• Company Overview
• News & Events
• Seismic Services
• Software Products
• Technical Information

• Home
• Contacts
• Locations
• Press Kit
• Links
• Site Map

• Processing Solutions
• Proprietary Solutions

Technical Information

• Technical Introduction
• 3D Kirchhoff PSTM
• Radon DMO
• Tracker™
• Silencer™
• 3D PSPI Migration

TRACKER

3D Autostatics Solutions for Difficult Data

Tracker calculates surface consistent short wave residual statics based on the stack power maximization method proposed by Ronen and Claerbout (Geophysics Vol. 50, No. 12, December 1985). Tracker was specifically designed to solve the most difficult statics problems, including those encountered in poor signal-to-noise data sets, areas with large residual statics, and data with steep or conflicting dips. Tracker works on both 2D and 3D data sets.

Conventional Autostatics

Although adjacent trace correlation methods still have some use in current industry practice, residual statics are calculated most commonly by cross correlation of individual traces with CDP stacked traces (or with pilot/model traces built from stacked traces). A time pick is derived for each trace, equating to the maximum of the cross correlation. A set of over determined and under constrained simultaneous equations is built for the correlations with one equation for every trace. This set of equations is generally solved using Gauss-Seidel iterations. The CDP stack or pilot traces can be updated with the statics calculated, and the correlation and complex matrix solution processes may be iterated.

Tracker Autostatics

Tracker is similar to conventional autostatics in that the basic method of cross correlating individual traces to model or pilot traces is primary. However, Tracker does not use Gauss-Seidel iteration. Instead it relies on a number of options and features which allow the user to defeat noise-related problems directly.

Pilot traces can be formed in three different ways. First, the conventional stacked trace can be used as the model. This option is useful on good quality data sets with minimal skip or edge problems; this program-path also corresponds most closely to the specific method described by Ronen-Claerbout. Second, the pilot trace can be constructed by applying FX deconvolution for 2D data sets or FXY for 3D. FX models work best when some pilot data noise reduction is desirable, and the fidelity of the pilot traces is a priority. Finally, the user can build models via a smoothed, dip-smart mixing of energy between CDP stacked traces, a method most often used for challenging data sets. For 3D cases, this model is areally smoothed.

Regardless of which model-building path is chosen, Tracker is unique in that two complete cross correlations for every input trace are performed in each iteration. First, the traces for each shot group are correlated with the appropriate model traces. Second, the shot statics obtained are applied to the data, and an updated model is built which is then correlated with all traces from each receiver group. Iterations are repeated as necessary. Most data sets, regardless of difficulty, will converge fully in three to five full iterations. The power of CDP stack is measured after each iteration, providing a useful tool for determining whether or not convergence is complete.

Tracker avoids an acknowledged weakness associated with conventional (Gauss-Seidel) approaches, e.g. the distortion-tendencies due to the noise inherent in time picks derived from individual trace correlations. In Tracker, time picks are made after the cross correlations are summed for each shot and receiver. By controlling which traces are used based on offset (or offset ranges), fold, frequency, or correlation quality, the user can also limit problems associated with noise. For example, users can specify what percent of the best quality correlations they want to be considered; the percent can vary with each iteration. Tracker allows the computation of CDP trim statics with any of its three model-building options. Partial line (2D) or partial grid (3D) processing is possible.

Conclusions

Tracker allows the user several sound model-building options for obtaining residual statics for routine data processing. Because it is specifically constructed for problem data, it does require a little more processing power than conventional autostatics. However, Tracker offers more effective options and has a clear advantage on tough data sets where conventional autostatics fail.

Tracker is a standard module in Vista for Windows and is also available as a ProMAX socket tool or as a SEG-Y standalone program capable of running on most computers.