Data needed by InterWell are:

• Seismic data (full stack or sub stacks amplitudes segy files)
• Optionally Interval seismic velocities (segy)
• Interpreted horizons on seismic (X, Y, TWT or IL, XL, TWT format)
• Well data (LAS2.0 or 3.0 files):
  o Elastic impedances for each sub stack
  o IP, IS and RHO logs
  o Markers

Prior to the inversion to check the quality of the imported data, the INTViewer platform which support the InterWell functionalities offers a wide set of visualization and QC tools such as statistic tools (histograms, cross plots, Frequency analysis) and viewers.

In the context of an elastic inversion, in each sub stack a same event can be located at slightly different time locations because of the NMO processing. To ensure the quality of the joint pre stack inversion it is mandatory to remove these residual misalignments and to align all sub stacks to the same reference (usually the mid angle stack or a full stack cube).

InterWell provides a specific tool based on a volumetric correction of the vertical misalignments between cubes with control of the amplitude accuracy. This tool proceeds to a global analysis of the misalignments then check the quality of the correction using a user's defined quality cutoff and finally writes a seismic cube aligned to the reference sub stack.

All sub stacks but the reference one must be processed to ensure a good reliability of the inversion results

Reliability of inversion results actually depends to a great extend on the wavelet used for modeling the seismic response during the process. The wavelet must then ensure the best possible match between the synthetic seismogram calculated at wells and the neighboring traces.

InterWell applies an hybrid (both deterministic and statistical) method for wavelet estimation that employs multi-well, multi-trace calibration. Wavelet obtained using this hybrid method are more accurate and fairly consistent as they also rely on well data.

As InterWell inversion process uses geological information it is mandatory to define the structural framework highlighted by the seismic data. This is done by selecting a set of interpreted horizons between which the deposition sequence is homogeneous and by defining the type of deposition between each used horizons.

The deposition between horizons defining a sequence can be parallel to the top horizon, parallel to the bottom horizon, parallel to another horizon or accommodating a progressive parallelization between both bottom and top horizon. Erosive horizons can be managed

In order to take into account the geological information, InterWell needs at each seismic sample a value of the impedances deduced from available wells data and structural framework. This is simply done by computing a volume of IP, IS and RHO data from an interpolation of all well logs along virtual deposition surfaces defined by the structural framework. This interpolation can be improved using interval velocity data (non stationary interpolation) particularly in the seismic part not investigated by well logs.

The computed initial model produce IP, IS, RHO cubes along with DIP_IL and DIP_XL cubes which will be used as a guide for the inversion process.

This is the final step of the classical InterWell workflow. In order to provide best optimal impedances it is mandatory to feed the algorithm with Seismic data, Geological data (the initial or "a priori" model) and a set of parameters which will tuned the algorithm according to the user's estimation of the data quality. These parameters are simply defined by the N/S ratio for each sub stack (defining thus the quality of the seismic), the Impedances standard deviation (allowing the inversion to more or less take into account the low frequency model) and a correlation distance which will smoothen the results according to the geological framework (following dip information).

The InterWell elastic inversion is a joint inversion, all the angle stacks are inverted together in order to ensure a better S impedance result and to maximize the robustness toward angle dependant noise.

The final results of the inversion are: IP, IS, RHO cubes, Synthetic seismic, Seismic Reflection coefficients  and difference between Observed and Synthetic Seismic for each sub stacks.

The optimal impedance model is one solution among a wide range of admissible models that fits the seismic data and the a priori information. Thanks to the Bayesian formalism of the inversion process implemented in InterWell, it is now possible to directly extract the a posteriori uncertainties on IP, IS and RHO (uncertainties on results depend on the "a priori" uncertainties taken into account). These uncertainties can be deterministically assessed from inversion results and parameters.

The main principle is to estimate the posterior covariance matrix. The exact computation of this matrix consists in inverting the Hessian of the objective function at the optimal solution.

This new functionality of InterWell allows the simultaneous inversion a several seismic vintages. The basic workflow implies:

• Compute sequential elastic inversion from angle gathers stacks performed individually for both base and monitors surveys
• Compute a scaling law to match the reference and monitor time basis in the joint inversion process (Warping factor). This is done by taking IP volumes from previous step
• Performed the joint elastic 4D inversion from stacked angle gathers after warping.

This enable the estimation of elastic properties from amplitudes variations.

A major gain in productivity is the consequence of the use of the same formalism and methodology used for post and pre stack inversion.