Practical, 3D surface-related multiple prediction (SMP)

Abstract

Prediction of surface multiples via 2D algorithms (eg, SRME) is now routine in data processing, and is often effective in removing those multiples when combined with an adaptive subtraction. There are, however, many situations in which the 2D assumptions made on the geology and the acquisition geometry are invalid to the extent that the resultant errors in the predicted multiples are too large to allow those multiples to be effectively subtracted. 3D algorithms have the potential to overcome these problems, and are therefore very attractive. However, there are many issues to be overcome when implementing a 3D algorithm for use on conventional data. These include inadequacies in the crossline sampling, the limited maximum crossline offset (aperture) of the recorded data, and irregularities in the acquisition geometry. This paper attempts to address these issues through the development of a methodology that predicts the required crossline aperture and the timing errors associated with a given acquisition geometry for a given mode of multiple. Knowledge of the aperture and timing errors is very useful in determining where a 3D prediction is appropriate, and in determining optimum parameters for that prediction. The results demonstrate the accuracy and value of the predicted errors, and the benefits of a 3D prediction over a 2D prediction for conventional marine data. 3D surface-related multiple prediction is practical for such datasets. The error analysis can also be used to optimize the acquisition geometry and hence the level of success of both 2D and 3D algorithms.