the common-shot gather before and after the correction for any point you click, and — in 3D — the statics along a chosen source line and receiver line.

Workflow

Six steps from a smeared stack to a coherent one.

1
Denoise first
Residual statics measure similarity, and noise is not similar. A denoise pass before the calculation is the difference between a correlation that finds the reflector and one that finds the noise.
2
Bring statics and velocity in
Sort to CDP-offset, load the refraction statics and the stacking velocity, and let the sub-sequence apply them — with AGC and a band-pass — before the module correlates anything.
3
Set the model and the window
Model aperture and super-bin aperture for the reference, calculation start and end time for the zone that matters, correlation length, maximum shift and a correlation threshold for what counts as a measurement.
4
Iterate
Run the cross-correlation, solution and application cycle, with re-weighted solutions for faster convergence, and create a stack every few iterations to watch the section improve.
5
QC on the maps
Read the source and receiver correction maps for spatial patterns that should not be there, compare gathers before and after, and re-run with a different window or shift limit if a cycle skip shows up.
6
Save, then apply
Save the solution to the project database or export it to ASCII, then apply it — surface-consistent, trim and structural corrections summed into one composite shift, forward for correction or in reverse to take it back out.
Parameters & execution

Every decision the solver makes is one you can see.

The controls that matter are the ones that define what is correlated, how far a shift is allowed to move, and when the iteration stops — alongside the memory settings that decide whether a 3D survey runs comfortably or not at all.

Solver parameters
What defines the measurement and bounds the answer
Model & stack
Model aperture in CMPs either side, super-bin aperture for summing near traces, a dip term in ms per trace and a maximum CMP shift — or the Multi-Focusing model with its dip range and number of directions.
Window
Calculation start and end time, drawn on the stack section, plus a correlation length that the module raises automatically if it is short relative to the maximum shift.
Limits
Maximum static shift, and a correlation threshold between 0 and 1 below which a measurement is not used in the solution at all.
Iterations
The number of cross-correlate, solve and apply cycles, plus iteratively re-weighted solutions at each decision stage for faster convergence.
Stabilising terms
The CMP term — unused, residual or cross-correlation — with its own correction window and a minimum-fold limit, and an optional residual moveout term with a smoothing aperture.
What is solved
Surface-consistent static shift, surface-consistent phase rotation with its own step, or both together; trend removal with a grid size and smoothing takes out the long wavelength.
Running & outputs
Memory, visualisation and what leaves the module
Memory
A RAM cache size and a maximum binning fold size the data blocks automatically, or block sizes in the inline and crossline directions can be set by hand.
Visualisation
Create stacks during the run, choose how often, and nominate the inline and crossline to display — the sections that make convergence visible.
Velocity update
Optionally update the velocity field as the statics improve, writing a Vrms model per iteration, with auto-picking and smoothing controls in the vertical and both lateral directions.
NMO
Take NMO-corrected gathers, or take gathers without NMO and let the module apply the velocity model itself, with a stretch factor and an optional shift to datum with a replacement velocity.
Continuity
Statics iterations are saved to a path you choose, and a previous result can be loaded to continue from rather than starting the survey again.
Application
Surface-consistent, trim and structural datasets are summed into one shift per trace, applied forward or in reverse, and the applied values can be written back into the trace headers.
Modules in this group
The residual-statics modules named on this page — not the full list
Statics - Surface Consistent Statics - reflection based Statics - horizon based MF statics - surface consistent Trim statics - Calculate Statics QC Apply static shifts
More g-Platform capabilities

Statics are what the stack is standing 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 Refraction statics First-break picking Velocity analysis Stack imaging Signal processing & denoise Multiple attenuation Deghosting, debubble & designature Deconvolution Time imaging Depth imaging FWI & RTM VSP processing Geometry assignment & QC Data & geometry formats
FAQ

Questions, answered.

What do residual statics correct that refraction statics does not?
Field, elevation and refraction statics come first and take out the bulk of the near-surface delay. What remains are short-wavelength anomalies: uniform time shifts, varying quickly from station to station, caused by the highly variable weathering zone under each source and each receiver. Those are the residual statics, and left in place they blur the stack, because traces that should sum constructively no longer align. The input is moveout-corrected CMP gathers, and the result is a measurable improvement in stack quality beyond what datum statics alone deliver.
How is the reference trace built?
From the CMP stack, over a model aperture you choose. An aperture of one means three traces take part: the centre trace is cross-correlated with the trace on either side to produce a model trace, which is stored, and the process moves to the next trace and repeats along the line. That model trace is then cross-correlated with each individual trace of the pre-stack CMP gather to measure its shift. A super-bin aperture can sum neighbouring traces first to raise the reference fold on low-fold data, and a dip term — or a Multi-Focusing model that scans several dip directions — keeps a dipping reflector from smearing in the sum.
How many iterations should a run use?
More is not better; watch the average stack energy improvement curve instead. If it has gone flat by the fifth iteration of a ten-iteration run, the next run can be limited to five or fewer. The module can create a stack every N iterations so the sections can be compared directly in the inline view, and velocities can be updated on each iteration as well, with a Vrms model written per iteration if you want to see how the velocity field moved with the statics.
What if the misalignment is not surface-consistent?
Then it is trim statics territory. Trim statics measure a short-period shift for each individual trace against a reference — typically a stack or super-gather model — and store the result in the trace headers or a trim file rather than decomposing it into source and receiver terms. A simple single-pass algorithm and an advanced iterative one are both available, the second refining the reference after each pass. Surface-consistent, trim and structural corrections are then summed into one composite shift when they are applied.
Get started

Recover the stack your data already contains.

Surface-consistent shifts, trim statics for what is left, and the maps and stack sections to judge both. Talk to Geomage about a demo, or take g-Platform for a trial run.

Download free trial Request pricing