1.5D Internal Multiple Prediction: an Application on Synthetic Data, Physical Modelling Data and Land Data Synthetics

Pan (Penny) Pan

A 1.5D implementation of the inverse scattering series internal multiple prediction algorithm is investigated with the challenges of land seismic data application in mind. This method does not require any subsurface information and is suitable for situations where there is close interference between primaries and internal multiples; however, in land environments, issues of noise, coupling and statics have led to fewer reported successes. The methodology is also computationally costly, with the cost increasing dramatically as the implementation makes the transition from its 1D form to 1.5D, 2D and ultimately 3D. With these issues in mind, the algorithm is examined using a step-by-step approach: first, by carrying out synthetic examples; second, by testing physical modelling data; and finally, by operating on well log synthetics from land data. In the synthetic environment a study is undertaken to determine under what circumstances lower-dimension versions of the prediction algorithm can be applied to higher dimension problems to take advantage of the computational speed. The effects of various ϵ values are analyzed. A method to mitigate large-dip artifacts noticeable in unfiltered 1.5D internal multiple prediction is developed. Applicability of these ideas to real measurements taken in a physical modelling experiment, and using realistic synthetic data produced from real well logs is confirmed.