Internal multiple prediction: an application on synthetic data, physical modeling data and field data

Melissa Judith Hernandez Quijada

In this work we examined and applied a method of internal multiple prediction based on the inverse scattering series. The internal multiple prediction algorithm predicts and then suppresses all order of internal multiples independent of the subsurface reflectors that generate them. In this thesis we promote a stepped approach to predicting multiples in a given field data set: first, by carrying out synthetic/numerical examples; second by carrying out tests on laboratory physical modeling data; and finally by testing prediction of a field data set suspected to be strongly contaminated with internal multiples. In the synthetic examples we draw conclusions about the central frequency of the seismic wavelet and the optimum choice for parameter epsilon (є). The physical modelling study, in which internal multiples are deliberately generated in order to be predicted, is the first of this kind. The results confirm the synthetic’s study conclusions regarding the estimation of epsilon (є), and motivated the development of a method for optimum estimation of epsilon (є) based on autocorrelation. In the land data study, the prediction allows us to confirm and precisely predict the presence of internal multiples in regions where they were expected.