3D Image Driven Internal Multiple Elimination

Abstract

Abstract In this paper, we present both synthetic and field data study of applying the image-driven method proposed by Antonio Pica and Laurie Delmas (2008) for internal multiple attenuation. The image-driven method includes two steps: prediction of internal multiples and subtracting of the predicted multiples from the input. The prediction step comprises of the following four procedures: 1) the recorded seismic data is muted and propagated backward-in-time to user defined fictitious depth regions; 2) secondary seismic sources are generated by multiplying the migrated image and the extrapolated wavefield by the first step; 3) the secondary sources are further propagated forward-in-time and downward-in-depth and the downward extrapolated wavefield is multiplied with the migrated image and turned into up-going reflected waves; 4) the up-going waves are further propagated to the surface and used as the predicted internal multiples, which will be adjusted and subtracted from the input by using adaptive filter. Following the four steps described above, an internal-multiple-attenuation module is developed by using the one way wave-equation extrapolation. The method requires a migrated image and a migration velocity model as inputs. Where, the image is used as the reflectivity model, which is needed by the second and third steps mentioned above. The tests on both synthetic and field data show this method can correctly predict internal multiples, even when the migration velocity model is inaccurate. The tolerance to velocity errors stems from the fact, to certain extent, that the migrated image made from the erroneous velocity can compensate the errors presented in the velocity model in terms of making the traveltime of predicted multiples correct. Compared with the data-driven method, the main advantage of the image-driven method is that it imposes fewer requirements on the acquisition geometry.