Reflection Tomography - PGS hyperTomo

PGS hyperTomo® is a wavelet oriented tomography solution that makes direct use of the wavelet level properties determined in PGS’ Beam system.

Listed below are the steps involved:

  • The input data is dip decomposed into wavelets using a process called PGS Dipscan.
  • In the migration step (using PGS hyperBeam®) these wavelets are mapped to the migrated location using the current earth model. A reconstruction workflow then allows these wavelets to be reconstructed in migrated space.
  • PGS 3DRNMO (3D residual normal moveout) determines the time shift to align wavelets with a reference stack.
  • Finally, hyperTomo updates the velocity model using the 3D RNMO shifts.

 

Reflection Tomography Workflow

 

With hyperTomo, both global and layer based updates are possible and require minimal processing time. The data is input into the inversion and this iterative approach to velocity model building utilizes initial large global updates to determine the regional velocity trend. This is followed by the gradual introduction of higher velocity variations which can either be global or layer based.

Wavelet Formation using Dipscan

The formation step of the wavelets is completed using PGSDipscan. A multidimensional slant stack decomposes the data into a range of seismic wavelets in time.

Small bin sizes for dip-scanning form higher frequency products, however large bin widths improve the signal to noise ratio in the slant stack which is particularly useful in areas of weak signal, such as below salt bodies.

Migration and Reconstruction

Given a wavelet's position, dip components as well as a current earth model it is possible to 3D ray trace from source to receiver.

From the dip components, the takeoff and arrival angle can be calculated.  By combining this angle information with midpoint and offset coordinates, rays can be traced from both the source and the receiver location to determine the intercept location and place the wavelets intercept.  This provides the best receiver position and other properties such as reflector dip and angle of incidence.

The migration in hyperBeam is a point-to-point mapping of wavelets. This means that the seismic wavelet decomposition forms a basis in both un-migrated and migrated space with the velocity model the mapping function. Migration is applied to each coherent wavelet without creating a traveltime table. This eliminates multipath traveltime issues, steep-dip or turning wave problems and the requirement of an aperture limit.

Following the migration the data is reconstructed in migrated depth space.

Wavelet Selection

Wavelet attributes calculated during Dipscan and hyperBeam migration can be used to filter the data for later use in the tomographic inversion (hyperTomo).

The attribute filtering is performed in migrated space. The attributes used for the wavelet filtering include:

  • angle of incidence
  • offset-dip
  • reflector dip
  • 3D semblance

Using the assigned wavelet attributes, the reconstruction can be easily and quickly manipulated to weigh down or exclude a subset of the data. This can be used, for example, for noise reduction by scanning angle of incidence (see figures below).

The aim is to obtain a dataset that has a high signal-to-noise ratio with no cycle skipping and minimal multiple energy prior to measuring the velocity error.

Data preconditioningAngle of incidence outer trace mute scans – 0-30° (red), 30-35° (yellow), 35-40° (green), 40-45° (blue), 45-50° (purple)

Wavelets can be rejected with X-Y dips consistent with a dominant multiple generating surface.  The example gathers below show significant multiple contamination but by selecting wavelets with a particular dip in the offset direction (offset-dip) this energy can be rejected.  

 

Internal hyperBeam de-multipleGathers showing significant multiple contamination.

 

Internal hyperBeam de-multipleWavelet attribute selection using offset-dip range rejects wavelets with a particular dip in the offset direction

PGS Smart Flood was developed to assist with the interpretation of salt flanks however its use is not restricted to salt, but can be used wherever there is a recumbence and a large velocity contrast.

It is part of the data discrimination process and is based on tagging. The tagging is done on the change in velocity defined by the interface between a high velocity (such as salt) and the background sedimentary model.  

Smart Flood is a salt flood that also preserves the sedimentary reflectors illuminated through sediments and improves interpretability by reducing the number of mis-positioned events (mode conversions, multiples, etc).

Wavelet Shift Calculation (3DRNMO)

After PGS Beam migration, the reconstruction yields a stack that gives a good estimate of the reflector position and orientation; this is referred to as the reference stack and represents the zero-offset traveltime.

Wavelets are aligned to this stack by calculating traveltime shifts; this process is called 3DRNMO and uses the reference stack and migrated depth gathers. 3DRNMO shifts wavelets in three dimensions in order to produce flat gathers. By adjusting the traveltime, the wavelet is effectively shifted along the norm of the migrated dip: a cross-correlation against the stack provides a semblance from which the peak is picked to allocate a time shift.

The process allocates two additional attributes to each wavelet, the traveltime shift required to align the wavelet and the semblance. The calculated travel time shifts are used for tomographic inversion.

Tomographic Inversion

The hyperTomo tomographic inversion works by iteratively updating velocities along source-receiver ray paths to minimize the observed residual velocity error. The subsurface is divided into cells. The objective is to update the velocities along the ray paths in an iterative manner until the velocity errors are minimized, such that the magnitude of the updates falls below a set tolerance.

Key advantages of hyperTomo:

  • Curve fitting is avoided as 3DRNMO supplies residuals for individual wavelets. This allows for accurate tracking of complex moveout and the resolution of small scale anomalies - essential for high resolution velocity model building.
  • Time is saved by point to point ray tracing rather tracing a suite of rays.