next up previous print clean
Next: REFERENCES Up: Guitton: Pattern-based multiple attenuation Previous: A 2-D field data

Discussion-Conclusion

I have presented a multiple-attenuation technique that aims to model and separate the noise and signal with time-variant, multidimensional prediction-error filters. These filters can be 2-D or 3-D thanks to the helical boundary conditions. The estimation of these filters incorporate weighting functions to cope with amplitude variations in the data and in the noise model. In addition, a masking operator is introduced in the signal/noise separation method in order to preserve areas where no multiples are present. Tests with a field data example from the Gulf of Mexico prove that the multiple attenuation works much better when 3-D filters are utilized, as opposed to 2-D filters. These 3-D filters allow a better signal/separation in areas where the multiple model is known to be inaccurate, e.g., short offset traces, diffracted multiples and off-plane/3-D events.


next up previous print clean
Next: REFERENCES Up: Guitton: Pattern-based multiple attenuation Previous: A 2-D field data
Stanford Exploration Project
7/8/2003