| Popis: |
Summary Land data is usually noisier and spatially sparser than marine data, which hampers the application of data-driven multiple attenuation methods. In this paper, we present a practical implementation of the Surface Related Multiple Elimination (SRME) algorithm for land processing. The processing flow can be generalized into four steps: preprocessing, data preconditioning, multiple prediction, and least-squares subtraction. The preprocessing step follows a conventional land processing flow except that data is corrected to the surface instead of a flat datum. Based on the characteristics of land data, the data is preconditioned by applying aggressive noise reduction and regularization to stabilize the multiple prediction. In the final step, least-squares subtraction in higher dimensions is applied to remove multiples and better preserve primaries. Synthetic and real data examples show the robustness of the processing flow. |
